Add files using upload-large-folder tool

05d9d5da933137c5402a176a469b618685c7e9142aa8972616ca5cdf0f6e53d1/crowd-code-9ff54a43-2a59-41a8-96bc-f7e46d5244651750887279734-2025_06_25-23.36.32.560/source.csv

@@ -0,0 +1,3 @@
+Sequence,Time,File,RangeOffset,RangeLength,Text,Language,Type
+1,18,"scripts/file_duplicate_checker.py",0,0,"import os\nfrom collections import defaultdict\nfrom tqdm import tqdm\n\ndef find_duplicate_filenames(root_dir):\n    filenames = defaultdict(list)\n    file_count = 0\n\n    # Use tqdm with manual update and no percentage/ETA bar\n    pbar = tqdm(desc=""Files scanned"", unit=""file"", dynamic_ncols=True, bar_format=""{desc}: {n_fmt}"")\n\n    # Walk the directory recursively\n    for dirpath, _, files in os.walk(root_dir):\n        for file in files:\n            full_path = os.path.join(dirpath, file)\n            if os.path.isfile(full_path):\n                filenames[file].append(full_path)\n                file_count += 1\n                pbar.update(1)\n\n    pbar.close()\n\n    # Print duplicates\n    duplicates = {name: paths for name, paths in filenames.items() if len(paths) > 1}\n    if duplicates:\n        print(""\nDuplicate filenames found:\n"")\n        for name, paths in duplicates.items():\n            print(f""Filename: {name}"")\n            for path in paths:\n                print(f"" - {path}"")\n            print()\n    else:\n        print(""\nNo duplicate filenames found."")\n\nif __name__ == ""__main__"":\n    import sys\n    if len(sys.argv) < 2:\n        print(""Usage: python find_duplicates.py <directory_path>"")\n    else:\n        find_duplicate_filenames(sys.argv[1])\n\n",python,tab
+2,521,"extension-output-pdoom-org.crowd-code-#1-crowd-code",0,0,"11:36:32 PM [info] Activating crowd-code\n11:36:32 PM [info] Welcome back tum_ind3695. Your user-id is '507ab0ec0dfe0c18ad7778dd15e072f92367194c94623114de802c8ed9c52e20'. Happy coding!\n11:36:32 PM [info] Recording started\n",Log,tab

58dff52cba2a091453cfbef6169091e684254819f0b9f334dbecea6a130284bc/crowd-code-495b78a3-8c04-4965-88fa-979c320df6561767630491949-2026_01_05-17.28.24.506/source.csv

@@ -0,0 +1,3 @@
+Sequence,Time,File,RangeOffset,RangeLength,Text,Language,Type
+2,464,"extension-output-pdoom-org.crowd-code-#1-crowd-code",0,0,"5:28:24 PM [info] Activating crowd-code\n5:28:24 PM [info] Recording started\n5:28:24 PM [info] Initializing git provider using file system watchers...\n5:28:24 PM [error] Not a git repository: EntryNotFound (FileSystemError): Error: ENOENT: no such file or directory, stat '/fast/project/HFMI_SynergyUnit/tab_model/data/qwen/.git'\n",Log,tab
+3,2064,"extension-output-pdoom-org.crowd-code-#1-crowd-code",329,0,"5:28:26 PM [info] Retrying git provider initialization...\n5:28:26 PM [error] Not a git repository: EntryNotFound (FileSystemError): Error: ENOENT: no such file or directory, stat '/fast/project/HFMI_SynergyUnit/tab_model/data/qwen/.git'\n",Log,content

58dff52cba2a091453cfbef6169091e684254819f0b9f334dbecea6a130284bc/crowd-code-df110f6f-a3c9-4641-90da-f30934c027e01767707463233-2026_01_06-14.51.27.372/source.csv

@@ -0,0 +1,64 @@
+Sequence,Time,File,RangeOffset,RangeLength,Text,Language,Type
+1,4,"tab-model-eval/src/evaluation/sglang_eval.py",0,0,"import asyncio\nimport json\nimport os\nimport sys\nimport subprocess\nimport time\nimport wandb\nfrom dataclasses import dataclass, field\nfrom typing import Dict, Any, List, Optional\n\nimport httpx\nimport tyro\nfrom openai import AsyncOpenAI, BadRequestError\nfrom tqdm.asyncio import tqdm_asyncio\n\nclass LocalLogger:\n    """"""A simple local logger that saves metrics to JSON files for later sync to wandb.""""""\n    \n    def __init__(self, log_dir: str, run_id: str, run_name: str, project: str, config: dict = None, tags: list = None):\n        self.log_dir = os.path.join(log_dir, run_id)\n        os.makedirs(self.log_dir, exist_ok=True)\n        self.run_id = run_id\n        self.run_name = run_name\n        self.project = project\n        self.config = config or {}\n        self.tags = tags or []\n        self.metrics_file = os.path.join(self.log_dir, ""metrics.jsonl"")\n        \n        # Save run metadata\n        metadata_file = os.path.join(self.log_dir, ""metadata.json"")\n        with open(metadata_file, ""w"") as f:\n            json.dump({\n                ""run_id"": run_id,\n                ""run_name"": run_name,\n                ""project"": project,\n                ""config"": config,\n                ""tags"": tags,\n                ""created_at"": time.strftime(""%Y-%m-%dT%H:%M:%S"")\n            }, f, indent=2)\n        \n        print(f""LocalLogger initialized. Logs will be saved to: {self.log_dir}"")\n    \n    def log(self, metrics: dict):\n        """"""Append metrics to the JSONL file.""""""\n        metrics_with_timestamp = {\n            ""timestamp"": time.strftime(""%Y-%m-%dT%H:%M:%S""),\n            **metrics\n        }\n        with open(self.metrics_file, ""a"") as f:\n            f.write(json.dumps(metrics_with_timestamp) + ""\n"")\n        print(f""Logged metrics to {self.metrics_file}: eval_step={metrics.get('eval_step', 'N/A')}"")\n    \n    def finish(self):\n        """"""Called when logging is complete.""""""\n        print(f""LocalLogger finished. All logs saved to: {self.log_dir}"")\n\n\n# ----------------------------\n# Argument definitions\n# ----------------------------\n@dataclass\nclass Args:\n    # Eval-related\n    wandb_project: str = ""llm-coding-agent""\n    wandb_name: str = ""validation_set_eval""\n    wandb_eval_type: str = ""next_action_validation_set""\n    wandb_tags: list[str] = field(default_factory=lambda: [""val_mini"", ""judge_eval""])\n    wandb_id: str | None = None\n    wandb_group: str = ""debug""\n    \n    # Single-file mode (backward compatible)\n    generations_file: str = """"\n    evaluations_file: str = """"\n    eval_step: int = 0\n    \n    # Batch mode: comma-separated lists of files and steps\n    # When these are provided, they take precedence over single-file args\n    generations_files: str = """"  # Comma-separated list of generation files\n    evaluations_files: str = """"  # Comma-separated list of evaluation output files\n    eval_steps: str = """"  # Comma-separated list of eval steps (integers)\n    \n    limit: int = -1\n    system_prompt_file: str = ""data/prompts/judge_system_prompt_v2.md""\n    judge_name: str = ""default""\n    judge_prompt_file: str = ""data/prompts/judge_prompt_v2.md""\n    judge_prompt_file_with_context: str = ""data/prompts/judge_prompt_v2_with_context.md""\n\n    # Local logging for offline mode\n    use_local_logger: bool = False\n    local_log_dir: str = ""data/eval/local_logs""\n\n    # Server-related (sglang)\n    judge_model_path: str = ""Qwen/Qwen3-Coder-30B-A3B-Instruct""\n    server_host: str = ""0.0.0.0""\n    server_port: int = 30000\n    context_length: int = 40960\n    problem_length: int = 40960\n    api_key: str = ""EMPTY""  # sglang's OpenAI-compatible server ignores this value\n    mem_fraction_static: float = 0.95\n    tp_size: int = 1\n\n    # Client-related\n    temperature: float = 0.7\n    top_p: float = 0.8\n    presence_penalty: float = 1.5\n    top_k: int = 20\n    min_p: float = 0.0\n    enable_thinking: bool = True\n\n    # HTTP / client config\n    concurrency: int = 16\n    max_connections: int = 256\n    keepalive: int = 60\n    max_attempts: int = 6\n    timeout: float = 30.0\n\n    # Control whether to launch server from this script\n    launch_server: bool = True\n    # Extra args passed to `sglang.launch_server` if needed\n    extra_server_args: Optional[List[str]] = None\n    \n    def get_eval_jobs(self) -> List[tuple[str, str, int]]:\n        """"""\n        Returns a list of (generations_file, evaluations_file, eval_step) tuples.\n        If batch mode args are provided, uses those. Otherwise falls back to single-file mode.\n        """"""\n        if self.generations_files and self.evaluations_files and self.eval_steps:\n            # Batch mode\n            gen_files = [f.strip() for f in self.generations_files.split("","") if f.strip()]\n            eval_files = [f.strip() for f in self.evaluations_files.split("","") if f.strip()]\n            steps = [int(s.strip()) for s in self.eval_steps.split("","") if s.strip()]\n            \n            if not (len(gen_files) == len(eval_files) == len(steps)):\n                raise ValueError(\n                    f""Batch mode requires equal-length lists for generations_files ({len(gen_files)}), ""\n                    f""evaluations_files ({len(eval_files)}), and eval_steps ({len(steps)})""\n                )\n            \n            return list(zip(gen_files, eval_files, steps))\n        elif self.generations_file and self.evaluations_file:\n            # Single-file mode (backward compatible)\n            return [(self.generations_file, self.evaluations_file, self.eval_step)]\n        else:\n            raise ValueError(\n                ""Either provide single-file args (generations_file, evaluations_file) ""\n                ""or batch args (generations_files, evaluations_files, eval_steps)""\n            )\n\n\n# ----------------------------\n# Dataset helpers\n# ----------------------------\ndef load_dataset(filepath):\n    with open(filepath, ""r"") as f:\n        return json.loads(f.read())\n\n\ndef estimate_token_count(messages: List[Dict[str, str]]) -> int:\n    """"""\n    Rough estimate of token count for a list of messages.\n    Assumes ~3 characters per token as a conservative estimate.\n    """"""\n    total_chars = sum(len(msg.get(""content"", """")) for msg in messages)\n    return total_chars // 3\n\n\ndef filter_tasks_by_context_length(\n    test_cases: List[Dict[str, Any]],\n    system_prompt: str,\n    prompt_template: str,\n    max_context_length: int = 40960,\n    problem_length: int = 40960,\n    buffer_tokens: int = 512,  # Reserve space for response\n    include_context: bool = False,\n) -> tuple[List[Dict[str, Any]], List[Dict[str, Any]]]:\n    """"""\n    Filter out test cases whose context would exceed the model's context length.\n    Returns (valid_cases, skipped_cases)\n    """"""\n    valid_cases = []\n    skipped_cases = []\n\n    for tc in test_cases:\n        # Estimate tokens for system prompt + context\n        messages = [{""role"": ""system"", ""content"": system_prompt}]\n        if include_context:\n            messages.extend(tc[""context""])\n        messages.append({""role"": ""user"", ""content"": prompt_template})\n        estimated_tokens = estimate_token_count(messages)\n\n        length = estimated_tokens + buffer_tokens\n        if length <= max_context_length and length <= problem_length:\n            valid_cases.append(tc)\n        else:\n            print(\n                f""Skipping {tc['task_id']}: estimated {estimated_tokens} tokens (limit: {max_context_length}, problem_length: {problem_length})""\n            )\n            skipped_cases.append(\n                {\n                    ""task_id"": tc[""task_id""],\n                    ""estimated_tokens"": estimated_tokens,\n                    ""reason"": ""context_too_long"",\n                }\n            )\n\n    return valid_cases, skipped_cases\n\n\n# ----------------------------\n# Eval logic\n# ----------------------------\nasync def evaluate_generated_command(\n    client: AsyncOpenAI,\n    sem: asyncio.Semaphore,\n    test_case: Dict[str, Any],\n    args: Args,\n    system_prompt: str,\n    prompt_template: str,\n    include_context: bool,\n) -> Dict[str, Any]:\n    """"""\n    Handles a single evaluation task with concurrency control and retries.\n    """"""\n    async with sem:\n        delay = 0.25\n\n        if test_case.get(""error"", None) is not None:\n            print(f""Returning failure object for task {test_case['task_id']} due to error"")\n            return {\n                ""task_id"": test_case[""task_id""],\n                ""error"": test_case[""error""],\n                ""is_correct"": 0,\n                ""average_score"": 0.0,\n            }\n\n        samples = test_case.get(""samples"", [])\n        if not samples:\n            print(f""Returning failure object for task {test_case['task_id']} due to no samples"")\n            return {\n                ""task_id"": test_case[""task_id""],\n                ""error"": ""No samples"",\n                ""is_correct"": 0,\n                ""average_score"": 0.0,\n            }\n\n        sample_results = []\n        for sample in samples:\n            for attempt in range(args.max_attempts):\n                try:\n                    format_dict = {\n                        ""expected"": test_case[""expected_command""],\n                        ""generated"": sample[""generated_command""],\n                    }\n                    if include_context:\n                        format_dict[""context""] = json.dumps(test_case[""context""], indent=2)\n                    prompt = prompt_template.format(**format_dict)\n\n                    messages = [\n                        {\n                            ""role"": ""system"",\n                            ""content"": system_prompt,\n                        },\n                        {""role"": ""user"", ""content"": prompt},\n                    ]\n\n                    resp = await client.chat.completions.create(\n                        model=args.judge_name,\n                        messages=messages,\n                        temperature=args.temperature,\n                        top_p=args.top_p,\n                        presence_penalty=args.presence_penalty,\n                        response_format={""type"": ""json_object""},\n                        extra_body={\n                            ""top_k"": args.top_k,\n                        },\n                    )\n\n                    thinking_trace = getattr(resp.choices[0].message, ""reasoning_content"", """")\n                    result = json.loads(resp.choices[0].message.content)\n                    equivalent = result.get(""equivalent"", 0)\n\n                    sample_results.append(\n                        {\n                            ""generated_command"": sample[""generated_command""],\n                            ""thinking_trace"": thinking_trace,\n                            ""evaluation_results"": result,\n                            ""equivalent"": equivalent,\n                            ""exact_match"": sample[""exact_match""],\n                        }\n                    )\n                    break\n\n                except BadRequestError as e:\n                    print(\n                        f""Returning failure object for task {test_case['task_id']} due to BadRequestError: {e}""\n                    )\n                    sample_results.append(\n                        {\n                            ""task_id"": test_case[""task_id""],\n                            ""error"": str(e),\n                            ""equivalent"": 0,\n                            ""exact_match"": 0,\n                        }\n                    )\n                    break\n\n                except Exception as e:\n                    print(f""Error on task {test_case['task_id']}: {e}"")\n                    if attempt == args.max_attempts - 1:\n                        print(f""Returning failure object for task {test_case['task_id']}"")\n                        sample_results.append(\n                            {\n                                ""task_id"": test_case[""task_id""],\n                                ""error"": str(e),\n                                ""equivalent"": 0,\n                            }\n                        )\n                    await asyncio.sleep(delay)\n                    delay *= 2\n\n        # Compute avg@n and pass@n\n        num_judge_matches = sum(s.get(""equivalent"", 0) for s in sample_results)\n        judge_avg_at_n = num_judge_matches / len(sample_results)\n        judge_pass_at_n = int(num_judge_matches > 0)\n        num_exact_matches = test_case.get(""num_exact_matches"", 0)\n\n        return {\n            ""task_id"": test_case[""task_id""],\n            ""context"": test_case[""context""],\n            ""expected_command"": test_case[""expected_command""],\n            ""sample_evaluations"": sample_results,\n            ""num_samples"": len(sample_results),\n            ""num_judge_matches"": num_judge_matches,\n            ""judge_avg_at_n"": judge_avg_at_n,\n            ""judge_pass_at_n"": judge_pass_at_n,\n            ""num_exact_matches"": num_exact_matches,\n        }\n\n\nasync def run_single_eval(\n    args: Args,\n    generations_file: str,\n    evaluations_file: str,\n    eval_step: int,\n    client: AsyncOpenAI,\n    sem: asyncio.Semaphore,\n    system_prompt: str,\n    prompt_template: str,\n    include_context: bool,\n    logger: Optional[LocalLogger] = None,\n) -> Dict[str, Any]:\n    """"""\n    Evaluate a single generations file and write results to evaluations file.\n    Uses shared client and semaphore for efficiency in batch mode.\n    Returns the evaluation scores dictionary.\n    """"""\n    print(f""\n{'='*60}"")\n    print(f""Evaluating: {generations_file}"")\n    print(f""Output: {evaluations_file}"")\n    print(f""Step: {eval_step}"")\n    print(f""{'='*60}"")\n    \n    loaded_data = load_dataset(generations_file)\n    test_cases = loaded_data[""generation_results""]\n\n    config_generations = loaded_data[""config_generations""]\n    config_evaluations = args.__dict__\n    metadata = {\n        ""config_generations"": config_generations,\n        ""config_evaluations"": config_evaluations,\n    }\n\n    if args.limit > 0:\n        test_cases = test_cases[: args.limit]\n\n    # Filter out tasks with context that's too long\n    test_cases, skipped_cases = filter_tasks_by_context_length(\n        test_cases,\n        system_prompt=system_prompt,\n        prompt_template=prompt_template,\n        max_context_length=args.context_length,\n        problem_length=args.problem_length,\n        buffer_tokens=512,\n        include_context=include_context,\n    )\n\n    print(f""\nFiltered dataset:"")\n    print(f""  Valid test cases: {len(test_cases)}"")\n    print(f""  Skipped (too long): {len(skipped_cases)}"")\n    print()\n\n    # Clean output\n    if os.path.exists(evaluations_file):\n        os.remove(evaluations_file)\n\n    tasks = [\n        evaluate_generated_command(\n            client, sem, tc, args, system_prompt, prompt_template, include_context\n        )\n        for tc in test_cases\n    ]\n\n    print(f""Running {len(test_cases)} test cases with concurrency={args.concurrency} ..."")\n    results: List[Dict[str, Any]] = []\n\n    # progress bar over async tasks\n    for coro in tqdm_asyncio.as_completed(tasks, total=len(tasks)):\n        results.append(await coro)\n\n    # sort the results by task_id\n    results.sort(key=lambda x: x[""task_id""])\n\n    os.makedirs(os.path.dirname(evaluations_file), exist_ok=True)\n    total_judge_avg_at_n = sum(r.get(""judge_avg_at_n"", 0) for r in results) / len(results)\n    total_judge_pass_at_n = sum(r.get(""judge_pass_at_n"", 0) for r in results)\n\n    total_exact_match_avg_at_n = loaded_data[""generation_scores""][""total_exact_match_avg_at_n""]\n    total_exact_match_pass_at_n = loaded_data[""generation_scores""][""total_exact_match_pass_at_n""]\n\n    # Prepare metrics to log\n    metrics_to_log = {\n        ""eval_step"": eval_step,\n        f""{args.wandb_eval_type}/total_test_cases"": len(test_cases),\n        f""{args.wandb_eval_type}/num_samples_per_task"": loaded_data[""config_generations""][\n            ""num_samples""\n        ],\n        f""{args.wandb_eval_type}/total_judge_avg_at_n"": total_judge_avg_at_n,\n        f""{args.wandb_eval_type}/total_judge_pass_at_n"": total_judge_pass_at_n,\n        f""{args.wandb_eval_type}/total_exact_match_avg_at_n"": total_exact_match_avg_at_n,\n        f""{args.wandb_eval_type}/total_exact_match_pass_at_n"": total_exact_match_pass_at_n,\n    }\n    \n    # Log metrics using appropriate logger\n    if args.use_local_logger:\n        logger.log(metrics_to_log)\n    else:\n        wandb.log(metrics_to_log)\n\n    with open(evaluations_file, ""w"") as f:\n        json.dump(\n            {\n                ""metadata"": metadata,\n                ""evaluation_scores"": {\n                    ""total_test_cases"": len(test_cases),\n                    ""num_samples_per_task"": loaded_data[""config_generations""][""num_samples""],\n                    ""total_judge_avg_at_n"": total_judge_avg_at_n,\n                    ""total_judge_pass_at_n"": total_judge_pass_at_n,\n                    ""total_exact_match_avg_at_n"": total_exact_match_avg_at_n,\n                    ""total_exact_match_pass_at_n"": total_exact_match_pass_at_n,\n                    ""max_attempts"": args.max_attempts,\n                },\n                ""generation_results"": results,\n            },\n            f,\n            indent=2,\n        )\n\n    print(""\n"" + ""="" * 50)\n    print(f""--- Evaluation Complete (step {eval_step}) ---"")\n    print(""="" * 50)\n    print(f""Total Test Cases: {len(test_cases)}"")\n    print(f""Total Judge Pass At N: {total_judge_pass_at_n}"")\n    print(f""Total Judge Avg At N: {total_judge_avg_at_n * 100:.2f}%"")\n    print(f""Total Exact Match Pass At N: {total_exact_match_pass_at_n}"")\n    print(f""Total Exact Match Avg At N: {total_exact_match_avg_at_n * 100:.2f}%"")\n    print(f""Evaluations output file: {evaluations_file}"")\n\n    return {\n        ""eval_step"": eval_step,\n        ""generations_file"": generations_file,\n        ""evaluations_file"": evaluations_file,\n        ""total_test_cases"": len(test_cases),\n        ""total_judge_avg_at_n"": total_judge_avg_at_n,\n        ""total_judge_pass_at_n"": total_judge_pass_at_n,\n        ""total_exact_match_avg_at_n"": total_exact_match_avg_at_n,\n        ""total_exact_match_pass_at_n"": total_exact_match_pass_at_n,\n    }\n\n\nasync def run_batch_eval(args: Args, base_url: str):\n    """"""\n    Run evaluation on multiple generation files with a single model load.\n    This avoids the overhead of loading/unloading the judge model for each checkpoint.\n    """"""\n    eval_jobs = args.get_eval_jobs()\n    \n    print(f""\n{'#'*60}"")\n    print(f""# BATCH EVALUATION MODE"")\n    print(f""# Processing {len(eval_jobs)} evaluation job(s)"")\n    print(f""{'#'*60}\n"")\n    \n    for i, (gen_file, eval_file, step) in enumerate(eval_jobs):\n        print(f""  [{i+1}/{len(eval_jobs)}] Step {step}: {gen_file}"")\n    print()\n    \n    # Load prompts once (shared across all evaluations)\n    with open(args.system_prompt_file, ""r"") as f:\n        system_prompt = f.read()\n\n    include_context = bool(args.judge_prompt_file_with_context)\n    judge_prompt_file = args.judge_prompt_file_with_context or args.judge_prompt_file\n\n    with open(judge_prompt_file, ""r"") as f:\n        prompt_template = f.read()\n\n    # Initialize logger (local or wandb) - shared across all evaluations\n    logger = None\n    if args.use_local_logger:\n        run_id = args.wandb_id or args.wandb_name\n        logger = LocalLogger(\n            log_dir=args.local_log_dir,\n            run_id=run_id,\n            run_name=args.wandb_name,\n            project=args.wandb_project,\n            config={""batch_mode"": True, ""num_jobs"": len(eval_jobs)},\n            tags=args.wandb_tags,\n        )\n    else:\n        wandb_init_kwargs = {\n            ""project"": args.wandb_project,\n            ""name"": args.wandb_name,\n            ""tags"": args.wandb_tags,\n            ""group"": args.wandb_group,\n            ""config"": {""batch_mode"": True, ""num_jobs"": len(eval_jobs)},\n        }\n\n        if args.wandb_id:\n            wandb_dir = os.path.join(os.getcwd(), ""eval_logs"", args.wandb_id)\n            os.makedirs(wandb_dir, exist_ok=True)\n            os.environ[""WANDB_DIR""] = wandb_dir\n            os.environ[""WANDB_RESUME""] = ""allow""\n            \n            wandb_init_kwargs.update(\n                {\n                    ""id"": args.wandb_id,\n                    ""resume"": ""allow"",\n                    ""dir"": wandb_dir,\n                }\n            )\n        wandb.init(**wandb_init_kwargs)\n\n    # Reuse a single HTTP/2 client with a large pool (shared across all evaluations)\n    http = httpx.AsyncClient(\n        http2=True,\n        timeout=httpx.Timeout(args.timeout, connect=10.0, read=args.timeout),\n        limits=httpx.Limits(\n            max_connections=args.max_connections,\n            max_keepalive_connections=args.max_connections,\n            keepalive_expiry=args.keepalive,\n        ),\n        headers={""Connection"": ""keep-alive""},\n    )\n    client = AsyncOpenAI(\n        base_url=base_url,\n        api_key=args.api_key,\n        http_client=http,\n    )\n    sem = asyncio.Semaphore(args.concurrency)\n\n    # Process each evaluation job sequentially\n    all_results = []\n    for i, (gen_file, eval_file, step) in enumerate(eval_jobs):\n        print(f""\n[{i+1}/{len(eval_jobs)}] Processing step {step}..."")\n        \n        result = await run_single_eval(\n            args=args,\n            generations_file=gen_file,\n            evaluations_file=eval_file,\n            eval_step=step,\n            client=client,\n            sem=sem,\n            system_prompt=system_prompt,\n            prompt_template=prompt_template,\n            include_context=include_context,\n            logger=logger,\n        )\n        all_results.append(result)\n\n    await http.aclose()\n    \n    # Finish logging\n    if args.use_local_logger:\n        logger.finish()\n    else:\n        wandb.finish()\n\n    # Print summary\n    print(""\n"" + ""#"" * 60)\n    print(""# BATCH EVALUATION SUMMARY"")\n    print(""#"" * 60)\n    for r in all_results:\n        print(f""  Step {r['eval_step']:>5}: Judge Avg@N = {r['total_judge_avg_at_n']*100:5.2f}%, ""\n              f""Pass@N = {r['total_judge_pass_at_n']}, ""\n              f""Exact Avg@N = {r['total_exact_match_avg_at_n']*100:5.2f}%"")\n    print(""#"" * 60)\n\n\n# ----------------------------\n# Server launch + waiting\n# ----------------------------\nasync def wait_for_server(base_url: str, timeout: float = 600.0) -> None:\n    """"""\n    Poll the server's OpenAI-compatible endpoint until it responds or timeout.\n    We'll try a lightweight call to /models.\n    """"""\n    print(f""Waiting for server at {base_url} ..."")\n    deadline = asyncio.get_event_loop().time() + timeout\n\n    async with httpx.AsyncClient() as client:\n        while True:\n            now = asyncio.get_event_loop().time()\n            if now > deadline:\n                raise RuntimeError(\n                    f""Server at {base_url} did not become ready within {timeout} seconds.""\n                )\n            try:\n                resp = await client.get(f""{base_url}/v1/models"", timeout=5.0)\n                if resp.status_code == 200:\n                    print(""Server is up."")\n                    return\n                else:\n                    print(f""Server not ready yet (status {resp.status_code}); retrying..."")\n            except Exception as e:\n                print(f""Server not ready yet ({e}); retrying..."")\n            await asyncio.sleep(10.0)\n\n\ndef launch_sglang_server(args: Args) -> subprocess.Popen:\n    """"""\n    Launch sglang server as a subprocess.\n    You should have `module load CUDA/12.8` and `source .venv/bin/activate`\n    done in your shell before running this script.\n    """"""\n    cmd = [\n        sys.executable,\n        ""-m"",\n        ""sglang.launch_server"",\n        ""--model-path"",\n        args.judge_model_path,\n        ""--host"",\n        args.server_host,\n        ""--port"",\n        str(args.server_port),\n        ""--context-length"",\n        str(args.context_length),\n        ""--mem-fraction-static"",\n        str(args.mem_fraction_static),\n        ""--tp-size"",\n        str(args.tp_size),\n    ]\n\n    if args.extra_server_args:\n        cmd.extend(args.extra_server_args)\n\n    print(""Launching sglang server:"")\n    print(""  "" + "" "".join(cmd))\n\n    env = os.environ.copy()\n    proc = subprocess.Popen(\n        cmd,\n        env=env,\n        stdout=sys.stdout,\n        stderr=sys.stderr,\n    )\n    return proc\n\n\n# ----------------------------\n# Main\n# ----------------------------\nasync def amain(args: Args):\n    base_url = f""http://{args.server_host}:{args.server_port}/v1""\n    print(f""Using server at {base_url}"")\n\n    server_proc: Optional[subprocess.Popen] = None\n    try:\n        if args.launch_server:\n            server_proc = launch_sglang_server(args)\n            await wait_for_server(f""http://{args.server_host}:{args.server_port}"")\n\n        await run_batch_eval(args, base_url=base_url)\n\n    finally:\n        if server_proc is not None:\n            print(""Shutting down sglang server ..."")\n            server_proc.terminate()\n            try:\n                server_proc.wait(timeout=30)\n            except subprocess.TimeoutExpired:\n                print(""Server did not exit in time; killing."")\n                server_proc.kill()\n\n\nif __name__ == ""__main__"":\n    args = tyro.cli(Args)\n    asyncio.run(amain(args))\n    print(""Done"")\n",python,tab
+2,237,"extension-output-pdoom-org.crowd-code-#1-crowd-code",0,0,"2:51:27 PM [info] Activating crowd-code\n2:51:27 PM [info] Recording started\n2:51:27 PM [info] Initializing git provider using file system watchers...\n2:51:27 PM [info] Git repository found\n2:51:27 PM [info] Git provider initialized successfully\n2:51:27 PM [info] Initial git state: [object Object]\n",Log,tab
+3,1773,"tab-model-eval/src/evaluation/sglang_eval.py",0,0,"",python,tab
+4,142831,"TERMINAL",0,0,"git diff",,terminal_command
+5,142985,"TERMINAL",0,0,"]633;C[?1h=\rdiff --git a/train.py b/train.py\r\ndeleted file mode 100644\r\nindex 9fb480e..0000000\r\n--- a/train.py\r\n+++ /dev/null\r\n@@ -1,106 +0,0 @@\r\n-import ray\r\n-from sglang.srt.constants import GPU_MEMORY_TYPE_KV_CACHE, GPU_MEMORY_TYPE_WEIGHTS\r\n-\r\n-try:\r\n-    from sglang.srt.constants import GPU_MEMORY_TYPE_CUDA_GRAPH\r\n-except ImportError:\r\n-    GPU_MEMORY_TYPE_CUDA_GRAPH = None\r\n:",,terminal_output
+6,144565,"TERMINAL",0,0,"...skipping...\r\ndiff --git a/train.py b/train.py\r\ndeleted file mode 100644\r\nindex 9fb480e..0000000\r\n--- a/train.py\r\n+++ /dev/null\r\n@@ -1,106 +0,0 @@\r\n-import ray\r\n-from sglang.srt.constants import GPU_MEMORY_TYPE_KV_CACHE, GPU_MEMORY_TYPE_WEIGHTS\r\n-\r\n-try:\r\n-    from sglang.srt.constants import GPU_MEMORY_TYPE_CUDA_GRAPH\r\n-except ImportError:\r\n-    GPU_MEMORY_TYPE_CUDA_GRAPH = None\r\n-\r\n:",,terminal_output
+7,144746,"TERMINAL",0,0,"...skipping...\r\ndiff --git a/train.py b/train.py\r\ndeleted file mode 100644\r\nindex 9fb480e..0000000\r\n--- a/train.py\r\n+++ /dev/null\r\n@@ -1,106 +0,0 @@\r\n-import ray\r\n-from sglang.srt.constants import GPU_MEMORY_TYPE_KV_CACHE, GPU_MEMORY_TYPE_WEIGHTS\r\n-\r\n-try:\r\n-    from sglang.srt.constants import GPU_MEMORY_TYPE_CUDA_GRAPH\r\n-except ImportError:\r\n-    GPU_MEMORY_TYPE_CUDA_GRAPH = None\r\n-\r\n-from miles.ray.placement_group import create_placement_groups, create_rollout_manager, create_training_models\r\n-from miles.utils.arguments import parse_args\r\n-from miles.utils.logging_utils import configure_logger\r\n-from miles.utils.misc import should_run_periodic_action\r\n:...skipping...\r\ndiff --git a/train.py b/train.py\r\ndeleted file mode 100644\r\nindex 9fb480e..0000000\r\n--- a/train.py\r\n+++ /dev/null\r\n@@ -1,106 +0,0 @@\r\n-import ray\r\n-from sglang.srt.constants import GPU_MEMORY_TYPE_KV_CACHE, GPU_MEMORY_TYPE_WEIGHTS\r\n-\r\n-try:\r\n-    from sglang.srt.constants import GPU_MEMORY_TYPE_CUDA_GRAPH\r\n-except ImportError:\r\n-    GPU_MEMORY_TYPE_CUDA_GRAPH = None\r\n-\r\n-from miles.ray.placement_group import create_placement_groups, create_rollout_manager, create_training_models\r\n-from miles.utils.arguments import parse_args\r\n-from miles.utils.logging_utils import configure_logger\r\n-from miles.utils.misc import should_run_periodic_action\r\n-from miles.utils.tracking_utils import init_tracking\r\n-\r\n-\r\n-def train(args):\r\n-    configure_logger()\r\n:...skipping...\r\ndiff --git a/train.py b/train.py\r\ndeleted file mode 100644\r\nindex 9fb480e..0000000\r\n--- a/train.py\r\n+++ /dev/null\r\n@@ -1,106 +0,0 @@\r\n-import ray\r\n-from sglang.srt.constants import GPU_MEMORY_TYPE_KV_CACHE, GPU_MEMORY_TYPE_WEIGHTS\r\n-\r\n-try:\r\n-    from sglang.srt.constants import GPU_MEMORY_TYPE_CUDA_GRAPH\r\n-except ImportError:\r\n-    GPU_MEMORY_TYPE_CUDA_GRAPH = None\r\n-\r\n-from miles.ray.placement_group import create_placement_groups, create_rollout_manager, create_training_models\r\n-from miles.utils.arguments import parse_args\r\n-from miles.utils.logging_utils import configure_logger\r\n-from miles.utils.misc import should_run_periodic_action\r\n-from miles.utils.tracking_utils import init_tracking\r\n-\r\n-\r\n-def train(args):\r\n-    configure_logger()\r\n-    # allocate the GPUs\r\n-    pgs = create_placement_groups(args)\r\n-    init_tracking(args)\r\n-\r\n-    # create the rollout manager, with sglang engines inside.\r\n-    # need to initialize rollout manager first to calculate num_rollout\r\n-    rollout_manager, num_rollout_per_epoch = create_rollout_manager(args, pgs[""rollout""])\r\n-\r\n-    # create the actor and critic models\r\n-    actor_model, critic_model = create_training_models(args, pgs, rollout_manager)\r\n-\r\n-    if args.offload_rollout:\r\n-        ray.get(rollout_manager.onload.remote(tags=[GPU_MEMORY_TYPE_WEIGHTS]))\r\n:...skipping...\r\ndiff --git a/train.py b/train.py\r\ndeleted file mode 100644\r\nindex 9fb480e..0000000\r\n--- a/train.py\r\n+++ /dev/null\r\n@@ -1,106 +0,0 @@\r\n-import ray\r\n-from sglang.srt.constants import GPU_MEMORY_TYPE_KV_CACHE, GPU_MEMORY_TYPE_WEIGHTS\r\n-\r\n-try:\r\n-    from sglang.srt.constants import GPU_MEMORY_TYPE_CUDA_GRAPH\r\n-except ImportError:\r\n-    GPU_MEMORY_TYPE_CUDA_GRAPH = None\r\n-\r\n-from miles.ray.placement_group import create_placement_groups, create_rollout_manager, create_training_models\r\n-from miles.utils.arguments import parse_args\r\n-from miles.utils.logging_utils import configure_logger\r\n-from miles.utils.misc import should_run_periodic_action\r\n-from miles.utils.tracking_utils import init_tracking\r\n-\r\n-\r\n-def train(args):\r\n-    configure_logger()\r\n-    # allocate the GPUs\r\n-    pgs = create_placement_groups(args)\r\n-    init_tracking(args)\r\n-\r\n-    # create the rollout manager, with sglang engines inside.\r\n-    # need to initialize rollout manager first to calculate num_rollout\r\n-    rollout_manager, num_rollout_per_epoch = create_rollout_manager(args, pgs[""rollout""])\r\n-\r\n-    # create the actor and critic models\r\n-    actor_model, critic_model = create_training_models(args, pgs, rollout_manager)\r\n-\r\n-    if args.offload_rollout:\r\n-        ray.get(rollout_manager.onload.remote(tags=[GPU_MEMORY_TYPE_WEIGHTS]))\r\n-\r\n-    # always update weight first so that sglang has the loaded weights from training.\r\n-    actor_model.update_weights()\r\n-\r\n:...skipping...\r\ndiff --git a/train.py b/train.py\r\ndeleted file mode 100644\r\nindex 9fb480e..0000000\r\n--- a/train.py\r\n+++ /dev/null\r\n@@ -1,106 +0,0 @@\r\n-import ray\r\n-from sglang.srt.constants import GPU_MEMORY_TYPE_KV_CACHE, GPU_MEMORY_TYPE_WEIGHTS\r\n-\r\n-try:\r\n-    from sglang.srt.constants import GPU_MEMORY_TYPE_CUDA_GRAPH\r\n-except ImportError:\r\n-    GPU_MEMORY_TYPE_CUDA_GRAPH = None\r\n-\r\n-from miles.ray.placement_group import create_placement_groups, create_rollout_manager, create_training_models\r\n-from miles.utils.arguments import parse_args\r\n-from miles.utils.logging_utils import configure_logger\r\n-from miles.utils.misc import should_run_periodic_action\r\n-from miles.utils.tracking_utils import init_tracking\r\n-\r\n-\r\n-def train(args):\r\n-    configure_logger()\r\n-    # allocate the GPUs\r\n-    pgs = create_placement_groups(args)\r\n-    init_tracking(args)\r\n-\r\n-    # create the rollout manager, with sglang engines inside.\r\n-    # need to initialize rollout manager first to calculate num_rollout\r\n-    rollout_manager, num_rollout_per_epoch = create_rollout_manager(args, pgs[""rollout""])\r\n-\r\n-    # create the actor and critic models\r\n-    actor_model, critic_model = create_training_models(args, pgs, rollout_manager)\r\n-\r\n-    if args.offload_rollout:\r\n-        ray.get(rollout_manager.onload.remote(tags=[GPU_MEMORY_TYPE_WEIGHTS]))\r\n-\r\n-    # always update weight first so that sglang has the loaded weights from training.\r\n-    actor_model.update_weights()\r\n-\r\n-    if args.check_weight_update_equal:\r\n-        ray.get(rollout_manager.check_weights.remote(action=""compare""))\r\n:...skipping...\r\ndiff --git a/train.py b/train.py\r\ndeleted file mode 100644\r\nindex 9fb480e..0000000\r\n--- a/train.py\r\n+++ /dev/null\r\n@@ -1,106 +0,0 @@\r\n-import ray\r\n-from sglang.srt.constants import GPU_MEMORY_TYPE_KV_CACHE, GPU_MEMORY_TYPE_WEIGHTS\r\n-\r\n-try:\r\n-    from sglang.srt.constants import GPU_MEMORY_TYPE_CUDA_GRAPH\r\n-except ImportError:\r\n-    GPU_MEMORY_TYPE_CUDA_GRAPH = None\r\n-\r\n-from miles.ray.placement_group import create_placement_groups, create_rollout_manager, create_training_models\r\n-from miles.utils.arguments import parse_args\r\n-from miles.utils.logging_utils import configure_logger\r\n-from miles.utils.misc import should_run_periodic_action\r\n-from miles.utils.tracking_utils import init_tracking\r\n-\r\n-\r\n-def train(args):\r\n-    configure_logger()\r\n-    # allocate the GPUs\r\n-    pgs = create_placement_groups(args)\r\n-    init_tracking(args)\r\n-\r\n-    # create the rollout manager, with sglang engines inside.\r\n-    # need to initialize rollout manager first to calculate num_rollout\r\n-    rollout_manager, num_rollout_per_epoch = create_rollout_manager(args, pgs[""rollout""])\r\n-\r\n-    # create the actor and critic models\r\n-    actor_model, critic_model = create_training_models(args, pgs, rollout_manager)\r\n-\r\n-    if args.offload_rollout:\r\n-        ray.get(rollout_manager.onload.remote(tags=[GPU_MEMORY_TYPE_WEIGHTS]))\r\n-\r\n-    # always update weight first so that sglang has the loaded weights from training.\r\n-    actor_model.update_weights()\r\n-\r\n-    if args.check_weight_update_equal:\r\n-        ray.get(rollout_manager.check_weights.remote(action=""compare""))\r\n-\r\n:...skipping...\r\ndiff --git a/train.py b/train.py\r\ndeleted file mode 100644\r\nindex 9fb480e..0000000\r\n--- a/train.py\r\n+++ /dev/null\r\n@@ -1,106 +0,0 @@\r\n-import ray\r\n-from sglang.srt.constants import GPU_MEMORY_TYPE_KV_CACHE, GPU_MEMORY_TYPE_WEIGHTS\r\n-\r\n-try:\r\n-    from sglang.srt.constants import GPU_MEMORY_TYPE_CUDA_GRAPH\r\n-except ImportError:\r\n-    GPU_MEMORY_TYPE_CUDA_GRAPH = None\r\n-\r\n-from miles.ray.placement_group import create_placement_groups, create_rollout_manager, create_training_models\r\n-from miles.utils.arguments import parse_args\r\n-from miles.utils.logging_utils import configure_logger\r\n-from miles.utils.misc import should_run_periodic_action\r\n-from miles.utils.tracking_utils import init_tracking\r\n-\r\n-\r\n-def train(args):\r\n-    configure_logger()\r\n-    # allocate the GPUs\r\n-    pgs = create_placement_groups(args)\r\n-    init_tracking(args)\r\n-\r\n-    # create the rollout manager, with sglang engines inside.\r\n-    # need to initialize rollout manager first to calculate num_rollout\r\n-    rollout_manager, num_rollout_per_epoch = create_rollout_manager(args, pgs[""rollout""])\r\n-\r\n-    # create the actor and critic models\r\n-    actor_model, critic_model = create_training_models(args, pgs, rollout_manager)\r\n-\r\n-    if args.offload_rollout:\r\n-        ray.get(rollout_manager.onload.remote(tags=[GPU_MEMORY_TYPE_WEIGHTS]))\r\n-\r\n-    # always update weight first so that sglang has the loaded weights from training.\r\n-    actor_model.update_weights()\r\n-\r\n-    if args.check_weight_update_equal:\r\n-        ray.get(rollout_manager.check_weights.remote(action=""compare""))\r\n-\r\n-    if args.offload_rollout:\r\n:",,terminal_output
+8,144862,"TERMINAL",0,0,"...skipping...\r\ndiff --git a/train.py b/train.py\r\ndeleted file mode 100644\r\nindex 9fb480e..0000000\r\n--- a/train.py\r\n+++ /dev/null\r\n@@ -1,106 +0,0 @@\r\n-import ray\r\n-from sglang.srt.constants import GPU_MEMORY_TYPE_KV_CACHE, GPU_MEMORY_TYPE_WEIGHTS\r\n-\r\n-try:\r\n-    from sglang.srt.constants import GPU_MEMORY_TYPE_CUDA_GRAPH\r\n-except ImportError:\r\n-    GPU_MEMORY_TYPE_CUDA_GRAPH = None\r\n-\r\n-from miles.ray.placement_group import create_placement_groups, create_rollout_manager, create_training_models\r\n-from miles.utils.arguments import parse_args\r\n-from miles.utils.logging_utils import configure_logger\r\n-from miles.utils.misc import should_run_periodic_action\r\n-from miles.utils.tracking_utils import init_tracking\r\n-\r\n-\r\n-def train(args):\r\n-    configure_logger()\r\n-    # allocate the GPUs\r\n-    pgs = create_placement_groups(args)\r\n-    init_tracking(args)\r\n-\r\n-    # create the rollout manager, with sglang engines inside.\r\n-    # need to initialize rollout manager first to calculate num_rollout\r\n-    rollout_manager, num_rollout_per_epoch = create_rollout_manager(args, pgs[""rollout""])\r\n-\r\n-    # create the actor and critic models\r\n-    actor_model, critic_model = create_training_models(args, pgs, rollout_manager)\r\n-\r\n-    if args.offload_rollout:\r\n-        ray.get(rollout_manager.onload.remote(tags=[GPU_MEMORY_TYPE_WEIGHTS]))\r\n-\r\n-    # always update weight first so that sglang has the loaded weights from training.\r\n-    actor_model.update_weights()\r\n-\r\n-    if args.check_weight_update_equal:\r\n-        ray.get(rollout_manager.check_weights.remote(action=""compare""))\r\n-\r\n-    if args.offload_rollout:\r\n-        if GPU_MEMORY_TYPE_CUDA_GRAPH is not None:\r\n:",,terminal_output
+9,145349,"TERMINAL",0,0,"\r-            ray.get(rollout_manager.onload.remote(tags=[GPU_MEMORY_TYPE_CUDA_GRAPH]))\r\n:",,terminal_output
+10,146004,"TERMINAL",0,0,"\rMdiff --git a/train.py b/train.py\r\n\r:",,terminal_output
+11,146352,"TERMINAL",0,0,"\r\r:",,terminal_output
+12,146548,"TERMINAL",0,0,"\r\r:",,terminal_output
+13,146884,"TERMINAL",0,0,"\r\r:",,terminal_output
+14,148417,"TERMINAL",0,0,"\r[?1l>]0;mahajan1@jwlogin21:~/projects/envcomp/miles",,terminal_output
+15,154778,"TERMINAL",0,0,"cd tab-model-eval/",,terminal_command
+16,154781,"TERMINAL",0,0,"]633;C]0;mahajan1@jwlogin21:~/projects/envcomp/miles/tab-model-eval",,terminal_output
+17,156747,"TERMINAL",0,0,"git diff",,terminal_command
+18,156780,"TERMINAL",0,0,"]633;C[?1h=\rdiff --git a/src/evaluation/sglang_generations.py b/src/evaluation/sglang_generations.py\r\nindex 554e563..822a89b 100644\r\n--- a/src/evaluation/sglang_generations.py\r\n+++ b/src/evaluation/sglang_generations.py\r\n@@ -39,10 +39,11 @@ class Args:\r\n     # Model-related\r\n     temperature: float = 0.7\r\n     top_p: float = 0.8\r\n-    presence_penalty: float = 1.5\r\n+    presence_penalty: float = 0.0\r\n     top_k: int = 20\r\n     min_p: float = 0.0\r\n     num_samples: int = 5\r\n+    max_new_tokens: int = 5000\r\n \r\n     # HTTP / client config\r\n     concurrency: int = 16\r\n@@ -144,6 +145,7 @@ async def generate_next_command(\r\n                     top_p=args.top_p,\r\n                     presence_penalty=args.presence_penalty,\r\n                     n=args.num_samples,\r\n+                    max_tokens=args.max_new_tokens,\r\n                     extra_body={\r\n                         ""top_k"": args.top_k,\r\n                     },\r\ndiff --git a/src/evaluation/sync_local_logs_to_wandb.py b/src/evaluation/sync_local_logs_to_wandb.py\r\nindex 50f9f10..c765579 100644\r\n--- a/src/evaluation/sync_local_logs_to_wandb.py\r\n+++ b/src/evaluation/sync_local_logs_to_wandb.py\r\n@@ -84,7 +84,7 @@ def sync_single_run(log_dir: str, dry_run: bool = False) -> bool:\r\n         run = wandb.init(\r\n             project=metadata[""project""],\r\n             name=metadata[""run_name""],\r\n-            id=metadata[""run_id""],\r\n+            id=f""evaluate_{metadata[""run_id""]}"",\r\n             config=metadata.get(""config"", {}),\r\n             tags=metadata.get(""tags"", []),\r\n             resume=""allow"",  # Resume if exists, create if not\r\n\r[?1l>]0;mahajan1@jwlogin21:~/projects/envcomp/miles/tab-model-eval",,terminal_output
+19,173057,"TERMINAL",0,0,"git diff",,terminal_command
+20,173060,"TERMINAL",0,0,"]633;C[?1h=\rdiff --git a/src/evaluation/sglang_generations.py b/src/evaluation/sglang_generations.py\r\nindex 554e563..822a89b 100644\r\n--- a/src/evaluation/sglang_generations.py\r\n+++ b/src/evaluation/sglang_generations.py\r\n@@ -39,10 +39,11 @@ class Args:\r\n     # Model-related\r\n     temperature: float = 0.7\r\n     top_p: float = 0.8\r\n-    presence_penalty: float = 1.5\r\n+    presence_penalty: float = 0.0\r\n     top_k: int = 20\r\n     min_p: float = 0.0\r\n     num_samples: int = 5\r\n+    max_new_tokens: int = 5000\r\n \r\n     # HTTP / client config\r\n     concurrency: int = 16\r\n@@ -144,6 +145,7 @@ async def generate_next_command(\r\n                     top_p=args.top_p,\r\n                     presence_penalty=args.presence_penalty,\r\n                     n=args.num_samples,\r\n+                    max_tokens=args.max_new_tokens,\r\n                     extra_body={\r\n                         ""top_k"": args.top_k,\r\n                     },\r\ndiff --git a/src/evaluation/sync_local_logs_to_wandb.py b/src/evaluation/sync_local_logs_to_wandb.py\r\nindex 50f9f10..c765579 100644\r\n--- a/src/evaluation/sync_local_logs_to_wandb.py\r\n+++ b/src/evaluation/sync_local_logs_to_wandb.py\r\n@@ -84,7 +84,7 @@ def sync_single_run(log_dir: str, dry_run: bool = False) -> bool:\r\n         run = wandb.init(\r\n             project=metadata[""project""],\r\n             name=metadata[""run_name""],\r\n-            id=metadata[""run_id""],\r\n+            id=f""evaluate_{metadata[""run_id""]}"",\r\n             config=metadata.get(""config"", {}),\r\n             tags=metadata.get(""tags"", []),\r\n             resume=""allow"",  # Resume if exists, create if not\r\n\r[?1l>]0;mahajan1@jwlogin21:~/projects/envcomp/miles/tab-model-eval",,terminal_output
+21,594981,"tab-model-eval/src/evaluation/sglang_eval.py",13453,0,"",python,selection_mouse
+22,594997,"tab-model-eval/src/evaluation/sglang_eval.py",13452,0,"",python,selection_command
+23,1090903,"tab-model-eval/src/evaluation/sglang_eval.py",13559,0,"",python,selection_mouse
+24,1217328,"tab-model-eval/src/evaluation/sglang_eval.py",13780,0,"",python,selection_mouse
+25,1217335,"tab-model-eval/src/evaluation/sglang_eval.py",13779,0,"",python,selection_command
+26,1217859,"tab-model-eval/src/evaluation/sglang_eval.py",13781,0,"",python,selection_mouse
+27,1219273,"tab-model-eval/src/evaluation/sglang_eval.py",13559,0,"",python,selection_mouse
+28,1220768,"tab-model-eval/src/evaluation/sglang_eval.py",13458,0,"",python,selection_mouse
+29,1220789,"tab-model-eval/src/evaluation/sglang_eval.py",13457,0,"",python,selection_command
+30,1222704,"tab-model-eval/src/evaluation/sglang_eval.py",13851,0,"",python,selection_mouse
+31,1224544,"tab-model-eval/src/evaluation/sglang_eval.py",14232,0,"",python,selection_mouse
+32,1226368,"tab-model-eval/src/evaluation/sglang_eval.py",14388,0,"",python,selection_mouse
+33,1227193,"tab-model-eval/src/evaluation/sglang_eval.py",14387,0,"",python,selection_mouse
+34,1227197,"tab-model-eval/src/evaluation/sglang_eval.py",14386,0,"",python,selection_command
+35,1227966,"tab-model-eval/src/evaluation/sglang_eval.py",14232,0,"",python,selection_mouse
+36,1228911,"tab-model-eval/src/evaluation/sglang_eval.py",13851,0,"",python,selection_mouse
+37,1230211,"tab-model-eval/src/evaluation/sglang_eval.py",13780,0,"",python,selection_mouse
+38,1230232,"tab-model-eval/src/evaluation/sglang_eval.py",13779,0,"",python,selection_command
+39,1230687,"tab-model-eval/src/evaluation/sglang_eval.py",13781,0,"",python,selection_mouse
+40,1231859,"tab-model-eval/src/evaluation/sglang_eval.py",13559,0,"",python,selection_mouse
+41,1285609,"tab-model-eval/src/evaluation/sync_local_logs_to_wandb.py",0,0,"#!/usr/bin/env python3\n""""""\nSync local evaluation logs to wandb.\n\nThis script reads local logs created by sglang_eval.py when using --use_local_logger\nand uploads them to wandb as a single run with all data points.\n\nUsage:\n    python sync_local_logs_to_wandb.py --log_dir data/eval/local_logs/eval_<RUN_ID>\n    \nOr to sync all runs in a directory:\n    python sync_local_logs_to_wandb.py --log_dir data/eval/local_logs --sync_all\n""""""\n\nimport argparse\nimport json\nimport os\nimport sys\nfrom pathlib import Path\n\nimport wandb\n\n\ndef load_metadata(log_dir: str) -> dict:\n    """"""Load run metadata from the log directory.""""""\n    metadata_file = os.path.join(log_dir, ""metadata.json"")\n    if not os.path.exists(metadata_file):\n        raise FileNotFoundError(f""Metadata file not found: {metadata_file}"")\n    \n    with open(metadata_file, ""r"") as f:\n        return json.load(f)\n\n\ndef load_metrics(log_dir: str) -> list:\n    """"""Load all metrics from the JSONL file.""""""\n    metrics_file = os.path.join(log_dir, ""metrics.jsonl"")\n    if not os.path.exists(metrics_file):\n        raise FileNotFoundError(f""Metrics file not found: {metrics_file}"")\n    \n    metrics = []\n    with open(metrics_file, ""r"") as f:\n        for line in f:\n            line = line.strip()\n            if line:\n                metrics.append(json.loads(line))\n    \n    return metrics\n\n\ndef sync_single_run(log_dir: str, dry_run: bool = False) -> bool:\n    """"""\n    Sync a single run's local logs to wandb.\n    \n    Returns True if successful, False otherwise.\n    """"""\n    print(f""\n{'='*60}"")\n    print(f""Syncing: {log_dir}"")\n    print(f""{'='*60}"")\n    \n    try:\n        # Load metadata and metrics\n        metadata = load_metadata(log_dir)\n        metrics = load_metrics(log_dir)\n        \n        if not metrics:\n            print(f""  Warning: No metrics found in {log_dir}"")\n            return False\n        \n        # Sort metrics by eval_step\n        metrics.sort(key=lambda x: x.get(""eval_step"", 0))\n        \n        print(f""  Run ID: {metadata['run_id']}"")\n        print(f""  Run Name: {metadata['run_name']}"")\n        print(f""  Project: {metadata['project']}"")\n        print(f""  Tags: {metadata.get('tags', [])}"")\n        print(f""  Data points: {len(metrics)}"")\n        print(f""  Steps: {[m.get('eval_step', 'N/A') for m in metrics]}"")\n        \n        if dry_run:\n            print(""  [DRY RUN] Would upload to wandb"")\n            return True\n        \n        # Initialize wandb run\n        run = wandb.init(\n            project=metadata[""project""],\n            name=metadata[""run_name""],\n            id=f""evaluate_{metadata[""run_id""]}"",\n            config=metadata.get(""config"", {}),\n            tags=metadata.get(""tags"", []),\n            resume=""allow"",  # Resume if exists, create if not\n        )\n        \n        # Log each metric with its step\n        for metric_entry in metrics:\n            # Remove timestamp for wandb logging\n            entry = {k: v for k, v in metric_entry.items() if k != ""timestamp""}\n            wandb.log(entry)\n        \n        wandb.finish()\n        print(f""  ✓ Successfully synced to wandb!"")\n        return True\n        \n    except Exception as e:\n        print(f""  ✗ Error syncing {log_dir}: {e}"")\n        return False\n\n\ndef find_run_dirs(base_dir: str) -> list:\n    """"""Find all run directories in the base directory.""""""\n    run_dirs = []\n    \n    for item in os.listdir(base_dir):\n        item_path = os.path.join(base_dir, item)\n        if os.path.isdir(item_path):\n            # Check if it's a valid run directory (has metadata.json)\n            metadata_file = os.path.join(item_path, ""metadata.json"")\n            if os.path.exists(metadata_file):\n                run_dirs.append(item_path)\n    \n    return sorted(run_dirs)\n\n\ndef main():\n    parser = argparse.ArgumentParser(\n        description=""Sync local evaluation logs to wandb"",\n        formatter_class=argparse.RawDescriptionHelpFormatter,\n        epilog=""""""\nExamples:\n    # Sync a single run\n    python sync_local_logs_to_wandb.py --log_dir data/eval/local_logs/eval_13032805\n    \n    # Sync all runs in a directory\n    python sync_local_logs_to_wandb.py --log_dir data/eval/local_logs --sync_all\n    \n    # Dry run to see what would be synced\n    python sync_local_logs_to_wandb.py --log_dir data/eval/local_logs --sync_all --dry_run\n""""""\n    )\n    \n    parser.add_argument(\n        ""--log_dir"",\n        type=str,\n        required=True,\n        help=""Path to the log directory (single run) or parent directory (with --sync_all)""\n    )\n    parser.add_argument(\n        ""--sync_all"",\n        action=""store_true"",\n        help=""Sync all runs found in the log_dir""\n    )\n    parser.add_argument(\n        ""--dry_run"",\n        action=""store_true"",\n        help=""Show what would be synced without actually uploading""\n    )\n    \n    args = parser.parse_args()\n    \n    if not os.path.exists(args.log_dir):\n        print(f""Error: Log directory not found: {args.log_dir}"")\n        sys.exit(1)\n    \n    if args.sync_all:\n        # Sync all runs in the directory\n        run_dirs = find_run_dirs(args.log_dir)\n        \n        if not run_dirs:\n            print(f""No valid run directories found in {args.log_dir}"")\n            sys.exit(1)\n        \n        print(f""Found {len(run_dirs)} run(s) to sync:"")\n        for d in run_dirs:\n            print(f""  - {os.path.basename(d)}"")\n        \n        success_count = 0\n        for run_dir in run_dirs:\n            if sync_single_run(run_dir, dry_run=args.dry_run):\n                success_count += 1\n        \n        print(f""\n{'='*60}"")\n        print(f""Sync complete: {success_count}/{len(run_dirs)} runs synced successfully"")\n        print(f""{'='*60}"")\n        \n    else:\n        # Sync single run\n        # Check if log_dir is the run directory or contains metadata.json\n        if os.path.exists(os.path.join(args.log_dir, ""metadata.json"")):\n            sync_single_run(args.log_dir, dry_run=args.dry_run)\n        else:\n            print(f""Error: {args.log_dir} does not appear to be a valid run directory"")\n            print(""  (missing metadata.json)"")\n            print(""\nHint: Use --sync_all to sync all runs in a parent directory"")\n            sys.exit(1)\n\n\nif __name__ == ""__main__"":\n    main()\n",python,tab
+42,1285723,"tab-model-eval/src/evaluation/sync_local_logs_to_wandb.py",2122,0,"",python,selection_mouse
+43,1285723,"tab-model-eval/src/evaluation/sync_local_logs_to_wandb.py",2121,0,"",python,selection_command
+44,1287340,"tab-model-eval/src/evaluation/sync_local_logs_to_wandb.py",2172,0,"",python,selection_command
+45,1287668,"tab-model-eval/src/evaluation/sync_local_logs_to_wandb.py",2222,0,"",python,selection_command
+46,1287669,"tab-model-eval/src/evaluation/sync_local_logs_to_wandb.py",2273,0,"",python,selection_command
+47,1287845,"tab-model-eval/src/evaluation/sync_local_logs_to_wandb.py",2305,0,"",python,selection_command
+48,1288024,"tab-model-eval/src/evaluation/sync_local_logs_to_wandb.py",2325,0,"",python,selection_command
+49,1288200,"tab-model-eval/src/evaluation/sync_local_logs_to_wandb.py",2376,0,"",python,selection_command
+50,1288227,"tab-model-eval/src/evaluation/sync_local_logs_to_wandb.py",2404,0,"",python,selection_command
+51,1288380,"tab-model-eval/src/evaluation/sync_local_logs_to_wandb.py",2413,0,"",python,selection_command
+52,1288508,"tab-model-eval/src/evaluation/sync_local_logs_to_wandb.py",2444,0,"",python,selection_command
+53,1288705,"tab-model-eval/src/evaluation/sync_local_logs_to_wandb.py",2470,0,"",python,selection_command
+54,1288886,"tab-model-eval/src/evaluation/sync_local_logs_to_wandb.py",2511,0,"",python,selection_command
+55,1288959,"tab-model-eval/src/evaluation/sync_local_logs_to_wandb.py",2550,0,"",python,selection_command
+56,1289067,"tab-model-eval/src/evaluation/sync_local_logs_to_wandb.py",2599,0,"",python,selection_command
+57,1289281,"tab-model-eval/src/evaluation/sync_local_logs_to_wandb.py",2646,0,"",python,selection_command
+58,1289664,"tab-model-eval/src/evaluation/sync_local_logs_to_wandb.py",2599,0,"",python,selection_command
+59,1290744,"tab-model-eval/src/evaluation/sync_local_logs_to_wandb.py",2618,0,"",python,selection_command
+60,1290929,"tab-model-eval/src/evaluation/sync_local_logs_to_wandb.py",2619,0,"",python,selection_command
+61,1292519,"tab-model-eval/src/evaluation/sync_local_logs_to_wandb.py",2570,0,"",python,selection_command
+62,1294253,"tab-model-eval/src/evaluation/sync_local_logs_to_wandb.py",2552,49,"",python,content
+63,1294259,"tab-model-eval/src/evaluation/sync_local_logs_to_wandb.py",2564,0,"",python,selection_command

58dff52cba2a091453cfbef6169091e684254819f0b9f334dbecea6a130284bc/crowd-code-f4bf9883-3801-446a-98c6-413295d94c701767091439743-2025_12_30-11.44.46.167/source.csv

@@ -0,0 +1,360 @@
+Sequence,Time,File,RangeOffset,RangeLength,Text,Language,Type
+2,266,"extension-output-pdoom-org.crowd-code-#1-crowd-code",0,0,"11:44:45 AM [info] Activating crowd-code\n11:44:46 AM [info] Recording started\n11:44:46 AM [info] Initializing git provider using file system watchers...\n11:44:46 AM [info] Git repository found\n11:44:46 AM [info] Git provider initialized successfully\n11:44:46 AM [info] Initial git state: [object Object]\n",Log,tab
+3,32786,"TERMINAL",0,0,"watch -n1 squeue --me",,terminal_command
+4,32830,"TERMINAL",0,0,"]633;C",,terminal_output
+5,33102,"TERMINAL",0,0,"[?1049h(B[?7hEvery 1.0s: squeue --mejwlogin23.juwels: Tue Dec 30 11:45:18 2025JOBID PARTITION     NAME     USER ST\tTIME  NODES NODELIST(REASON)13032808   booster qwen_32b mahajan1 PD\t0:00     64 (Resources)13032806   booster qwen_32b mahajan1  R    8:12:30     64 jwb[0200-0204,0213-0217,0219-0224,0226-0236,0245-0264,0266-0268,0277-0285,0287-0291]",,terminal_output
+6,34350,"TERMINAL",0,0,"201",,terminal_output
+7,35610,"TERMINAL",0,0,"12",,terminal_output
+8,36722,"TERMINAL",0,0,"[?1049l\r[?1l>]0;mahajan1@jwlogin23:~/projects/mahajan1/miles",,terminal_output
+9,37647,"TERMINAL",0,0,"squeue --me -o ""%.10i %.16P %.60j %.8u %.8T %.10M %.9l %.6D %R""",,terminal_command
+10,37695,"TERMINAL",0,0,"]633;C",,terminal_output
+11,37896,"TERMINAL",0,0,"     JOBID        PARTITION                                                         NAME     USER    STATE       TIME TIME_LIMI  NODES NODELIST(REASON)\r\n  13032808          booster                                       qwen_32b_no_lora_bs_32 mahajan1  PENDING       0:00 1-00:00:00     64 (Resources)\r\n  13032806          booster                                          qwen_32b_lora_bs_32 mahajan1  RUNNING    8:12:35 1-00:00:00     64 jwb[0200-0204,0213-0217,0219-0224,0226-0236,0245-0264,0266-0268,0277-0285,0287-0291]\r\n]0;mahajan1@jwlogin23:~/projects/mahajan1/miles",,terminal_output
+12,58404,"TERMINAL",0,0,"bash",,terminal_focus
+13,58830,"TERMINAL",0,0,"bash",,terminal_focus
+14,59615,"TERMINAL",0,0,"scancel 13032808",,terminal_command
+15,59662,"TERMINAL",0,0,"]633;C",,terminal_output
+16,59854,"TERMINAL",0,0,"]0;mahajan1@jwlogin23:~/projects/mahajan1/miles",,terminal_output
+17,68172,"TERMINAL",0,0,"sacct --format=""JobID%15,JobName%30,Partition%16,AllocCPUS%3,State%12,Elapsed%10,Timelimit%10"" --starttime $(date -d ""last week"" +%Y-%m-%d)",,terminal_command
+18,68218,"TERMINAL",0,0,"]633;C",,terminal_output
+19,68474,"TERMINAL",0,0,"          JobID                        JobName        Partition All        State    Elapsed  Timelimit \r\n--------------- ------------------------------ ---------------- --- ------------ ---------- ---------- \r\n     13024059.0                             sh                   16       FAILED   00:00:22            \r\n     13024059.1                             sh                   16       FAILED   00:01:49            \r\n     13024059.2                             sh                   16       FAILED   00:03:59            \r\n     13024095.0                       torchrun                   16       FAILED   00:00:15            \r\n     13024095.2                             sh                   16    COMPLETED   00:00:03            \r\n     13024095.4                             sh                   16       FAILED   00:06:09            \r\n     13024095.7                             sh                   16       FAILED   00:00:35            \r\n     13024209.0                             sh                   16       FAILED   00:03:47            \r\n       13024229            qwen_1.7b_lora_bs_8          booster   0       FAILED   00:00:00 1-00:00:00 \r\n       13024250           qwen_1.7b_lora_bs_16          booster 192       FAILED   00:15:49 1-00:00:00 \r\n     13024250.0                             sh                   20       FAILED   00:15:27            \r\n       13024251           qwen_1.7b_lora_bs_32          booster 192       FAILED   00:08:02 1-00:00:00 \r\n     13024251.0                             sh                   20       FAILED   00:07:41            \r\n       13024252            qwen_1.7b_lora_bs_8          booster 192       FAILED   00:15:38 1-00:00:00 \r\n     13024252.0                             sh                   20       FAILED   00:15:17            \r\n       13024253           qwen_1.7b_lora_bs_16          booster 192    COMPLETED   06:02:31 1-00:00:00 \r\n     13024253.0                             sh                   20    COMPLETED   06:02:09            \r\n       13024254           qwen_1.7b_lora_bs_32          booster 192       FAILED   00:06:09 1-00:00:00 \r\n     13024254.0                             sh                   20       FAILED   00:05:48            \r\n       13024255            qwen_1.7b_lora_bs_8          booster 192       FAILED   00:01:48 1-00:00:00 \r\n     13024255.0                             sh                   20       FAILED   00:01:27            \r\n       13024256             qwen_4b_lora_bs_16          booster 384       FAILED   00:15:39 1-00:00:00 \r\n     13024256.0                             sh                   40       FAILED   00:15:17            \r\n       13024257             qwen_4b_lora_bs_32          booster 384       FAILED   00:06:48 1-00:00:00 \r\n     13024257.0                             sh                   40       FAILED   00:06:26            \r\n       13024258              qwen_4b_lora_bs_8          booster 384       FAILED   00:03:43 1-00:00:00 \r\n     13024258.0                             sh                   40       FAILED   00:03:21            \r\n       13024259             qwen_4b_lora_bs_16          booster 384       FAILED   00:05:27 1-00:00:00 \r\n     13024259.0                             sh                   40       FAILED   00:05:05            \r\n       13024260             qwen_4b_lora_bs_32          booster 384       FAILED   00:06:27 1-00:00:00 \r\n     13024260.0                             sh                   40       FAILED   00:06:06            \r\n       13024261              qwen_4b_lora_bs_8          booster 384       FAILED   00:04:01 1-00:00:00 \r\n     13024261.0                             sh                   40       FAILED   00:03:39            \r\n       13025339           qwen_1.7b_lora_bs_16          booster 192    COMPLETED   04:13:45 1-00:00:00 \r\n     13025339.0                             sh                   20    COMPLETED   04:13:21            \r\n       13025340           qwen_1.7b_lora_bs_32          booster 192       FAILED   00:07:09 1-00:00:00 \r\n     13025340.0                             sh                   20       FAILED   00:06:47            \r\n       13025341            qwen_1.7b_lora_bs_8          booster 192    COMPLETED   04:18:27 1-00:00:00 \r\n     13025341.0                             sh                   20    COMPLETED   04:18:04            \r\n       13025342           qwen_1.7b_lora_bs_16          booster 192    COMPLETED   05:58:27 1-00:00:00 \r\n     13025342.0                             sh                   20    COMPLETED   05:58:04            \r\n       13025343           qwen_1.7b_lora_bs_32          booster 192       FAILED   00:05:17 1-00:00:00 \r\n     13025343.0                             sh                   20       FAILED   00:04:55            \r\n       13025344            qwen_1.7b_lora_bs_8          booster 192       FAILED   00:00:49 1-00:00:00 \r\n     13025344.0                             sh                   20       FAILED   00:00:27            \r\n       13025345             qwen_4b_lora_bs_16          booster 384       FAILED   00:32:33 1-00:00:00 \r\n     13025345.0                             sh                   40       FAILED   00:32:11            \r\n       13025346             qwen_4b_lora_bs_32          booster 384       FAILED   00:05:49 1-00:00:00 \r\n     13025346.0                             sh                   40       FAILED   00:05:26            \r\n       13025347              qwen_4b_lora_bs_8          booster 384       FAILED   00:02:41 1-00:00:00 \r\n     13025347.0                             sh                   40       FAILED   00:02:19            \r\n       13025348             qwen_4b_lora_bs_16          booster 384       FAILED   00:04:21 1-00:00:00 \r\n     13025348.0                             sh                   40       FAILED   00:03:59            \r\n       13025349             qwen_4b_lora_bs_32          booster 384       FAILED   00:05:28 1-00:00:00 \r\n     13025349.0                             sh                   40       FAILED   00:05:05            \r\n       13025350              qwen_4b_lora_bs_8          booster 384       FAILED   00:02:59 1-00:00:00 \r\n     13025350.0                             sh                   40       FAILED   00:02:37            \r\n       13025365            qwen_1.7b_lora_bs_8          booster 192    COMPLETED   04:20:44 1-00:00:00 \r\n     13025365.0                             sh                   20    COMPLETED   04:20:20            \r\n     13030348.0                             sh                   16       FAILED   00:06:31            \r\n     13030348.3                             sh                   16       FAILED   00:03:42            \r\n       13030386           qwen_1.7b_lora_bs_16          booster 192    COMPLETED   03:52:07 1-00:00:00 \r\n     13030386.0                             sh                   20    COMPLETED   03:51:44            \r\n       13030387           qwen_1.7b_lora_bs_32          booster 384    COMPLETED   04:40:04 1-00:00:00 \r\n     13030387.0                             sh                   40    COMPLETED   04:39:40            \r\n       13030388            qwen_1.7b_lora_bs_8          booster 192    COMPLETED   04:01:53 1-00:00:00 \r\n     13030388.0                             sh                   20    COMPLETED   04:01:31            \r\n       13030389           qwen_1.7b_lora_bs_16          booster 192    COMPLETED   05:40:57 1-00:00:00 \r\n     13030389.0                             sh                   20    COMPLETED   05:40:33            \r\n       13030390           qwen_1.7b_lora_bs_32          booster 384    COMPLETED   03:42:05 1-00:00:00 \r\n     13030390.0                             sh                   40    COMPLETED   03:41:41            \r\n       13030391            qwen_1.7b_lora_bs_8          booster 192    COMPLETED   05:55:16 1-00:00:00 \r\n     13030391.0                             sh                   20    COMPLETED   05:54:52            \r\n       13030392             qwen_4b_lora_bs_16          booster 384    COMPLETED   04:41:52 1-00:00:00 \r\n     13030392.0                             sh                   40    COMPLETED   04:41:31            \r\n       13030393             qwen_4b_lora_bs_32          booster 768    COMPLETED   03:11:49 1-00:00:00 \r\n     13030393.0                             sh                   80    COMPLETED   03:11:27            \r\n       13030394              qwen_4b_lora_bs_8          booster 192    COMPLETED   07:12:58 1-00:00:00 \r\n     13030394.0                             sh                   20    COMPLETED   07:12:37            \r\n       13030395             qwen_4b_lora_bs_16          booster 384    COMPLETED   07:03:11 1-00:00:00 \r\n     13030395.0                             sh                   40    COMPLETED   07:02:50            \r\n       13030396             qwen_4b_lora_bs_32          booster 768    COMPLETED   04:35:42 1-00:00:00 \r\n     13030396.0                             sh                   80    COMPLETED   04:35:20            \r\n       13030397              qwen_4b_lora_bs_8          booster 192    COMPLETED   14:08:39 1-00:00:00 \r\n     13030397.0                             sh                   20    COMPLETED   14:08:18            \r\n       13030398             qwen_8b_lora_bs_16          booster 384    COMPLETED   07:15:28 1-00:00:00 \r\n     13030398.0                             sh                   40    COMPLETED   07:15:06            \r\n       13030399             qwen_8b_lora_bs_32          booster 768    COMPLETED   04:40:58 1-00:00:00 \r\n     13030399.0                             sh                   80    COMPLETED   04:40:36            \r\n       13030400              qwen_8b_lora_bs_8          booster 192    COMPLETED   11:59:57 1-00:00:00 \r\n     13030400.0                             sh                   20    COMPLETED   11:59:36            \r\n       13030401             qwen_8b_lora_bs_16          booster 384       FAILED   00:06:33 1-00:00:00 \r\n     13030401.0                             sh                   40       FAILED   00:06:12            \r\n       13030402             qwen_8b_lora_bs_32          booster 768       FAILED   00:05:45 1-00:00:00 \r\n     13030402.0                             sh                   80       FAILED   00:05:23            \r\n       13030403              qwen_8b_lora_bs_8          booster 192       FAILED   00:06:14 1-00:00:00 \r\n     13030403.0                             sh                   20       FAILED   00:05:52            \r\n     13031068.0                             sh                   16       FAILED   00:04:00            \r\n     13031068.1                             sh                   16       FAILED   00:02:57            \r\n     13031068.2                             sh                   16       FAILED   00:02:54            \r\n     13031093.1                             sh                   32       FAILED   00:00:52            \r\n     13031093.2                             sh                   32       FAILED   00:00:51            \r\n     13031093.3                             sh                   32       FAILED   00:00:49            \r\n     13031093.4                             sh                   32       FAILED   00:00:48            \r\n     13031854.0                             sh                   40       FAILED   00:52:06            \r\n     13031856.0                             sh                  160       FAILED   00:17:59            \r\n     13031902.4                             sh                   32       FAILED   00:04:16            \r\n       13031978           qwen_8b_no_lora_bs_8          booster 384       FAILED   01:09:06 1-00:00:00 \r\n     13031978.0                             sh                   40       FAILED   01:08:44            \r\n       13031979          qwen_8b_no_lora_bs_16          booster 768       FAILED   12:08:49 1-00:00:00 \r\n     13031979.0                             sh                   80       FAILED   12:08:24            \r\n       13031981          qwen_8b_no_lora_bs_32          booster 1536       FAILED   07:43:12 1-00:00:00 \r\n     13031981.0                             sh                  160       FAILED   07:42:47            \r\n       13031982             qwen_32b_lora_bs_8          booster 768       FAILED   03:27:04 1-00:00:00 \r\n     13031982.0                             sh                   80       FAILED   03:26:39            \r\n       13032780           qwen_8b_no_lora_bs_8          booster 384       FAILED   17:33:36 1-00:00:00 \r\n     13032780.0                             sh                   40       FAILED   17:33:12            \r\n       13032800           qwen_1.7b_lora_bs_16          booster 192    COMPLETED   03:54:13 1-00:00:00 \r\n     13032800.0                             sh                   20    COMPLETED   03:53:47            \r\n       13032801           qwen_1.7b_lora_bs_32          booster 384    COMPLETED   04:40:04 1-00:00:00 \r\n     13032801.0                             sh                   40    COMPLETED   04:39:41            \r\n       13032802            qwen_1.7b_lora_bs_8          booster 192    COMPLETED   04:04:53 1-00:00:00 \r\n     13032802.0                             sh                   20    COMPLETED   04:04:28            \r\n       13032803           qwen_1.7b_lora_bs_16          booster 192    COMPLETED   05:39:56 1-00:00:00 \r\n     13032803.0                             sh                   20    COMPLETED   05:39:30            \r\n       13032804           qwen_1.7b_lora_bs_32          booster 384    COMPLETED   03:43:42 1-00:00:00 \r\n     13032804.0                             sh                   40    COMPLETED   03:43:18            \r\n       13032805            qwen_1.7b_lora_bs_8          booster 192    COMPLETED   05:56:43 1-00:00:00 \r\n     13032805.0                             sh                   20    COMPLETED   05:56:19            \r\n       13032806            qwen_32b_lora_bs_32          booster 6144      RUNNING   08:13:05 1-00:00:00 \r\n     13032806.0                             sh                  640      RUNNING   08:12:42            \r\n       13032807             qwen_32b_lora_bs_8          booster 768      TIMEOUT 1-00:00:30 1-00:00:00 \r\n       13032809             qwen_4b_lora_bs_16          booster 384    COMPLETED   04:47:21 1-00:00:00 \r\n     13032809.0                             sh                   40    COMPLETED   04:46:57            \r\n       13032810             qwen_4b_lora_bs_32          booster 768    COMPLETED   03:11:21 1-00:00:00 \r\n     13032810.0                             sh                   80    COMPLETED   03:10:55            \r\n       13032811              qwen_4b_lora_bs_8          booster 192    COMPLETED   07:12:50 1-00:00:00 \r\n     13032811.0                             sh                   20    COMPLETED   07:12:27            \r\n       13032812             qwen_4b_lora_bs_16          booster 384    COMPLETED   07:06:10 1-00:00:00 \r\n     13032812.0                             sh                   40    COMPLETED   07:05:47            \r\n       13032813             qwen_4b_lora_bs_32          booster 768    COMPLETED   04:43:02 1-00:00:00 \r\n     13032813.0                             sh                   80    COMPLETED   04:42:39            \r\n       13032814              qwen_4b_lora_bs_8          booster 192    COMPLETED   13:51:27 1-00:00:00 \r\n     13032814.0                             sh                   20    COMPLETED   13:51:03            \r\n       13032815             qwen_8b_lora_bs_16          booster 384    COMPLETED   07:21:14 1-00:00:00 \r\n     13032815.0                             sh                   40    COMPLETED   07:20:52            \r\n       13032816             qwen_8b_lora_bs_32          booster 768    COMPLETED   04:44:36 1-00:00:00 \r\n     13032816.0                             sh                   80    COMPLETED   04:44:13            \r\n       13032817              qwen_8b_lora_bs_8          booster 192    COMPLETED   12:05:55 1-00:00:00 \r\n     13032817.0                             sh                   20    COMPLETED   12:05:33            \r\n       13032818          qwen_8b_no_lora_bs_16          booster 768    COMPLETED   08:07:43 1-00:00:00 \r\n     13032818.0                             sh                   80    COMPLETED   08:07:20            \r\n       13032820          qwen_8b_no_lora_bs_32          booster 1536    COMPLETED   04:46:00 1-00:00:00 \r\n     13032820.0                             sh                  160    COMPLETED   04:45:35            \r\n       13032821           qwen_8b_no_lora_bs_8          booster 384    COMPLETED   17:30:39 1-00:00:00 \r\n     13032821.0                             sh                   40    COMPLETED   17:30:13            \r\n       13034357          qwen_8b_no_lora_bs_32          booster 6144    COMPLETED   07:14:13 1-00:00:00 \r\n     13034357.0                             sh                  640    COMPLETED   07:13:46            \r\n]0;mahajan1@jwlogin23:~/projects/mahajan1/miles",,terminal_output
+20,636927,"TERMINAL",0,0,"git diff",,terminal_command
+21,637017,"TERMINAL",0,0,"]633;C[?1h=\rdiff --git a/miles/backends/fsdp_utils/actor.py b/miles/backends/fsdp_utils/actor.py\r\nindex f500a88..57cc178 100644\r\n--- a/miles/backends/fsdp_utils/actor.py\r\n+++ b/miles/backends/fsdp_utils/actor.py\r\n@@ -973,7 +973,10 @@ def get_logprob_and_entropy_with_cp(\r\n         )\r\n         log_probs_full = torch.log_softmax(shifted_logits, dim=-1)\r\n         probs = torch.softmax(shifted_logits, dim=-1)\r\n-        entropy = -(probs * log_probs_full).sum(dim=-1)\r\n+        # mentropy = -(probs * log_probs_full).sum(dim=-1)\r\n+        p_log_p = probs * log_probs_full\r\n+        p_log_p = torch.nan_to_num(p_log_p, nan=0.0) \r\n+        entropy = -p_log_p.sum(dim=-1)\r\n         return local_log_probs, entropy\r\n \r\n     chunk_size = logits.shape[0]\r\n@@ -1003,7 +1006,10 @@ def get_logprob_and_entropy_with_cp(\r\n     shifted_logits = logits[:-1, :] if cp_rank == cp_size - 1 else logits\r\n     log_probs_full = torch.log_softmax(shifted_logits, dim=-1)\r\n     probs = torch.softmax(shifted_logits, dim=-1)\r\n-    entropy = -(probs * log_probs_full).sum(dim=-1)\r\n+    # entropy = -(probs * log_probs_full).sum(dim=-1)\r\n+    p_log_p = probs * log_probs_full\r\n+    p_log_p = torch.nan_to_num(p_log_p, nan=0.0) \r\n+    entropy = -p_log_p.sum(dim=-1)\r\n \r\n     # Pad entropy for the last rank\r\n     if cp_rank == cp_size - 1:\r\ndiff --git a/miles/backends/fsdp_utils/checkpoint.py b/miles/backends/fsdp_utils/checkpoint.py\r\nindex 3846bd9..758ccec 100644\r\n--- a/miles/backends/fsdp_utils/checkpoint.py\r\n+++ b/miles/backends/fsdp_utils/checkpoint.py\r\n@@ -25,16 +25,17 @@ class ModelState(Stateful):\r\n         self.keys_filter = keys_filter\r\n \r\n     def state_dict(self):\r\n-        model_state_dict, _ = get_state_dict(self.model, optimizers=[])\r\n+        options = StateDictOptions(full_state_dict=False, cpu_offload=True)\r\n+        model_state_dict, _ = get_state_dict(self.model, optimizers=[], options=options)\r\n         if self.keys_filter:\r\n             model_state_dict = {k: v for k, v in model_state_dict.items() if self.keys_filter(k)}\r\n         return {""model"": model_state_dict}\r\n \r\n:",,terminal_output
+22,667321,"TERMINAL",0,0,"\r     def load_state_dict(self, state_dict):\r\n:",,terminal_output
+23,667610,"TERMINAL",0,0,"\r-        options = None\r\n:",,terminal_output
+24,667810,"TERMINAL",0,0,"\r+        options = StateDictOptions(cpu_offload=True)\r\n:",,terminal_output
+25,667947,"TERMINAL",0,0,"\r         if self.keys_filter:\r\n:",,terminal_output
+26,668099,"TERMINAL",0,0,"\r             # For filtered loading (e.g., LoRA), use strict=False to allow partial loading\r\n:",,terminal_output
+27,668262,"TERMINAL",0,0,"\r-            options = StateDictOptions(strict=False)\r\n:",,terminal_output
+28,669549,"TERMINAL",0,0,"\r+            options.strict = False\r\n:",,terminal_output
+29,671741,"TERMINAL",0,0,"\r/\r         probs = torch.softmax(shifted_logits, dim=-1)\r\n-        entropy = -(probs * log_probs_full).sum(dim=-1)\r\n+        # mentropy = -(probs * log_probs_full).sum(dim=-1)\r\n+        p_log_p = probs * log_probs_full\r\n+        p_log_p = torch.nan_to_num(p_log_p, nan=0.0) \r\n+        entropy = -p_log_p.sum(dim=-1)\r\n         return local_log_probs, entropy\r\n \r\n     chunk_size = logits.shape[0]\r\n@@ -1003,7 +1006,10 @@ def get_logprob_and_entropy_with_cp(\r\n     shifted_logits = logits[:-1, :] if cp_rank == cp_size - 1 else logits\r\n     log_probs_full = torch.log_softmax(shifted_logits, dim=-1)\r\n     probs = torch.softmax(shifted_logits, dim=-1)\r\n-    entropy = -(probs * log_probs_full).sum(dim=-1)\r\n+    # entropy = -(probs * log_probs_full).sum(dim=-1)\r\n+    p_log_p = probs * log_probs_full\r\n+    p_log_p = torch.nan_to_num(p_log_p, nan=0.0) \r\n+    entropy = -p_log_p.sum(dim=-1)\r\n \r\n     # Pad entropy for the last rank\r\n     if cp_rank == cp_size - 1:\r\ndiff --git a/miles/backends/fsdp_utils/checkpoint.py b/miles/backends/fsdp_utils/checkpoint.py\r\nindex 3846bd9..758ccec 100644\r\n--- a/miles/backends/fsdp_utils/checkpoint.py\r\n+++ b/miles/backends/fsdp_utils/checkpoint.py\r\n@@ -25,16 +25,17 @@ class ModelState(Stateful):\r\n         self.keys_filter = keys_filter\r\n \r\n     def state_dict(self):\r\n-        model_state_dict, _ = get_state_dict(self.model, optimizers=[])\r\n+        options = StateDictOptions(full_state_dict=False, cpu_offload=True)\r\n+        model_state_dict, _ = get_state_dict(self.model, optimizers=[], options=options)\r\n         if self.keys_filter:\r\n             model_state_dict = {k: v for k, v in model_state_dict.items() if self.keys_filter(k)}\r\n         return {""model"": model_state_dict}\r\n \r\n     def load_state_dict(self, state_dict):\r\n-        options = None\r\n+        options = StateDictOptions(cpu_offload=True)\r\n         if self.keys_filter:\r\n             # For filtered loading (e.g., LoRA), use strict=False to allow partial loading\r\n-            options = StateDictOptions(strict=False)\r\n+            options.strict = False\r\n         set_state_dict(\r\n             self.model, optimizers=[], \r\n             model_state_dict=state_dict[""model""], \r\n@@ -52,16 +53,17 @@ class OptimizerState(Stateful):\r\n         self.keys_filter = keys_filter\r\n \r\n     def state_dict(self):\r\n-        _, optimizer_state_dict = get_state_dict(self.model, optimizers=self.optimizer)\r\n+        options = StateDictOptions(full_state_dict=False, cpu_offload=True)\r\n+        _, optimizer_state_dict = get_state_dict(self.model, optimizers=self.optimizer, options=options)\r\n         if self.keys_filter:\r\n             optimizer_state_dict = {k: v for k, v in optimizer_state_dict.items() if self.keys_filter(k)}\r\n         return {""optim"": optimizer_state_dict}\r\n \r\n     def load_state_dict(self, state_dict):\r\n-        options = None\r\n+        options = StateDictOptions(cpu_offload=True)\r\n         if self.keys_filter:\r\n             # For filtered loading (e.g., LoRA), use strict=False to allow partial loading\r\n-            options = StateDictOptions(strict=False)\r\n+            options.strict = False\r\n         set_state_dict(\r\n             self.model, optimizers=self.optimizer, \r\n             model_state_dict=None, \r\n@@ -137,12 +139,14 @@ def load(actor: Any) -> dict[str, Any] | None:\r\n         keys_filter = lambda k: ""lora_"" in k\r\n         logger.info(""[FSDP] LoRA mode: loading only LoRA weights from checkpoint"")\r\n \r\n+    dp_group = getattr(actor, ""dp_group"", None)\r\n+\r\n:",,terminal_output
+30,671917,"TERMINAL",0,0,"\r/\r     # Load model weights (always)\r\n     model_state = ModelState(actor.model, keys_filter=keys_filter)\r\n     state_dict = {""model_state"": model_state}\r\n \r\n     try:\r\n-        dcp.load(state_dict=state_dict, checkpoint_id=str(model_dir))\r\n+        dcp.load(state_dict=state_dict, checkpoint_id=str(model_dir), process_group=dp_group)\r\n         logger.info(f""[FSDP] Loaded model from {model_dir}"")\r\n:",,terminal_output
+31,672882,"TERMINAL",0,0,"\r     except Exception as e:\r\n:",,terminal_output
+32,673068,"TERMINAL",0,0,"\r         logger.error(f""[FSDP] Failed to load model from {model_dir}: {e}"")\r\n:",,terminal_output
+33,673274,"TERMINAL",0,0,"\r@@ -154,7 +158,7 @@ def load(actor: Any) -> dict[str, Any] | None:\r\n:",,terminal_output
+34,673736,"TERMINAL",0,0,"\r         optimizer_state = OptimizerState(actor.model, actor.optimizer, keys_filter=keys_filter)\r\n:",,terminal_output
+35,673888,"TERMINAL",0,0,"\r         optim_state_dict = {""optim_state"": optimizer_state}\r\n:",,terminal_output
+36,674054,"TERMINAL",0,0,"\rM+            options.strict = False\r\n\r:",,terminal_output
+37,674195,"TERMINAL",0,0,"\rM-            options = StateDictOptions(strict=False)\r\n\r:",,terminal_output
+38,674369,"TERMINAL",0,0,"\rM             # For filtered loading (e.g., LoRA), use strict=False to allow partial loading\r\n\r:",,terminal_output
+39,674526,"TERMINAL",0,0,"\rM         if self.keys_filter:\r\n\r:",,terminal_output
+40,674685,"TERMINAL",0,0,"\rM+        options = StateDictOptions(cpu_offload=True)\r\n\r:",,terminal_output
+41,674867,"TERMINAL",0,0,"\rM-        options = None\r\n\r:",,terminal_output
+42,675016,"TERMINAL",0,0,"\rM     def load_state_dict(self, state_dict):\r\n\r:",,terminal_output
+43,675255,"TERMINAL",0,0,"\rM \r\n\r:\r-        dcp.load(state_dict=state_dict, checkpoint_id=str(model_dir))\r\n:",,terminal_output
+44,675356,"TERMINAL",0,0,"\r+        dcp.load(state_dict=state_dict, checkpoint_id=str(model_dir), process_group=dp_group)\r\n:",,terminal_output
+45,675536,"TERMINAL",0,0,"\r         logger.info(f""[FSDP] Loaded model from {model_dir}"")\r\n:",,terminal_output
+46,675686,"TERMINAL",0,0,"\r     except Exception as e:\r\n:",,terminal_output
+47,675855,"TERMINAL",0,0,"\r         logger.error(f""[FSDP] Failed to load model from {model_dir}: {e}"")\r\n:",,terminal_output
+48,676331,"TERMINAL",0,0,"\r@@ -154,7 +158,7 @@ def load(actor: Any) -> dict[str, Any] | None:\r\n:",,terminal_output
+49,676333,"TERMINAL",0,0,"\r         optimizer_state = OptimizerState(actor.model, actor.optimizer, keys_filter=keys_filter)\r\n:",,terminal_output
+50,676456,"TERMINAL",0,0,"\r         optim_state_dict = {""optim_state"": optimizer_state}\r\n:",,terminal_output
+51,676528,"TERMINAL",0,0,"\r         try:\r\n:",,terminal_output
+52,678986,"TERMINAL",0,0,"\r-            dcp.load(state_dict=optim_state_dict, checkpoint_id=str(optimizer_dir))\r\n:",,terminal_output
+53,679154,"TERMINAL",0,0,"\r+            dcp.load(state_dict=optim_state_dict, checkpoint_id=str(optimizer_dir), process_group=dp_group)\r\n:",,terminal_output
+54,679330,"TERMINAL",0,0,"\r             logger.info(f""[FSDP] Loaded optimizer from {optimizer_dir}"")\r\n:",,terminal_output
+55,680566,"TERMINAL",0,0,"\r         except Exception as e:\r\n:",,terminal_output
+56,680751,"TERMINAL",0,0,"\r             logger.warning(f""[FSDP] Failed to load optimizer from {optimizer_dir}: {e}"")\r\n:",,terminal_output
+57,680959,"TERMINAL",0,0,"\r@@ -167,7 +171,7 @@ def load(actor: Any) -> dict[str, Any] | None:\r\n:",,terminal_output
+58,681132,"TERMINAL",0,0,"\r         lr_scheduler_state = LRSchedulerState(actor.lr_scheduler)\r\n:",,terminal_output
+59,681510,"TERMINAL",0,0,"\r         lr_scheduler_state_dict = {""lr_scheduler_state"": lr_scheduler_state}\r\n:",,terminal_output
+60,681691,"TERMINAL",0,0,"\r         try:\r\n:",,terminal_output
+61,681898,"TERMINAL",0,0,"\r-            dcp.load(state_dict=lr_scheduler_state_dict, checkpoint_id=str(lr_scheduler_dir))\r\n:",,terminal_output
+62,683126,"TERMINAL",0,0,"\r+            dcp.load(state_dict=lr_scheduler_state_dict, checkpoint_id=str(lr_scheduler_dir), process_group=dp_group)\r\n:",,terminal_output
+63,683584,"TERMINAL",0,0,"\r             logger.info(f""[FSDP] Loaded LR scheduler from {lr_scheduler_dir}"")\r\n:",,terminal_output
+64,684050,"TERMINAL",0,0,"\r         except Exception as e:\r\n:",,terminal_output
+65,689555,"TERMINAL",0,0,"\r             logger.warning(f""[FSDP] Failed to load LR scheduler from {lr_scheduler_dir}: {e}"")\r\n:",,terminal_output
+66,689744,"TERMINAL",0,0,"\r@@ -238,22 +242,29 @@ def save(actor: Any, iteration: int, keys_filter=None) -> None:\r\n:",,terminal_output
+67,689842,"TERMINAL",0,0,"\r         lr_scheduler_dir.mkdir(parents=True, exist_ok=True)\r\n:",,terminal_output
+68,689991,"TERMINAL",0,0,"\r     dist.barrier()\r\n:",,terminal_output
+69,690170,"TERMINAL",0,0,"\r \r\n:",,terminal_output
+70,690374,"TERMINAL",0,0,"\r+    cp_size = getattr(actor, ""cp_size"", 1)\r\n:",,terminal_output
+71,690578,"TERMINAL",0,0,"\r+    cp_rank = getattr(actor, ""cp_rank"", 0)\r\n:",,terminal_output
+72,690764,"TERMINAL",0,0,"\r+    dp_group = getattr(actor, ""dp_group"", None)\r\n:",,terminal_output
+73,690910,"TERMINAL",0,0,"\r+\r\n:",,terminal_output
+74,691091,"TERMINAL",0,0,"\r     # Save model weights\r\n:",,terminal_output
+75,691283,"TERMINAL",0,0,"\r-    model_state = ModelState(actor.model, keys_filter=keys_filter)\r\n:",,terminal_output
+76,691440,"TERMINAL",0,0,"\r-    state_dict = {""model_state"": model_state}\r\n:",,terminal_output
+77,691614,"TERMINAL",0,0,"\r-    dcp.save(state_dict, checkpoint_id=str(model_dir))\r\n:",,terminal_output
+78,691932,"TERMINAL",0,0,"\r+    if cp_rank == 0:\r\n:",,terminal_output
+79,692156,"TERMINAL",0,0,"\r+        model_state = ModelState(actor.model, keys_filter=keys_filter)\r\n:",,terminal_output
+80,692216,"TERMINAL",0,0,"\r+        state_dict = {""model_state"": model_state}\r\n:",,terminal_output
+81,692405,"TERMINAL",0,0,"\r+        dcp.save(state_dict, checkpoint_id=str(model_dir), process_group=dp_group)\r\n:",,terminal_output
+82,692546,"TERMINAL",0,0,"\r \r\n:",,terminal_output
+83,692692,"TERMINAL",0,0,"\r     # Save optimizer state\r\n:",,terminal_output
+84,692864,"TERMINAL",0,0,"\r     if hasattr(actor, ""optimizer"") and actor.optimizer is not None:\r\n:",,terminal_output
+85,692992,"TERMINAL",0,0,"\r-        optimizer_state = OptimizerState(actor.model, actor.optimizer, keys_filter=keys_filter)\r\n:",,terminal_output
+86,693163,"TERMINAL",0,0,"\r-        optim_state_dict = {""optim_state"": optimizer_state}\r\n:",,terminal_output
+87,693318,"TERMINAL",0,0,"\r-        dcp.save(optim_state_dict, checkpoint_id=str(optimizer_dir))\r\n:",,terminal_output
+88,693476,"TERMINAL",0,0,"\r+        if cp_rank == 0:\r\n:",,terminal_output
+89,693632,"TERMINAL",0,0,"\r+            optimizer_state = OptimizerState(actor.model, actor.optimizer, keys_filter=keys_filter)\r\n:",,terminal_output
+90,693796,"TERMINAL",0,0,"\r+            optim_state_dict = {""optim_state"": optimizer_state}\r\n:",,terminal_output
+91,693948,"TERMINAL",0,0,"\r+            dcp.save(optim_state_dict, checkpoint_id=str(optimizer_dir), process_group=dp_group)\r\n:",,terminal_output
+92,694169,"TERMINAL",0,0,"\r \r\n:",,terminal_output
+93,694342,"TERMINAL",0,0,"\r     # Save LR scheduler state\r\n:",,terminal_output
+94,694546,"TERMINAL",0,0,"\r     if hasattr(actor, ""lr_scheduler"") and actor.lr_scheduler is not None:\r\n:",,terminal_output
+95,694704,"TERMINAL",0,0,"\r-        lr_scheduler_state = LRSchedulerState(actor.lr_scheduler)\r\n:",,terminal_output
+96,694886,"TERMINAL",0,0,"\r-        lr_scheduler_state_dict = {""lr_scheduler_state"": lr_scheduler_state}\r\n:",,terminal_output
+97,695066,"TERMINAL",0,0,"\r-        dcp.save(lr_scheduler_state_dict, checkpoint_id=str(lr_scheduler_dir))\r\n:",,terminal_output
+98,695258,"TERMINAL",0,0,"\r+        if cp_rank == 0:\r\n:",,terminal_output
+99,695477,"TERMINAL",0,0,"\r+            lr_scheduler_state = LRSchedulerState(actor.lr_scheduler)\r\n:",,terminal_output
+100,696626,"TERMINAL",0,0,"\r+            lr_scheduler_state_dict = {""lr_scheduler_state"": lr_scheduler_state}\r\n:",,terminal_output
+101,696810,"TERMINAL",0,0,"\r+            dcp.save(lr_scheduler_state_dict, checkpoint_id=str(lr_scheduler_dir), process_group=dp_group)\r\n:",,terminal_output
+102,697057,"TERMINAL",0,0,"\r \r\n:",,terminal_output
+103,697176,"TERMINAL",0,0,"\r     if dist.get_rank() == 0:\r\n:",,terminal_output
+104,697392,"TERMINAL",0,0,"\r         rng_state = {""torch"": torch.get_rng_state()}\r\n:",,terminal_output
+105,697841,"TERMINAL",0,0,"\rM-            dcp.load(state_dict=lr_scheduler_state_dict, checkpoint_id=str(lr_scheduler_dir))\r\n\r:",,terminal_output
+106,698011,"TERMINAL",0,0,"\rM         try:\r\n\r:",,terminal_output
+107,698192,"TERMINAL",0,0,"\rM         lr_scheduler_state_dict = {""lr_scheduler_state"": lr_scheduler_state}\r\n\r:",,terminal_output
+108,698360,"TERMINAL",0,0,"\rM         lr_scheduler_state = LRSchedulerState(actor.lr_scheduler)\r\n\r:",,terminal_output
+109,698514,"TERMINAL",0,0,"\rM@@ -167,7 +171,7 @@ def load(actor: Any) -> dict[str, Any] | None:\r\n\r:",,terminal_output
+110,698770,"TERMINAL",0,0,"\rM             logger.warning(f""[FSDP] Failed to load optimizer from {optimizer_dir}: {e}"")\r\n\r:",,terminal_output
+111,700540,"TERMINAL",0,0,"\r+            lr_scheduler_state = LRSchedulerState(actor.lr_scheduler)\r\n:",,terminal_output
+112,700714,"TERMINAL",0,0,"\r+            lr_scheduler_state_dict = {""lr_scheduler_state"": lr_scheduler_state}\r\n:",,terminal_output
+113,700880,"TERMINAL",0,0,"\r+            dcp.save(lr_scheduler_state_dict, checkpoint_id=str(lr_scheduler_dir), process_group=dp_group)\r\n:",,terminal_output
+114,701043,"TERMINAL",0,0,"\r \r\n:",,terminal_output
+115,701235,"TERMINAL",0,0,"\r     if dist.get_rank() == 0:\r\n:",,terminal_output
+116,701443,"TERMINAL",0,0,"\r         rng_state = {""torch"": torch.get_rng_state()}\r\n:",,terminal_output
+117,701578,"TERMINAL",0,0,"\rdiff --git a/train_sft.py b/train_sft.py\r\n:",,terminal_output
+118,701927,"TERMINAL",0,0,"\rindex e494e0d..1f9422a 100644\r\n:",,terminal_output
+119,702105,"TERMINAL",0,0,"\r--- a/train_sft.py\r\n:",,terminal_output
+120,702288,"TERMINAL",0,0,"\r+++ b/train_sft.py\r\n:",,terminal_output
+121,702489,"TERMINAL",0,0,"\r@@ -536,7 +536,7 @@ class SFTTrainer:\r\n:",,terminal_output
+122,702811,"TERMINAL",0,0,"\r         """"""Execute one training step.""""""\r\n:",,terminal_output
+123,702969,"TERMINAL",0,0,"\r         # Prepare model inputs\r\n:\r         model_args = self._get_model_inputs_args(packed_batch)\r\n:",,terminal_output
+124,703176,"TERMINAL",0,0,"\r-        logits = self.model(**model_args).logits.squeeze(0).float()\r\n:",,terminal_output
+125,703331,"TERMINAL",0,0,"\r+        logits = self.model(**model_args).logits.squeeze(0)\r\n:",,terminal_output
+126,703528,"TERMINAL",0,0,"\r \r\n:",,terminal_output
+127,703651,"TERMINAL",0,0,"\r         # Compute log probs and entropy (unified for both CP and non-CP modes)\r\n:",,terminal_output
+128,703831,"TERMINAL",0,0,"\r         log_probs, entropy_result = get_logprob_and_entropy_with_cp(\r\n:",,terminal_output
+129,704083,"TERMINAL",0,0,"\r@@ -662,7 +662,7 @@ class SFTTrainer:\r\n:",,terminal_output
+130,704199,"TERMINAL",0,0,"\r     def _val_step(self, packed_batch):\r\n:",,terminal_output
+131,704477,"TERMINAL",0,0,"\r         model_args = self._get_model_inputs_args(packed_batch)\r\n:",,terminal_output
+132,704756,"TERMINAL",0,0,"\r         with torch.no_grad():\r\n:",,terminal_output
+133,704855,"TERMINAL",0,0,"\r-            logits = self.model(**model_args).logits.squeeze(0).float()\r\n:",,terminal_output
+134,705001,"TERMINAL",0,0,"\r+            logits = self.model(**model_args).logits.squeeze(0)\r\n:",,terminal_output
+135,705140,"TERMINAL",0,0,"\r \r\n:",,terminal_output
+136,705631,"TERMINAL",0,0,"\r         # Compute log probs and entropy (unified for both CP and non-CP modes)\r\n:\r         log_probs, entropy_result = get_logprob_and_entropy_with_cp(\r\n:\r@@ -718,6 +718,8 @@ class SFTTrainer:\r\n:",,terminal_output
+137,705803,"TERMINAL",0,0,"\r             if should_run_periodic_action(\r\n:",,terminal_output
+138,705960,"TERMINAL",0,0,"\r                 rollout_id, self.args.save_interval, self.num_rollout_per_epoch\r\n:",,terminal_output
+139,706149,"TERMINAL",0,0,"\r             ):\r\n:",,terminal_output
+140,706304,"TERMINAL",0,0,"\r+                torch.cuda.empty_cache() # <--- ADD THIS LINE\r\n:",,terminal_output
+141,706464,"TERMINAL",0,0,"\r+                torch.distributed.barrier() # <--- ADD THIS LINE (ensures all ranks pause here)\r\n:",,terminal_output
+142,706632,"TERMINAL",0,0,"\r                 self.save_model(rollout_id)\r\n:",,terminal_output
+143,706823,"TERMINAL",0,0,"\r \r\n:",,terminal_output
+144,707000,"TERMINAL",0,0,"\r             # Calculate val loss periodically\r\n(END)",,terminal_output
+145,707156,"TERMINAL",0,0,"\r\r(END)",,terminal_output
+146,707327,"TERMINAL",0,0,"\r\r(END)",,terminal_output
+147,707491,"TERMINAL",0,0,"\r\r(END)",,terminal_output
+148,707911,"TERMINAL",0,0,"\r\r(END)\r\r(END)",,terminal_output
+149,708124,"TERMINAL",0,0,"\r\r(END)",,terminal_output
+150,708223,"TERMINAL",0,0,"\r\r(END)",,terminal_output
+151,708411,"TERMINAL",0,0,"\r\r(END)",,terminal_output
+152,708580,"TERMINAL",0,0,"\r\r(END)",,terminal_output
+153,708763,"TERMINAL",0,0,"\r\r(END)",,terminal_output
+154,708944,"TERMINAL",0,0,"\r\r(END)",,terminal_output
+155,709113,"TERMINAL",0,0,"\r\r(END)",,terminal_output
+156,709281,"TERMINAL",0,0,"\r\r(END)",,terminal_output
+157,709456,"TERMINAL",0,0,"\r\r(END)",,terminal_output
+158,709627,"TERMINAL",0,0,"\r\r(END)",,terminal_output
+159,709793,"TERMINAL",0,0,"\r\r(END)",,terminal_output
+160,709955,"TERMINAL",0,0,"\r\r(END)",,terminal_output
+161,710142,"TERMINAL",0,0,"\r\r(END)",,terminal_output
+162,710310,"TERMINAL",0,0,"\r\r(END)",,terminal_output
+163,710474,"TERMINAL",0,0,"\r\r(END)",,terminal_output
+164,710662,"TERMINAL",0,0,"\r\r(END)",,terminal_output
+165,710847,"TERMINAL",0,0,"\r\r(END)",,terminal_output
+166,711034,"TERMINAL",0,0,"\r\r(END)",,terminal_output
+167,711180,"TERMINAL",0,0,"\r\r(END)",,terminal_output
+168,711340,"TERMINAL",0,0,"\r\r(END)",,terminal_output
+169,711526,"TERMINAL",0,0,"\r\r(END)",,terminal_output
+170,711703,"TERMINAL",0,0,"\r\r(END)",,terminal_output
+171,711880,"TERMINAL",0,0,"\r\r(END)",,terminal_output
+172,712094,"TERMINAL",0,0,"\r\r(END)",,terminal_output
+173,712287,"TERMINAL",0,0,"\r\r(END)",,terminal_output
+174,712502,"TERMINAL",0,0,"\r\r(END)",,terminal_output
+175,712686,"TERMINAL",0,0,"\r\r(END)",,terminal_output
+176,712936,"TERMINAL",0,0,"\r\r(END)",,terminal_output
+177,713827,"TERMINAL",0,0,"\rM \r\n\r:",,terminal_output
+178,714201,"TERMINAL",0,0,"\r             # Calculate val loss periodically\r\n(END)",,terminal_output
+179,714621,"TERMINAL",0,0,"\rM \r\n\r:",,terminal_output
+180,714822,"TERMINAL",0,0,"\r             # Calculate val loss periodically\r\n(END)",,terminal_output
+181,715054,"TERMINAL",0,0,"\rM \r\n\r:",,terminal_output
+182,715224,"TERMINAL",0,0,"\r             # Calculate val loss periodically\r\n(END)",,terminal_output
+183,715416,"TERMINAL",0,0,"\rM \r\n\r:",,terminal_output
+184,715542,"TERMINAL",0,0,"\r             # Calculate val loss periodically\r\n(END)",,terminal_output
+185,715703,"TERMINAL",0,0,"\rM \r\n\r:",,terminal_output
+186,715845,"TERMINAL",0,0,"\r             # Calculate val loss periodically\r\n(END)",,terminal_output
+187,715987,"TERMINAL",0,0,"\rM \r\n\r:",,terminal_output
+188,716103,"TERMINAL",0,0,"\r             # Calculate val loss periodically\r\n(END)",,terminal_output
+189,716246,"TERMINAL",0,0,"\rM \r\n\r:",,terminal_output
+190,716320,"TERMINAL",0,0,"\r             # Calculate val loss periodically\r\n(END)",,terminal_output
+191,1085087,"TERMINAL",0,0,"\rM \r\n\r:",,terminal_output
+192,1085463,"TERMINAL",0,0,"\r             # Calculate val loss periodically\r\n(END)",,terminal_output
+193,1085642,"TERMINAL",0,0,"\r\r(END)",,terminal_output
+194,1090638,"TERMINAL",0,0,"\r[?1l>]0;mahajan1@jwlogin23:~/projects/mahajan1/miles",,terminal_output
+195,1101278,"TERMINAL",0,0,"cd ${SCRATCH}",,terminal_command
+196,1101284,"TERMINAL",0,0,"]633;C]0;mahajan1@jwlogin23:/p/scratch/envcomp",,terminal_output
+197,1101627,"TERMINAL",0,0,"ls --color=auto",,terminal_command
+198,1101648,"TERMINAL",0,0,"]633;Cidm  logs  mihir  nguyen31  yll\r\n]0;mahajan1@jwlogin23:/p/scratch/envcomp",,terminal_output
+199,1115895,"TERMINAL",0,0,"cd mihir/huggingface/shared_data/",,terminal_command
+200,1115899,"TERMINAL",0,0,"]633;C]0;mahajan1@jwlogin23:/p/scratch/envcomp/mihir/huggingface/shared_data",,terminal_output
+201,1117660,"TERMINAL",0,0,"l",,terminal_command
+202,1117667,"TERMINAL",0,0,"]633;Cbash: l: command not found\r\n]0;mahajan1@jwlogin23:/p/scratch/envcomp/mihir/huggingface/shared_data",,terminal_output
+203,1118314,"TERMINAL",0,0,"ls --color=auto",,terminal_command
+204,1118319,"TERMINAL",0,0,"]633;C13024253  13025345  13030388  13030392  13030396  13030400  13031982  13032802  13032806  13032811  13032815  13032820\r\n13025339  13025365  13030389  13030393  13030397  13031978  13032780  13032803  13032807  13032812  13032816  13032821\r\n13025341  13030386  13030390  13030394  13030398  13031979  13032800  13032804  13032809  13032813  13032817  13034357\r\n13025342  13030387  13030391  13030395  13030399  13031981  13032801  13032805  13032810  13032814  13032818  qwen3-600M-fsdp-1116-noref\r\n]0;mahajan1@jwlogin23:/p/scratch/envcomp/mihir/huggingface/shared_data",,terminal_output
+205,1123484,"TERMINAL",0,0,"cd 13034357",,terminal_command
+206,1124592,"TERMINAL",0,0,"ls --color=auto",,terminal_command
+207,1124605,"TERMINAL",0,0,"]633;Ccheckpoints\r\n]0;mahajan1@jwlogin23:/p/scratch/envcomp/mihir/huggingface/shared_data/13034357",,terminal_output
+208,1126601,"TERMINAL",0,0,"cd checkpoints/",,terminal_command
+209,1126898,"TERMINAL",0,0,"ls --color=auto",,terminal_command
+210,1126969,"TERMINAL",0,0,"]633;Citer_0001000  iter_0002000  iter_0003000  iter_0004000  iter_0005000  latest_checkpointed_iteration.txt  rollout\r\n]0;mahajan1@jwlogin23:/p/scratch/envcomp/mihir/huggingface/shared_data/13034357/checkpoints",,terminal_output
+211,1130166,"TERMINAL",0,0,"ls --color=auto iter_0001000/",,terminal_command
+212,1130169,"TERMINAL",0,0,"]633;Clr_scheduler  meta.json  model  optimizer  rng.pt\r\n]0;mahajan1@jwlogin23:/p/scratch/envcomp/mihir/huggingface/shared_data/13034357/checkpoints",,terminal_output
+213,1306846,"TERMINAL",0,0,"ls --color=auto",,terminal_command
+214,1306849,"TERMINAL",0,0,"]633;Citer_0001000  iter_0002000  iter_0003000  iter_0004000  iter_0005000  latest_checkpointed_iteration.txt  rollout\r\n]0;mahajan1@jwlogin23:/p/scratch/envcomp/mihir/huggingface/shared_data/13034357/checkpoints",,terminal_output
+215,1309990,"TERMINAL",0,0,"du -h .",,terminal_command
+216,1310043,"TERMINAL",0,0,"]633;C",,terminal_output
+217,1310226,"TERMINAL",0,0,"40K\t./iter_0003000/lr_scheduler\r\n",,terminal_output
+218,1310370,"TERMINAL",0,0,"62G\t./iter_0003000/optimizer\r\n",,terminal_output
+219,1310513,"TERMINAL",0,0,"31G\t./iter_0003000/model\r\n92G\t./iter_0003000\r\n",,terminal_output
+220,1310776,"TERMINAL",0,0,"40K\t./iter_0004000/lr_scheduler\r\n62G\t./iter_0004000/optimizer\r\n31G\t./iter_0004000/model\r\n92G\t./iter_0004000\r\n3.0K\t./rollout\r\n",,terminal_output
+221,1310892,"TERMINAL",0,0,"40K\t./iter_0005000/lr_scheduler\r\n62G\t./iter_0005000/optimizer\r\n",,terminal_output
+222,1310944,"TERMINAL",0,0,"31G\t./iter_0005000/model\r\n92G\t./iter_0005000\r\n",,terminal_output
+223,1311035,"TERMINAL",0,0,"40K\t./iter_0001000/lr_scheduler\r\n",,terminal_output
+224,1311105,"TERMINAL",0,0,"62G\t./iter_0001000/optimizer\r\n",,terminal_output
+225,1311234,"TERMINAL",0,0,"31G\t./iter_0001000/model\r\n92G\t./iter_0001000\r\n40K\t./iter_0002000/lr_scheduler\r\n62G\t./iter_0002000/optimizer\r\n31G\t./iter_0002000/model\r\n92G\t./iter_0002000\r\n459G\t.\r\n]0;mahajan1@jwlogin23:/p/scratch/envcomp/mihir/huggingface/shared_data/13034357/checkpoints",,terminal_output
+226,1316248,"TERMINAL",0,0,"cd ..",,terminal_command
+227,1316250,"TERMINAL",0,0,"]633;C]0;mahajan1@jwlogin23:/p/scratch/envcomp/mihir/huggingface/shared_data/13034357",,terminal_output
+228,1316616,"TERMINAL",0,0,"ls --color=auto",,terminal_command
+229,1316624,"TERMINAL",0,0,"]633;Ccheckpoints\r\n]0;mahajan1@jwlogin23:/p/scratch/envcomp/mihir/huggingface/shared_data/13034357",,terminal_output
+230,1317548,"TERMINAL",0,0,"cd ..",,terminal_command
+231,1317790,"TERMINAL",0,0,"ls --color=auto",,terminal_command
+232,1317811,"TERMINAL",0,0,"]633;C13024253  13025345  13030388  13030392  13030396  13030400  13031982  13032802  13032806  13032811  13032815  13032820\r\n13025339  13025365  13030389  13030393  13030397  13031978  13032780  13032803  13032807  13032812  13032816  13032821\r\n13025341  13030386  13030390  13030394  13030398  13031979  13032800  13032804  13032809  13032813  13032817  13034357\r\n13025342  13030387  13030391  13030395  13030399  13031981  13032801  13032805  13032810  13032814  13032818  qwen3-600M-fsdp-1116-noref\r\n]0;mahajan1@jwlogin23:/p/scratch/envcomp/mihir/huggingface/shared_data",,terminal_output
+233,1350656,"TERMINAL",0,0,"cd 13032805",,terminal_command
+234,1351095,"TERMINAL",0,0,"ls --color=auto",,terminal_command
+235,1351111,"TERMINAL",0,0,"]633;Ccheckpoints\r\n]0;mahajan1@jwlogin23:/p/scratch/envcomp/mihir/huggingface/shared_data/13032805",,terminal_output
+236,1353850,"TERMINAL",0,0,"du -h .",,terminal_command
+237,1353898,"TERMINAL",0,0,"]633;C",,terminal_output
+238,1354011,"TERMINAL",0,0,"36K\t./checkpoints/iter_0014000/lr_scheduler\r\n",,terminal_output
+239,1354098,"TERMINAL",0,0,"13G\t./checkpoints/iter_0014000/optimizer\r\n",,terminal_output
+240,1354277,"TERMINAL",0,0,"7.6G\t./checkpoints/iter_0014000/model\r\n21G\t./checkpoints/iter_0014000\r\n36K\t./checkpoints/iter_0015000/lr_scheduler\r\n13G\t./checkpoints/iter_0015000/optimizer\r\n7.6G\t./checkpoints/iter_0015000/model\r\n21G\t./checkpoints/iter_0015000\r\n",,terminal_output
+241,1354394,"TERMINAL",0,0,"36K\t./checkpoints/iter_0007000/lr_scheduler\r\n",,terminal_output
+242,1354469,"TERMINAL",0,0,"13G\t./checkpoints/iter_0007000/optimizer\r\n7.6G\t./checkpoints/iter_0007000/model\r\n21G\t./checkpoints/iter_0007000\r\n",,terminal_output
+243,1354534,"TERMINAL",0,0,"36K\t./checkpoints/iter_0003000/lr_scheduler\r\n",,terminal_output
+244,1354590,"TERMINAL",0,0,"13G\t./checkpoints/iter_0003000/optimizer\r\n",,terminal_output
+245,1354684,"TERMINAL",0,0,"7.6G\t./checkpoints/iter_0003000/model\r\n21G\t./checkpoints/iter_0003000\r\n",,terminal_output
+246,1354740,"TERMINAL",0,0,"36K\t./checkpoints/iter_0011000/lr_scheduler\r\n",,terminal_output
+247,1355757,"TERMINAL",0,0,"13G\t./checkpoints/iter_0011000/optimizer\r\n7.6G\t./checkpoints/iter_0011000/model\r\n21G\t./checkpoints/iter_0011000\r\n36K\t./checkpoints/iter_0013000/lr_scheduler\r\n13G\t./checkpoints/iter_0013000/optimizer\r\n7.6G\t./checkpoints/iter_0013000/model\r\n21G\t./checkpoints/iter_0013000\r\n36K\t./checkpoints/iter_0017000/lr_scheduler\r\n13G\t./checkpoints/iter_0017000/optimizer\r\n7.6G\t./checkpoints/iter_0017000/model\r\n21G\t./checkpoints/iter_0017000\r\n36K\t./checkpoints/iter_0012000/lr_scheduler\r\n13G\t./checkpoints/iter_0012000/optimizer\r\n7.6G\t./checkpoints/iter_0012000/model\r\n21G\t./checkpoints/iter_0012000\r\n36K\t./checkpoints/iter_0016000/lr_scheduler\r\n13G\t./checkpoints/iter_0016000/optimizer\r\n7.6G\t./checkpoints/iter_0016000/model\r\n21G\t./checkpoints/iter_0016000\r\n36K\t./checkpoints/iter_0008000/lr_scheduler\r\n13G\t./checkpoints/iter_0008000/optimizer\r\n7.6G\t./checkpoints/iter_0008000/model\r\n21G\t./checkpoints/iter_0008000\r\n36K\t./checkpoints/iter_0020000/lr_scheduler\r\n13G\t./checkpoints/iter_0020000/optimizer\r\n7.6G\t./checkpoints/iter_0020000/model\r\n21G\t./checkpoints/iter_0020000\r\n36K\t./checkpoints/iter_0010000/lr_scheduler\r\n13G\t./checkpoints/iter_0010000/optimizer\r\n7.6G\t./checkpoints/iter_0010000/model\r\n21G\t./checkpoints/iter_0010000\r\n36K\t./checkpoints/iter_0018000/lr_scheduler\r\n13G\t./checkpoints/iter_0018000/optimizer\r\n7.6G\t./checkpoints/iter_0018000/model\r\n21G\t./checkpoints/iter_0018000\r\n36K\t./checkpoints/iter_0004000/lr_scheduler\r\n13G\t./checkpoints/iter_0004000/optimizer\r\n7.6G\t./checkpoints/iter_0004000/model\r\n21G\t./checkpoints/iter_0004000\r\n36K\t./checkpoints/iter_0009000/lr_scheduler\r\n13G\t./checkpoints/iter_0009000/optimizer\r\n7.6G\t./checkpoints/iter_0009000/model\r\n21G\t./checkpoints/iter_0009000\r\n11K\t./checkpoints/rollout\r\n",,terminal_output
+248,1355997,"TERMINAL",0,0,"36K\t./checkpoints/iter_0005000/lr_scheduler\r\n13G\t./checkpoints/iter_0005000/optimizer\r\n7.6G\t./checkpoints/iter_0005000/model\r\n21G\t./checkpoints/iter_0005000\r\n36K\t./checkpoints/iter_0019000/lr_scheduler\r\n13G\t./checkpoints/iter_0019000/optimizer\r\n7.6G\t./checkpoints/iter_0019000/model\r\n21G\t./checkpoints/iter_0019000\r\n",,terminal_output
+249,1356176,"TERMINAL",0,0,"36K\t./checkpoints/iter_0001000/lr_scheduler\r\n13G\t./checkpoints/iter_0001000/optimizer\r\n7.6G\t./checkpoints/iter_0001000/model\r\n21G\t./checkpoints/iter_0001000\r\n36K\t./checkpoints/iter_0002000/lr_scheduler\r\n13G\t./checkpoints/iter_0002000/optimizer\r\n7.6G\t./checkpoints/iter_0002000/model\r\n21G\t./checkpoints/iter_0002000\r\n36K\t./checkpoints/iter_0006000/lr_scheduler\r\n13G\t./checkpoints/iter_0006000/optimizer\r\n7.6G\t./checkpoints/iter_0006000/model\r\n21G\t./checkpoints/iter_0006000\r\n409G\t./checkpoints\r\n409G\t.\r\n]0;mahajan1@jwlogin23:/p/scratch/envcomp/mihir/huggingface/shared_data/13032805",,terminal_output
+250,1379857,"TERMINAL",0,0,"cd ..",,terminal_command
+251,1379863,"TERMINAL",0,0,"]633;C]0;mahajan1@jwlogin23:/p/scratch/envcomp/mihir/huggingface/shared_data",,terminal_output
+252,1380382,"TERMINAL",0,0,"sl",,terminal_command
+253,1380389,"TERMINAL",0,0,"]633;Cbash: sl: command not found\r\n]0;mahajan1@jwlogin23:/p/scratch/envcomp/mihir/huggingface/shared_data",,terminal_output
+254,1381246,"TERMINAL",0,0,"ls --color=auto",,terminal_command
+255,1381264,"TERMINAL",0,0,"]633;C13024253  13025345  13030388  13030392  13030396  13030400  13031982  13032802  13032806  13032811  13032815  13032820\r\n13025339  13025365  13030389  13030393  13030397  13031978  13032780  13032803  13032807  13032812  13032816  13032821\r\n13025341  13030386  13030390  13030394  13030398  13031979  13032800  13032804  13032809  13032813  13032817  13034357\r\n13025342  13030387  13030391  13030395  13030399  13031981  13032801  13032805  13032810  13032814  13032818  qwen3-600M-fsdp-1116-noref\r\n]0;mahajan1@jwlogin23:/p/scratch/envcomp/mihir/huggingface/shared_data",,terminal_output
+256,1383247,"TERMINAL",0,0,"cd 13032801",,terminal_command
+257,1383654,"TERMINAL",0,0,"ls --color=auto",,terminal_command
+258,1383660,"TERMINAL",0,0,"]633;Ccheckpoints\r\n]0;mahajan1@jwlogin23:/p/scratch/envcomp/mihir/huggingface/shared_data/13032801",,terminal_output
+259,1386050,"TERMINAL",0,0,"du -h .",,terminal_command
+260,1386098,"TERMINAL",0,0,"]633;C",,terminal_output
+261,1386235,"TERMINAL",0,0,"40K\t./checkpoints/iter_0004500/lr_scheduler\r\n",,terminal_output
+262,1386350,"TERMINAL",0,0,"17K\t./checkpoints/iter_0004500/optimizer\r\n",,terminal_output
+263,1386414,"TERMINAL",0,0,"15M\t./checkpoints/iter_0004500/model\r\n26M\t./checkpoints/iter_0004500\r\n",,terminal_output
+264,1386558,"TERMINAL",0,0,"40K\t./checkpoints/iter_0003000/lr_scheduler\r\n",,terminal_output
+265,1386652,"TERMINAL",0,0,"17K\t./checkpoints/iter_0003000/optimizer\r\n",,terminal_output
+266,1386746,"TERMINAL",0,0,"15M\t./checkpoints/iter_0003000/model\r\n26M\t./checkpoints/iter_0003000\r\n",,terminal_output
+267,1386851,"TERMINAL",0,0,"40K\t./checkpoints/iter_0001500/lr_scheduler\r\n",,terminal_output
+268,1387056,"TERMINAL",0,0,"17K\t./checkpoints/iter_0001500/optimizer\r\n",,terminal_output
+269,1387059,"TERMINAL",0,0,"15M\t./checkpoints/iter_0001500/model\r\n26M\t./checkpoints/iter_0001500\r\n",,terminal_output
+270,1387119,"TERMINAL",0,0,"40K\t./checkpoints/iter_0000500/lr_scheduler\r\n",,terminal_output
+271,1387212,"TERMINAL",0,0,"17K\t./checkpoints/iter_0000500/optimizer\r\n",,terminal_output
+272,1387275,"TERMINAL",0,0,"15M\t./checkpoints/iter_0000500/model\r\n26M\t./checkpoints/iter_0000500\r\n",,terminal_output
+273,1387343,"TERMINAL",0,0,"40K\t./checkpoints/iter_0004000/lr_scheduler\r\n",,terminal_output
+274,1387418,"TERMINAL",0,0,"17K\t./checkpoints/iter_0004000/optimizer\r\n",,terminal_output
+275,1387565,"TERMINAL",0,0,"15M\t./checkpoints/iter_0004000/model\r\n26M\t./checkpoints/iter_0004000\r\n5.5K\t./checkpoints/rollout\r\n40K\t./checkpoints/iter_0003500/lr_scheduler\r\n17K\t./checkpoints/iter_0003500/optimizer\r\n15M\t./checkpoints/iter_0003500/model\r\n26M\t./checkpoints/iter_0003500\r\n",,terminal_output
+276,1387623,"TERMINAL",0,0,"40K\t./checkpoints/iter_0005000/lr_scheduler\r\n",,terminal_output
+277,1387686,"TERMINAL",0,0,"17K\t./checkpoints/iter_0005000/optimizer\r\n",,terminal_output
+278,1388042,"TERMINAL",0,0,"15M\t./checkpoints/iter_0005000/model\r\n26M\t./checkpoints/iter_0005000\r\n40K\t./checkpoints/iter_0002500/lr_scheduler\r\n17K\t./checkpoints/iter_0002500/optimizer\r\n15M\t./checkpoints/iter_0002500/model\r\n26M\t./checkpoints/iter_0002500\r\n40K\t./checkpoints/iter_0001000/lr_scheduler\r\n17K\t./checkpoints/iter_0001000/optimizer\r\n15M\t./checkpoints/iter_0001000/model\r\n26M\t./checkpoints/iter_0001000\r\n40K\t./checkpoints/iter_0002000/lr_scheduler\r\n17K\t./checkpoints/iter_0002000/optimizer\r\n15M\t./checkpoints/iter_0002000/model\r\n26M\t./checkpoints/iter_0002000\r\n259M\t./checkpoints\r\n259M\t.\r\n]0;mahajan1@jwlogin23:/p/scratch/envcomp/mihir/huggingface/shared_data/13032801",,terminal_output
+279,1408798,"TERMINAL",0,0,"cd ../13032804",,terminal_command
+280,1411088,"TERMINAL",0,0,"du -h .",,terminal_command
+281,1411139,"TERMINAL",0,0,"]633;C",,terminal_output
+282,1411322,"TERMINAL",0,0,"40K\t./checkpoints/iter_0003000/lr_scheduler\r\n",,terminal_output
+283,1411415,"TERMINAL",0,0,"13G\t./checkpoints/iter_0003000/optimizer\r\n",,terminal_output
+284,1411482,"TERMINAL",0,0,"7.6G\t./checkpoints/iter_0003000/model\r\n21G\t./checkpoints/iter_0003000\r\n",,terminal_output
+285,1411583,"TERMINAL",0,0,"40K\t./checkpoints/iter_0004000/lr_scheduler\r\n13G\t./checkpoints/iter_0004000/optimizer\r\n",,terminal_output
+286,1411644,"TERMINAL",0,0,"7.6G\t./checkpoints/iter_0004000/model\r\n21G\t./checkpoints/iter_0004000\r\n",,terminal_output
+287,1411707,"TERMINAL",0,0,"3.0K\t./checkpoints/rollout\r\n",,terminal_output
+288,1411830,"TERMINAL",0,0,"40K\t./checkpoints/iter_0005000/lr_scheduler\r\n13G\t./checkpoints/iter_0005000/optimizer\r\n7.6G\t./checkpoints/iter_0005000/model\r\n21G\t./checkpoints/iter_0005000\r\n",,terminal_output
+289,1411994,"TERMINAL",0,0,"40K\t./checkpoints/iter_0001000/lr_scheduler\r\n13G\t./checkpoints/iter_0001000/optimizer\r\n",,terminal_output
+290,1412114,"TERMINAL",0,0,"7.6G\t./checkpoints/iter_0001000/model\r\n21G\t./checkpoints/iter_0001000\r\n40K\t./checkpoints/iter_0002000/lr_scheduler\r\n13G\t./checkpoints/iter_0002000/optimizer\r\n7.6G\t./checkpoints/iter_0002000/model\r\n21G\t./checkpoints/iter_0002000\r\n103G\t./checkpoints\r\n103G\t.\r\n]0;mahajan1@jwlogin23:/p/scratch/envcomp/mihir/huggingface/shared_data/13032804",,terminal_output
+291,1704506,"TERMINAL",0,0,"cd ${SCRATCH}",,terminal_command
+292,1704512,"TERMINAL",0,0,"]633;C]0;mahajan1@jwlogin23:/p/scratch/envcomp",,terminal_output
+293,1707885,"TERMINAL",0,0,"cd /p/home/jusers/mahajan1/juwels/projects/mahajan1/miles",,terminal_command
+294,1707894,"TERMINAL",0,0,"]633;C]0;mahajan1@jwlogin23:~/projects/mahajan1/miles",,terminal_output
+295,1711713,"TERMINAL",0,0,"source .venv/bin/activate",,terminal_command
+296,1711752,"TERMINAL",0,0,"]633;C]0;mahajan1@jwlogin23:~/projects/mahajan1/miles",,terminal_output
+297,1712931,"TERMINAL",0,0,"ls --color=auto",,terminal_command
+298,1712934,"TERMINAL",0,0,"]633;Cbuild_conda.sh               core.jwb0234.juwels.4036136  core.jwb0246.juwels.1957296  core.jwb0248.juwels.1760203  docs              slurm\r\ncore.jwb0234.juwels.4019759  core.jwb0234.juwels.4036137  core.jwb0246.juwels.1957297  core.jwb0248.juwels.1760204  examples          tests\r\ncore.jwb0234.juwels.4019760  core.jwb0234.juwels.4036138  core.jwb0246.juwels.1957298  core.jwb0248.juwels.1760205  imgs              tmp.txt\r\ncore.jwb0234.juwels.4019761  core.jwb0234.juwels.4036139  core.jwb0246.juwels.1957299  core.jwb0248.juwels.1760206  LICENSE           tools\r\ncore.jwb0234.juwels.4019762  core.jwb0246.juwels.1953481  core.jwb0246.juwels.1969807  core.jwb0248.juwels.1762730  miles             train_async.py\r\ncore.jwb0234.juwels.4021039  core.jwb0246.juwels.1953482  core.jwb0246.juwels.1969808  core.jwb0248.juwels.1762732  miles.egg-info    train.py\r\ncore.jwb0234.juwels.4021040  core.jwb0246.juwels.1953483  core.jwb0246.juwels.1969809  core.jwb0248.juwels.1762733  miles_plugins     train_sft.py\r\ncore.jwb0234.juwels.4021041  core.jwb0246.juwels.1953484  core.jwb0246.juwels.1969810  core.jwb0248.juwels.1775280  pyproject.toml    wandb\r\ncore.jwb0234.juwels.4021042  core.jwb0246.juwels.1954767  core.jwb0248.juwels.1758912  core.jwb0248.juwels.1775281  README.md\r\ncore.jwb0234.juwels.4023546  core.jwb0246.juwels.1954768  core.jwb0248.juwels.1758913  core.jwb0248.juwels.1775282  requirements.txt\r\ncore.jwb0234.juwels.4023547  core.jwb0246.juwels.1954769  core.jwb0248.juwels.1758914  core.jwb0248.juwels.1775283  scripts\r\ncore.jwb0234.juwels.4023548  core.jwb0246.juwels.1954770  core.jwb0248.juwels.1758915  docker                       setup.py\r\n]0;mahajan1@jwlogin23:~/projects/mahajan1/miles",,terminal_output
+299,1717970,"TERMINAL",0,0,"cd /p/scratch/envcomp/logs",,terminal_command
+300,1717991,"TERMINAL",0,0,"]633;C]0;mahajan1@jwlogin23:/p/scratch/envcomp/logs",,terminal_output
+301,1718311,"TERMINAL",0,0,"ls --color=auto",,terminal_command
+302,1718317,"TERMINAL",0,0,"]633;CANTA_CLAUS\r\n]0;mahajan1@jwlogin23:/p/scratch/envcomp/logs",,terminal_output
+303,1719731,"TERMINAL",0,0,"cd ANTA_CLAUS/wandb",,terminal_command
+304,1726328,"TERMINAL",0,0,"wandb sync offline-run-202512*",,terminal_command
+305,1726368,"TERMINAL",0,0,"]633;C",,terminal_output
+306,1733834,"TERMINAL",0,0,"Find logs at: /tmp/debug-cli.mahajan1.log\r\n",,terminal_output
+307,1735300,"TERMINAL",0,0,"Syncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/srhvpy1x ... ",,terminal_output
+308,1775668,"TERMINAL",0,0,"done.\r\n",,terminal_output
+309,1776254,"TERMINAL",0,0,"Syncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/s3vt6j03 ... ",,terminal_output
+310,1854902,"TERMINAL",0,0,"done.\r\n",,terminal_output
+311,1855347,"TERMINAL",0,0,"Syncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/hdeu863c ... ",,terminal_output
+312,2607375,"TERMINAL",0,0,"done.\r\nSyncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/txpn35pr ... ",,terminal_output
+313,2607523,"TERMINAL",0,0,"done.\r\nSyncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/2tmrcqyr ... done.\r\nSyncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/cynfzb05 ... done.\r\nSyncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/gbiw7ser ... done.\r\nSyncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/vwqvlawr ... done.\r\nSyncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/qy3yond7 ... done.\r\nSyncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/naygks5j ... done.\r\nSyncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/oiw5w79y ... ",,terminal_output
+314,2607756,"TERMINAL",0,0,"done.\r\nSyncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/srqg6je9 ... done.\r\nSyncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/0nh03pm4 ... done.\r\nSyncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/7zbgvp5x ... done.\r\nSyncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/lq3zgm03 ... done.\r\nSyncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/lu9cvgax ... done.\r\nSyncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/p2t8lqnz ... done.\r\nSyncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/r04esu69 ... done.\r\nSyncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/ae6935ok ... done.\r\nSyncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/8pfgcefs ... done.\r\nSyncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/jl7irwel ... done.\r\nSyncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/bmylq4k5 ... done.\r\nSyncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/ik1q9zz5 ... done.\r\nSyncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/7qsqbv0l ... done.\r\nSyncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/4riovnzx ... done.\r\nSyncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/r5gobf6z ... done.\r\nSyncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/h4yo7jku ... done.\r\nSyncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/p1w5hpc9 ... done.\r\nSyncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/8eru4ijj ... done.\r\nSyncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/05geexep ... done.\r\nSyncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/97n8prwm ... done.\r\nSyncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/995dj0ls ... done.\r\nSyncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/r8kd7td9 ... done.\r\nSyncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/b3ubrvgn ... done.\r\nSyncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/h7d0nu82 ... \r\n\r\n\r\n",,terminal_output
+315,2615353,"TERMINAL",0,0,"done.\r\n",,terminal_output
+316,2616042,"TERMINAL",0,0,"Syncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/dxnhrjgl ... ",,terminal_output
+317,2688171,"TERMINAL",0,0,"done.\r\n",,terminal_output
+318,2688814,"TERMINAL",0,0,"Syncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/8uqk41di ... ",,terminal_output
+319,2713210,"TERMINAL",0,0,"done.\r\n",,terminal_output
+320,2713832,"TERMINAL",0,0,"Syncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/qycikhf6 ... ",,terminal_output
+321,2738549,"TERMINAL",0,0,"done.\r\n",,terminal_output
+322,2739164,"TERMINAL",0,0,"Syncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/my1c4y3r ... ",,terminal_output
+323,2780388,"TERMINAL",0,0,"done.\r\n",,terminal_output
+324,2780999,"TERMINAL",0,0,"Syncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/xfl44nex ... ",,terminal_output
+325,2852143,"TERMINAL",0,0,"done.\r\n",,terminal_output
+326,2852868,"TERMINAL",0,0,"Syncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/1izuyptv ... ",,terminal_output
+327,2876023,"TERMINAL",0,0,"done.\r\n",,terminal_output
+328,2876796,"TERMINAL",0,0,"Syncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/z4ouqxad ... ",,terminal_output
+329,2916322,"TERMINAL",0,0,"done.\r\n",,terminal_output
+330,2917040,"TERMINAL",0,0,"Syncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/rcrggu5s ... ",,terminal_output
+331,2987540,"TERMINAL",0,0,"done.\r\n",,terminal_output
+332,2988288,"TERMINAL",0,0,"Syncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/dgnm2rjh ... ",,terminal_output
+333,3062857,"TERMINAL",0,0,"done.\r\n",,terminal_output
+334,3063479,"TERMINAL",0,0,"Syncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/sovqvfna ... ",,terminal_output
+335,3088467,"TERMINAL",0,0,"done.\r\n",,terminal_output
+336,3089066,"TERMINAL",0,0,"Syncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/a4qul0re ... ",,terminal_output
+337,3127410,"TERMINAL",0,0,"done.\r\n",,terminal_output
+338,3128025,"TERMINAL",0,0,"Syncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/cxqahaj0 ... ",,terminal_output
+339,3151842,"TERMINAL",0,0,"done.\r\n",,terminal_output
+340,3152439,"TERMINAL",0,0,"Syncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/ernte0x4 ... ",,terminal_output
+341,3175417,"TERMINAL",0,0,"done.\r\n",,terminal_output
+342,3175951,"TERMINAL",0,0,"Syncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/5t3dv9s2 ... ",,terminal_output
+343,3214658,"TERMINAL",0,0,"done.\r\n",,terminal_output
+344,3215432,"TERMINAL",0,0,"Syncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/iismd3s9 ... ",,terminal_output
+345,3253920,"TERMINAL",0,0,"done.\r\n",,terminal_output
+346,3254578,"TERMINAL",0,0,"Syncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/22c5kj9q ... ",,terminal_output
+347,3294181,"TERMINAL",0,0,"done.\r\n",,terminal_output
+348,3294793,"TERMINAL",0,0,"Syncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/54umjcrz ... ",,terminal_output
+349,3365370,"TERMINAL",0,0,"done.\r\n",,terminal_output
+350,3366016,"TERMINAL",0,0,"Syncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/axchcl5p ... ",,terminal_output
+351,3439595,"TERMINAL",0,0,"done.\r\n",,terminal_output
+352,3439949,"TERMINAL",0,0,"Syncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/zeltecf3 ... ",,terminal_output
+353,3513487,"TERMINAL",0,0,"done.\r\n",,terminal_output
+354,3514079,"TERMINAL",0,0,"Syncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/xtx69n1v ... ",,terminal_output
+355,3552448,"TERMINAL",0,0,"done.\r\n",,terminal_output
+356,3553097,"TERMINAL",0,0,"Syncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/fb070njd ... ",,terminal_output
+357,3576478,"TERMINAL",0,0,"done.\r\n",,terminal_output
+358,3577199,"TERMINAL",0,0,"Syncing: https://wandb.ai/instant-uv/crowd-pilot-miles/runs/w5ikcawf ... ",,terminal_output
+359,3595221,"TERMINAL",0,0,"done.\r\n",,terminal_output
+360,3595805,"TERMINAL",0,0,"]0;mahajan1@jwlogin23:/p/scratch/envcomp/logs/ANTA_CLAUS/wandb",,terminal_output

927a8af5474e5654810c00ce2e09fd2de87d3e5722f33fa1090d867db114e403/crowd-code-242fa472-b5db-492d-8b66-f482468772b21757500459062-2025_09_10-12.34.42.52/source.csv

@@ -0,0 +1,4 @@
+Sequence,Time,File,RangeOffset,RangeLength,Text,Language,Type
+1,7,"train_dynamics.py",0,0,"from dataclasses import dataclass, field\nimport os\nimport time\n\nimport einops\nfrom flax.training import orbax_utils\nfrom flax.training.train_state import TrainState\nimport optax\nimport orbax\nimport numpy as np\nimport jax\nimport jax.numpy as jnp\nimport tyro\nimport wandb\n\nfrom genie import Genie, restore_genie_components\nfrom utils.dataloader import get_dataloader\n\nts = int(time.time())\n\n\n@dataclass\nclass Args:\n    # Experiment\n    num_steps: int = 200_000\n    seed: int = 0\n    seq_len: int = 16\n    image_channels: int = 3\n    image_resolution: int = 64\n    data_dir: str = ""data/coinrun_episodes""\n    # Optimization\n    batch_size: int = 36\n    min_lr: float = 3e-6\n    max_lr: float = 3e-5\n    warmup_steps: int = 5000\n    # Tokenizer\n    tokenizer_dim: int = 512\n    latent_patch_dim: int = 32\n    num_patch_latents: int = 1024\n    patch_size: int = 4\n    tokenizer_num_blocks: int = 8\n    tokenizer_num_heads: int = 8\n    tokenizer_checkpoint: str = """"\n    # LAM\n    lam_dim: int = 512\n    latent_action_dim: int = 32\n    num_latent_actions: int = 6\n    lam_patch_size: int = 16\n    lam_num_blocks: int = 8\n    lam_num_heads: int = 8\n    lam_checkpoint: str = """"\n    # Dynamics\n    dyna_dim: int = 512\n    dyna_num_blocks: int = 12\n    dyna_num_heads: int = 8\n    dropout: float = 0.0\n    mask_limit: float = 0.5\n    # Logging\n    log: bool = False\n    entity: str = """"\n    project: str = """"\n    log_interval: int = 5\n    log_image_interval: int = 250\n    ckpt_dir: str = """"\n    log_checkpoint_interval: int = 25000\n    log_gradients: bool = False\n    name: str = """"\n    tags: list[str] = field(default_factory=lambda: [""dynamics""])\n    wandb_id: str = """"\n\nargs = tyro.cli(Args)\n\n\ndef dynamics_loss_fn(params, state, inputs):\n    """"""Compute masked dynamics loss""""""\n    outputs = state.apply_fn(\n        params, inputs, training=True, rngs={""dropout"": inputs[""dropout_rng""]}\n    )\n    mask = outputs[""mask""]\n    ce_loss = optax.softmax_cross_entropy_with_integer_labels(\n        outputs[""token_logits""], outputs[""video_tokens""]\n    )\n    ce_loss = (mask * ce_loss).sum() / mask.sum()\n    acc = outputs[""token_logits""].argmax(-1) == outputs[""video_tokens""]\n    acc = (mask * acc).sum() / mask.sum()\n    select_probs = jax.nn.softmax(outputs[""token_logits""])\n    metrics = dict(\n        cross_entropy_loss=ce_loss,\n        masked_token_accuracy=acc,\n        select_logit=outputs[""token_logits""].max(-1).mean(),\n        select_p=select_probs.max(-1).mean(),\n        entropy=jax.scipy.special.entr(select_probs).sum(-1).mean(),\n    )\n    return ce_loss, (outputs[""recon""], metrics)\n\n\n@jax.jit\ndef train_step(state, inputs):\n    """"""Update state and compute metrics""""""\n    grad_fn = jax.value_and_grad(dynamics_loss_fn, has_aux=True, allow_int=True)\n    (loss, (recon, metrics)), grads = grad_fn(state.params, state, inputs)\n    state = state.apply_gradients(grads=grads)\n    if args.log_gradients:\n        metrics[""gradients_std/""] = jax.tree.map(\n            lambda x: x.std(), grads[""params""][""dynamics""]\n        )\n    return state, loss, recon, metrics\n\n\nif __name__ == ""__main__"":\n    rng = jax.random.PRNGKey(args.seed)\n    if args.log:\n        wandb_init_kwargs = {\n            ""entity"": args.entity,\n            ""project"": args.project,\n            ""name"": args.name,\n            ""tags"": args.tags,\n            ""group"": ""debug"",\n            ""config"": args,\n        }\n        if args.wandb_id:\n            wandb_init_kwargs.update(\n                {\n                    ""id"": args.wandb_id,\n                    ""resume"": ""allow"",\n                }\n            )\n        wandb.init(**wandb_init_kwargs)\n\n    # --- Initialize model ---\n    genie = Genie(\n        # Tokenizer\n        in_dim=args.image_channels,\n        tokenizer_dim=args.tokenizer_dim,\n        latent_patch_dim=args.latent_patch_dim,\n        num_patch_latents=args.num_patch_latents,\n        patch_size=args.patch_size,\n        tokenizer_num_blocks=args.tokenizer_num_blocks,\n        tokenizer_num_heads=args.tokenizer_num_heads,\n        # LAM\n        lam_dim=args.lam_dim,\n        latent_action_dim=args.latent_action_dim,\n        num_latent_actions=args.num_latent_actions,\n        lam_patch_size=args.lam_patch_size,\n        lam_num_blocks=args.lam_num_blocks,\n        lam_num_heads=args.lam_num_heads,\n        # Dynamics\n        dyna_dim=args.dyna_dim,\n        dyna_num_blocks=args.dyna_num_blocks,\n        dyna_num_heads=args.dyna_num_heads,\n        dropout=args.dropout,\n        mask_limit=args.mask_limit,\n    )\n    rng, _rng = jax.random.split(rng)\n    image_shape = (args.image_resolution, args.image_resolution, args.image_channels)\n    dummy_inputs = dict(\n        videos=jnp.zeros(\n            (args.batch_size, args.seq_len, *image_shape), dtype=jnp.float32\n        ),\n        mask_rng=_rng,\n    )\n    rng, _rng = jax.random.split(rng)\n    init_params = genie.init(_rng, dummy_inputs)\n    init_params = restore_genie_components(\n        init_params, args.tokenizer_checkpoint, args.lam_checkpoint\n    )\n\n    # --- Initialize optimizer ---\n    lr_schedule = optax.warmup_cosine_decay_schedule(\n        args.min_lr, args.max_lr, args.warmup_steps, args.num_steps\n    )\n    tx = optax.adamw(learning_rate=lr_schedule, b1=0.9, b2=0.9, weight_decay=1e-4)\n    train_state = TrainState.create(apply_fn=genie.apply, params=init_params, tx=tx)\n\n    # --- TRAIN LOOP ---\n    dataloader = get_dataloader(args.data_dir, args.seq_len, args.batch_size)\n    step = 0\n    while step < args.num_steps:\n        for videos in dataloader:\n            # --- Train step ---\n            rng, _rng, _mask_rng = jax.random.split(rng, 3)\n            inputs = dict(\n                videos=videos,\n                action=jnp.zeros((args.batch_size, args.seq_len), dtype=jnp.float32),\n                dropout_rng=_rng,\n                mask_rng=_mask_rng,\n            )\n            train_state, loss, recon, metrics = train_step(train_state, inputs)\n            print(f""Step {step}, loss: {loss}"")\n            step += 1\n\n            # --- Logging ---\n            if args.log:\n                if step % args.log_interval == 0:\n                    wandb.log({""loss"": loss, ""step"": step, **metrics})\n                if step % args.log_image_interval == 0:\n                    gt_seq = inputs[""videos""][0]\n                    recon_seq = recon[0].clip(0, 1)\n                    comparison_seq = jnp.concatenate((gt_seq, recon_seq), axis=1)\n                    comparison_seq = einops.rearrange(\n                        comparison_seq * 255, ""t h w c -> h (t w) c""\n                    )\n                    log_images = dict(\n                        image=wandb.Image(np.asarray(gt_seq[15])),\n                        recon=wandb.Image(np.asarray(recon_seq[15])),\n                        true_vs_recon=wandb.Image(\n                            np.asarray(comparison_seq.astype(np.uint8))\n                        ),\n                    )\n                    wandb.log(log_images)\n                if step % args.log_checkpoint_interval == 0:\n                    ckpt = {""model"": train_state}\n                    orbax_checkpointer = orbax.checkpoint.PyTreeCheckpointer()\n                    save_args = orbax_utils.save_args_from_target(ckpt)\n                    orbax_checkpointer.save(\n                        os.path.join(os.getcwd(), args.ckpt_dir, f""genie_{ts}_{step}""),\n                        ckpt,\n                        save_args=save_args,\n                    )\n            if step >= args.num_steps:\n                break\n",python,tab
+2,479,"extension-output-pdoom-org.crowd-code-#1-crowd-code",0,0,"12:34:42 PM [info] Activating crowd-code\n12:34:42 PM [info] Recording started\n12:34:42 PM [info] Initializing git provider using file system watchers...\n",Log,tab
+3,1114,"extension-output-pdoom-org.crowd-code-#1-crowd-code",153,0,"12:34:42 PM [info] Git repository found\n12:34:42 PM [info] Git provider initialized successfully\n12:34:42 PM [info] Initial git state: [object Object]\n",Log,content

927a8af5474e5654810c00ce2e09fd2de87d3e5722f33fa1090d867db114e403/crowd-code-4b6051ce-1cfc-4dca-874b-0d0d7270d33f1753454394749-2025_07_25-16.42.21.379/source.csv

@@ -0,0 +1,125 @@
+Sequence,Time,File,RangeOffset,RangeLength,Text,Language,Type
+2,1366,"extension-output-pdoom-org.crowd-code-#1-crowd-code",0,0,"4:42:21 PM [info] Activating crowd-code\n4:42:21 PM [info] Recording started\n4:42:21 PM [info] Initializing git provider using file system watchers...\n4:42:21 PM [info] Git repository found\n4:42:21 PM [info] Git provider initialized successfully\n4:42:21 PM [info] Initial git state: [object Object]\n",Log,tab
+3,2902,"TERMINAL",0,0,"bash",,terminal_focus
+4,33144,"slurm/jobs/mihir/horeka/mask_prob_fix/train_dynamics_8_nodes.sbatch",0,0,"#!/usr/bin/env bash\n\n#SBATCH --nodes=8\n#SBATCH --ntasks-per-node=4\n#SBATCH --time=48:00:00\n#SBATCH --partition=accelerated\n#SBATCH --cpus-per-task=5\n#SBATCH --gres=gpu:4\n#SBATCH --output=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/logs/logs_mihir/maskgit-maskprob-fix/dynamics-cotraining/%x_%j.log\n#SBATCH --error=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/logs/logs_mihir/maskgit-maskprob-fix/dynamics-cotraining/%x_%j.log\n#SBATCH --job-name=train_dynamics_maskprob_fix_8_node\n#SBATCH --requeue\n#SBATCH --signal=b:usr1@300 # 5 min before timeout\n\n# --- signal trap to requeue job before timeout ---\nrequeue_job() {\n    echo ""[$(date)] caught sigusr1 (timeout warning), requeueing slurm job $SLURM_JOB_ID...""\n    # optional: trigger checkpoint saving here\n    # e.g., touch $checkpoint_dir/requeue_trigger\n    scontrol requeue $SLURM_JOB_ID\n    exit 0\n}\n\ntrap requeue_job sigusr1\n\n# set checkpoint flag based on restart count\nrestart_count=$(scontrol show job $SLURM_JOB_ID | grep -o 'Restarts=[0-9]*' | cut -d'=' -f2)\n\nif [ $restart_count -eq 0 ]; then\n    restore_ckpt_flag=""--no-restore-ckpt""\nelse\n    restore_ckpt_flag=""--restore-ckpt""\nfi\n\n\n\n# Log the sbatch script\ncat $0\n\nmodule unload mpi/openmpi/5.0\nmodule unload devel/cuda/12.4\nsource .venv/bin/activate\n\narray_records_dir=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/data_new/open_ai_minecraft_arrayrecords_chunked\n\njob_name=$SLURM_JOB_NAME\nslurm_job_id=$SLURM_JOB_ID\n\nCHECKPOINT_DIR=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/checkpoints/maskgit-maskprob-fix/$job_name/$slurm_job_id\nmkdir -p $CHECKPOINT_DIR\n\n# tokenizer with the new structure supporting larger ffn_dim\ntokenizer_ckpt_dir=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/data/checkpoints/train_tokenizer_lr_sweep_1e-4_larger_ffn/\n\nenv | grep SLURM\n\nsrun python train_dynamics.py \\n    --save_ckpt \\n    $restore_ckpt_flag \\n    --wandb_id $SLURM_JOB_ID \\n    --ckpt_dir $CHECKPOINT_DIR \\n    --batch_size=384 \\n    --init_lr=0 \\n    --max_lr=8e-5 \\n    --log_image_interval=1000 \\n    --log \\n    --log_checkpoint_interval=1000 \\n    --name=dynamics-maskprob-fix-8-node-$slurm_job_id \\n    --tags dynamics maskprob-fix 8-node \\n    --entity instant-uv \\n    --project jafar \\n    --dyna_dim=1024 \\n    --dyna_num_blocks=16 \\n    --dyna_num_heads=16 \\n    --dyna_ffn_dim=4096 \\n    --tokenizer_checkpoint=$tokenizer_ckpt_dir \\n    --data_dir $array_records_dir &\n\nchild_pid=$!\n\nwait $child_pid\n",shellscript,tab
+5,51851,"TERMINAL",0,0,"bash",,terminal_focus
+6,56221,"TERMINAL",0,0,"fsacct_week",,terminal_command
+7,56266,"TERMINAL",0,0,"]633;E;2025-07-25 16:43:16 fsacct_week;406cfb31-2341-454a-afa8-cae7781806b2]633;C",,terminal_output
+8,56266,"TERMINAL",0,0,"          JobID                        JobName        Partition All        State    Elapsed  Timelimit \r\n--------------- ------------------------------ ---------------- --- ------------ ---------- ---------- \r\n        3358457     train_dyn_yolorun_new_arch      accelerated  48       FAILED   00:00:28 2-00:00:00 \r\n        3359334                           wrap      accelerated   6      TIMEOUT   10:00:29   10:00:00 \r\n        3359338                           wrap      accelerated   6      TIMEOUT   10:00:16   10:00:00 \r\n        3359343 train_dyn_new_arch-bugfixed-s+      accelerated  48    COMPLETED   23:19:14 2-00:00:00 \r\n        3359349 train_dyn_new_arch-bugfixed-t+      accelerated  48    COMPLETED 1-01:00:55 2-00:00:00 \r\n        3365873 train_dynamics_overfit_sample+      accelerated   6    COMPLETED   01:26:52 2-00:00:00 \r\n        3365876 train_dynamics_overfit_sample+      accelerated   6    COMPLETED   01:40:07 2-00:00:00 \r\n        3366883 train_dynamics_overfit_sample+      accelerated   6    COMPLETED   01:33:31 2-00:00:00 \r\n        3371238 train_dynamics_maskprob_fix_2+      accelerated  48      RUNNING 1-16:09:01 2-00:00:00 \r\n        3372629 train_dynamics_maskprob_fix_8+      accelerated 192    COMPLETED 1-02:29:22 2-00:00:00 \r\n        3372631 train_dynamics_maskprob_fix_2+      accelerated  48    COMPLETED 1-01:17:59 2-00:00:00 \r\n        3372931          train_dyn_causal_180M  dev_accelerated   6       FAILED   00:00:33   00:10:00 \r\n        3372932          train_dyn_causal_255M  dev_accelerated   6       FAILED   00:00:29   00:10:00 \r\n        3372934          train_dyn_causal_356M  dev_accelerated   6       FAILED   00:00:29   00:10:00 \r\n        3372936          train_dyn_causal_500M  dev_accelerated   6       FAILED   00:00:29   00:10:00 \r\n        3372969          train_dyn_causal_180M  dev_accelerated   6       FAILED   00:02:11   00:10:00 \r\n        3372970          train_dyn_causal_255M  dev_accelerated   6       FAILED   00:02:24   00:10:00 \r\n        3372971          train_dyn_causal_356M  dev_accelerated   6       FAILED   00:02:08   00:10:00 \r\n        3372972          train_dyn_causal_500M  dev_accelerated   6       FAILED   00:02:09   00:10:00 \r\n        3373107          train_dyn_causal_180M  dev_accelerated   6    COMPLETED   00:06:15   00:10:00 \r\n        3373108          train_dyn_causal_255M  dev_accelerated   6    COMPLETED   00:07:14   00:10:00 \r\n        3373109          train_dyn_causal_356M  dev_accelerated   6       FAILED   00:04:17   00:10:00 \r\n        3373110          train_dyn_causal_500M  dev_accelerated   6       FAILED   00:04:59   00:10:00 \r\n        3373400                           wrap      accelerated   6    COMPLETED   00:04:34   02:00:00 \r\n        3373404                           wrap      accelerated   6    COMPLETED   00:04:38   02:00:00 \r\n        3373407   train_dynamics_causal_2_node      accelerated  48      RUNNING   05:58:48 2-00:00:00 \r\n        3373408   train_dynamics_causal_8_node      accelerated 192      RUNNING   05:58:48 2-00:00:00 \r\n        3373409                           wrap      accelerated   6    COMPLETED   00:41:24   02:00:00 \r\n        3373410                           wrap      accelerated   6    COMPLETED   00:42:56   02:00:00 \r\n        3371237 train_dynamics_maskprob_fix_8+      accelerated 192      RUNNING   00:45:54 2-00:00:00 \r\n]0;tum_cte0515@hkn1990:~/Projects/jafar]633;D;0",,terminal_output
+9,489274,"slurm/jobs/mihir/horeka/mask_prob_fix/train_dynamics_8_nodes.sbatch",0,0,"",shellscript,tab
+10,492538,"TERMINAL",0,0,"bash",,terminal_focus
+11,496388,"TERMINAL",0,0,"cd $ws_dir",,terminal_command
+12,496810,"TERMINAL",0,0,"ls",,terminal_command
+13,496859,"TERMINAL",0,0,"]633;E;2025-07-25 16:50:38 ls;adbf53fe-397b-40d3-9339-94ea79afad56]633;C",,terminal_output
+14,496990,"TERMINAL",0,0,"checkpoints  count_items.sh  data  data_new  huggingface  logs  possibly_corrupt_files_in_this_workspace.txt  scripts\r\n]0;tum_cte0515@hkn1990:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared]633;D;0",,terminal_output
+15,501267,"TERMINAL",0,0,"cd checkpoints/",,terminal_command
+16,501293,"TERMINAL",0,0,"]633;E;2025-07-25 16:50:42 cd checkpoints/;adbf53fe-397b-40d3-9339-94ea79afad56]633;C]0;tum_cte0515@hkn1990:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/checkpoints]633;D;0",,terminal_output
+17,501578,"TERMINAL",0,0,"ls",,terminal_command
+18,501628,"TERMINAL",0,0,"]633;E;2025-07-25 16:50:42 ls;adbf53fe-397b-40d3-9339-94ea79afad56]633;C",,terminal_output
+19,501871,"TERMINAL",0,0,"0000     3291405  3292335  3296571  3297606  3299069  3301026  3310436  3316022             lam_ckpt_dir                      train_dyn_causal_500M\r\n3290283  3292213  3292336  3296573  3297671  3299258  3301027  3310437  big-runs            lam_main_test                     train_dyn_new_arch-bugfixed-spatial-shift\r\n3290284  3292221  3292337  3296574  3297693  3299259  3301029  3311671  causal              maskgit-maskprob-fix              train_dyn_new_arch-bugfixed-temporal-shift\r\n3290295  3292258  3292338  3296575  3297706  3299272  3301030  3311672  checkpoints_alfred  tokenizer                         train_dyn_yolorun_new_arch\r\n3290296  3292328  3292339  3297569  3297727  3299579  3301031  3313562  coinrun             tokenizer_ckpt_dir                train_lam_minecraft_overfit_sample\r\n3290366  3292329  3294600  3297575  3299016  3300233  3306801  3313563  debug               train_dynamics_lr_schedule_const  train_tokenizer_batch_size_scaling_16_node\r\n3290367  3292330  3294601  3297576  3299062  3300290  3307618  3313564  dyn                 train_dynamics_lr_schedule_cos    train_tokenizer_minecraft_overfit_sample\r\n3290391  3292331  3294602  3297577  3299063  3300658  3307619  3313565  dynamics_ckpt_dir   train_dynamics_lr_schedule_wsd    wrap\r\n3290392  3292332  3294603  3297578  3299065  3300663  3309662  3313570  interactive         train_dyn_causal_180M\r\n3290439  3292333  3296502  3297582  3299066  3300672  3309663  3313571  lam                 train_dyn_causal_255M\r\n3290440  3292334  3296540  3297586  3299068  3301025  3309699  3313572  lam-1-action        train_dyn_causal_356M\r\n]0;tum_cte0515@hkn1990:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/checkpoints]633;D;0",,terminal_output
+20,503515,"slurm/jobs/mihir/horeka/mask_prob_fix/train_dynamics_8_nodes.sbatch",0,0,"",shellscript,tab
+21,506478,"slurm/jobs/mihir/horeka/mask_prob_fix/train_dynamics_8_nodes.sbatch",1559,0,"",shellscript,selection_mouse
+22,506611,"slurm/jobs/mihir/horeka/mask_prob_fix/train_dynamics_8_nodes.sbatch",1557,8,"maskprob",shellscript,selection_mouse
+23,511718,"TERMINAL",0,0,"cd maskgit-maskprob-fix/",,terminal_command
+24,512173,"TERMINAL",0,0,"ls",,terminal_command
+25,512216,"TERMINAL",0,0,"]633;E;2025-07-25 16:50:53 ls;adbf53fe-397b-40d3-9339-94ea79afad56]633;C",,terminal_output
+26,512275,"TERMINAL",0,0,"interactive  train_dynamics_maskprob_fix_2_node  train_dynamics_maskprob_fix_2_node_80M  train_dynamics_maskprob_fix_8_node  train_dynamics_maskprob_fix_8_node_80M\r\n]0;tum_cte0515@hkn1990:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/checkpoints/maskgit-maskprob-fix]633;D;0",,terminal_output
+27,520223,"TERMINAL",0,0,"cd train_dynamics_maskprob_fix_",,terminal_command
+28,520228,"TERMINAL",0,0,"]633;E;2025-07-25 16:51:01 cd train_dynamics_maskprob_fix_;adbf53fe-397b-40d3-9339-94ea79afad56]633;Cbash: cd: train_dynamics_maskprob_fix_: No such file or directory\r\n]0;tum_cte0515@hkn1990:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/checkpoints/maskgit-maskprob-fix]633;D;1",,terminal_output
+29,522776,"TERMINAL",0,0,"cd train_dynamics_maskprob_fix_8_node",,terminal_command
+30,522802,"TERMINAL",0,0,"]633;E;2025-07-25 16:51:04 cd train_dynamics_maskprob_fix_8_node;adbf53fe-397b-40d3-9339-94ea79afad56]633;C]0;tum_cte0515@hkn1990:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/checkpoints/maskgit-maskprob-fix/train_dynamics_maskprob_fix_8_node]633;D;0",,terminal_output
+31,523278,"TERMINAL",0,0,"ls",,terminal_command
+32,523301,"TERMINAL",0,0,"]633;E;2025-07-25 16:51:04 ls;adbf53fe-397b-40d3-9339-94ea79afad56]633;C3370788  3371237\r\n]0;tum_cte0515@hkn1990:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/checkpoints/maskgit-maskprob-fix/train_dynamics_maskprob_fix_8_node]633;D;0",,terminal_output
+33,536976,"TERMINAL",0,0,"cd 3371237/",,terminal_command
+34,537010,"TERMINAL",0,0,"]633;E;2025-07-25 16:51:18 cd 3371237/;adbf53fe-397b-40d3-9339-94ea79afad56]633;C]0;tum_cte0515@hkn1990:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/checkpoints/maskgit-maskprob-fix/train_dynamics_maskprob_fix_8_node/3371237]633;D;0",,terminal_output
+35,537454,"TERMINAL",0,0,"ls",,terminal_command
+36,537509,"TERMINAL",0,0,"]633;E;2025-07-25 16:51:18 ls;adbf53fe-397b-40d3-9339-94ea79afad56]633;C016000  017000  018000\r\n]0;tum_cte0515@hkn1990:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/checkpoints/maskgit-maskprob-fix/train_dynamics_maskprob_fix_8_node/3371237]633;D;0",,terminal_output
+37,557600,"slurm/jobs/mihir/horeka/mask_prob_fix/train_dynamics_8_nodes.sbatch",0,0,"",shellscript,tab
+38,560516,"slurm/jobs/mihir/horeka/mask_prob_fix/train_dynamics_8_nodes.sbatch",2060,0,"",shellscript,selection_mouse
+39,560673,"slurm/jobs/mihir/horeka/mask_prob_fix/train_dynamics_8_nodes.sbatch",2059,4,"1000",shellscript,selection_mouse
+40,561723,"slurm/jobs/mihir/horeka/mask_prob_fix/train_dynamics_8_nodes.sbatch",2147,0,"",shellscript,selection_mouse
+41,562486,"slurm/jobs/mihir/horeka/mask_prob_fix/train_dynamics_8_nodes.sbatch",2144,3,"fix",shellscript,selection_mouse
+42,563079,"slurm/jobs/mihir/horeka/mask_prob_fix/train_dynamics_8_nodes.sbatch",2110,0,"",shellscript,selection_mouse
+43,563243,"slurm/jobs/mihir/horeka/mask_prob_fix/train_dynamics_8_nodes.sbatch",2108,4,"1000",shellscript,selection_mouse
+44,579775,"train_dynamics.py",0,0,"from dataclasses import dataclass, field\nimport os\n\nimport einops\nfrom flax.training.train_state import TrainState\nfrom jax.sharding import Mesh, PartitionSpec, NamedSharding\nfrom jax.experimental.mesh_utils import create_device_mesh\nimport optax\nimport orbax.checkpoint as ocp\nimport numpy as np\nimport dm_pix as pix\nimport jax\nimport jax.numpy as jnp\nimport tyro\nimport wandb\nimport grain\n\nfrom genie import Genie, restore_genie_components\nfrom utils.dataloader import get_dataloader\nfrom utils.lr_utils import get_lr_schedule\nfrom utils.parameter_utils import count_parameters_by_component\n\n\n\n@dataclass\nclass Args:\n    # Experiment\n    num_steps: int = 200_000\n    seed: int = 0\n    seq_len: int = 16\n    image_channels: int = 3\n    image_height: int = 90\n    image_width: int = 160\n    data_dir: str = """"\n    save_ckpt: bool = False\n    restore_ckpt: bool = False\n    # Optimization\n    batch_size: int = 36\n    init_lr: float = 0.0\n    max_lr: float = 3e-5\n    decay_end: float = 0.0\n    wsd_decay_steps: int = (\n        10000  # NOTE: wsd_decay_steps will only be used when using a wsd-schedule\n    )\n    warmup_steps: int = 5000\n    lr_schedule: str = ""wsd""  # supported options: wsd, cos\n    # Tokenizer\n    tokenizer_dim: int = 512\n    tokenizer_ffn_dim: int = 2048\n    latent_patch_dim: int = 32\n    num_patch_latents: int = 1024\n    patch_size: int = 4\n    tokenizer_num_blocks: int = 4\n    tokenizer_num_heads: int = 8\n    tokenizer_checkpoint: str = """"\n    # LAM\n    lam_dim: int = 512\n    lam_ffn_dim: int = 2048\n    latent_action_dim: int = 32\n    num_latent_actions: int = 6\n    lam_patch_size: int = 16\n    lam_num_blocks: int = 4\n    lam_num_heads: int = 8\n    lam_checkpoint: str = """"\n    # Dynamics\n    dyna_dim: int = 512\n    dyna_ffn_dim: int = 2048\n    dyna_num_blocks: int = 6\n    dyna_num_heads: int = 8\n    dropout: float = 0.0\n    mask_limit: float = 0.5\n    use_maskgit: bool = False\n    param_dtype: jnp.dtype = jnp.float32\n    dtype: jnp.dtype = jnp.bfloat16\n    use_flash_attention: bool = True\n    # Logging\n    log: bool = False\n    entity: str = """"\n    project: str = """"\n    name: str = ""train_dynamics""\n    tags: list[str] = field(default_factory=lambda: [""dynamics""])\n    log_interval: int = 5\n    log_image_interval: int = 250\n    ckpt_dir: str = """"\n    log_checkpoint_interval: int = 25000\n    log_checkpoint_keep_period: int = 20000\n    log_gradients: bool = False\n    wandb_id: str = """"\n\n\nargs = tyro.cli(Args)\n\n\ndef dynamics_loss_fn(params, state, inputs):\n    """"""Compute masked dynamics loss""""""\n    inputs[""videos""] = inputs[""videos""].astype(args.dtype) / 255.0\n    outputs = state.apply_fn(\n        params,\n        inputs,\n        training=True,\n        rngs={""params"": inputs[""rng""], ""dropout"": inputs[""dropout_rng""]},\n    )\n    mask = outputs[""mask""]\n    outputs[""token_logits""] = outputs[""token_logits""].astype(jnp.float32)\n    outputs[""recon""] = outputs[""recon""].astype(jnp.float32)\n    logits = outputs[""token_logits""]\n    targets = outputs[""video_tokens""]\n\n    # if not args.use_maskgit:\n    # logits = outputs[""token_logits""][:, :, :-1]\n    # targets = outputs[""video_tokens""][:, :, 1:]\n    # mask = outputs[""mask""][:, :, 1:]\n\n    ce_loss = optax.softmax_cross_entropy_with_integer_labels(logits, targets)\n    ce_loss = (mask * ce_loss).sum() / mask.sum()\n    acc = logits.argmax(-1) == targets\n    acc = (mask * acc).sum() / mask.sum()\n    select_probs = jax.nn.softmax(logits)\n    gt = inputs[""videos""].clip(0, 1).reshape(-1, *inputs[""videos""].shape[2:])\n    recon = outputs[""recon""].clip(0, 1).reshape(-1, *outputs[""recon""].shape[2:])\n    psnr = pix.psnr(gt, recon).mean()  # type: ignore\n    ssim = pix.ssim(gt, recon).mean()  # type: ignore\n    psnr = pix.psnr(gt, recon).mean()  # type: ignore\n    ssim = pix.ssim(gt, recon).mean()  # type: ignore\n    _, index_counts_lam = jnp.unique_counts(\n        jnp.ravel(outputs[""lam_indices""]), size=args.num_latent_actions, fill_value=0\n    )\n    _, index_counts_tokenizer = jnp.unique_counts(\n        jnp.ravel(outputs[""video_tokens""]), size=args.num_patch_latents, fill_value=0\n    )\n    codebook_usage_lam = (index_counts_lam != 0).mean()\n    codebook_usage_tokenizer = (index_counts_tokenizer != 0).mean()\n    metrics = dict(\n        cross_entropy_loss=ce_loss,\n        masked_token_accuracy=acc,\n        select_logit=logits.max(-1).mean(),\n        select_p=select_probs.max(-1).mean(),\n        entropy=jax.scipy.special.entr(select_probs).sum(-1).mean(),\n        psnr=psnr,\n        ssim=ssim,\n        codebook_usage_lam=codebook_usage_lam,\n        codebook_usage_tokenizer=codebook_usage_tokenizer,\n    )\n    return ce_loss, (outputs[""recon""], metrics)\n\n\n@jax.jit\ndef train_step(state, inputs):\n    """"""Update state and compute metrics""""""\n    grad_fn = jax.value_and_grad(dynamics_loss_fn, has_aux=True, allow_int=True)\n    (loss, (recon, metrics)), grads = grad_fn(state.params, state, inputs)\n    state = state.apply_gradients(grads=grads)\n    if args.log_gradients:\n        metrics[""gradients_std/""] = jax.tree.map(\n            lambda x: x.std(), grads[""params""][""dynamics""]\n        )\n    return state, loss, recon, metrics\n\n\nif __name__ == ""__main__"":\n    jax.distributed.initialize()\n    num_devices = jax.device_count()\n    if num_devices == 0:\n        raise ValueError(""No JAX devices found."")\n    print(f""Running on {num_devices} devices."")\n\n    if args.batch_size % num_devices != 0:\n        raise ValueError(\n            f""Global batch size {args.batch_size} must be divisible by ""\n            f""number of devices {num_devices}.""\n        )\n\n    per_device_batch_size_for_init = args.batch_size // num_devices\n\n    rng = jax.random.PRNGKey(args.seed)\n\n    # --- Initialize model ---\n    genie = Genie(\n        # Tokenizer\n        in_dim=args.image_channels,\n        tokenizer_dim=args.tokenizer_dim,\n        tokenizer_ffn_dim=args.tokenizer_ffn_dim,\n        latent_patch_dim=args.latent_patch_dim,\n        num_patch_latents=args.num_patch_latents,\n        patch_size=args.patch_size,\n        tokenizer_num_blocks=args.tokenizer_num_blocks,\n        tokenizer_num_heads=args.tokenizer_num_heads,\n        # LAM\n        lam_dim=args.lam_dim,\n        lam_ffn_dim=args.lam_ffn_dim,\n        latent_action_dim=args.latent_action_dim,\n        num_latent_actions=args.num_latent_actions,\n        lam_patch_size=args.lam_patch_size,\n        lam_num_blocks=args.lam_num_blocks,\n        lam_num_heads=args.lam_num_heads,\n        lam_co_train=not args.lam_checkpoint,\n        # Dynamics\n        dyna_dim=args.dyna_dim,\n        dyna_ffn_dim=args.dyna_ffn_dim,\n        dyna_num_blocks=args.dyna_num_blocks,\n        dyna_num_heads=args.dyna_num_heads,\n        dropout=args.dropout,\n        mask_limit=args.mask_limit,\n        use_maskgit=args.use_maskgit,\n        param_dtype=args.param_dtype,\n        dtype=args.dtype,\n        use_flash_attention=args.use_flash_attention,\n    )\n    rng, _rng = jax.random.split(rng)\n    image_shape = (args.image_height, args.image_width, args.image_channels)\n    dummy_inputs = dict(\n        videos=jnp.zeros(\n            (per_device_batch_size_for_init, args.seq_len, *image_shape),\n            dtype=args.dtype,\n        ),\n        action=jnp.zeros(\n            (per_device_batch_size_for_init, args.seq_len), dtype=args.dtype\n        ),\n        mask_rng=_rng,\n    )\n    rng, _rng = jax.random.split(rng)\n    init_params = genie.init(_rng, dummy_inputs)\n\n    param_counts = count_parameters_by_component(init_params)\n\n    if args.log and jax.process_index() == 0:\n        wandb_init_kwargs = {\n            ""entity"": args.entity,\n            ""project"": args.project,\n            ""name"": args.name,\n            ""tags"": args.tags,\n            ""group"": ""debug"",\n            ""config"": args,\n        }\n\n        if args.wandb_id:\n            wandb_init_kwargs.update(\n                {\n                    ""id"": args.wandb_id,\n                    ""resume"": ""allow"",\n                }\n            )\n        wandb.init(**wandb_init_kwargs)\n\n        wandb.config.update({""model_param_count"": param_counts})\n\n    print(""Parameter counts:"")\n    print(param_counts)\n\n    # --- Initialize optimizer ---\n    lr_schedule = get_lr_schedule(\n        args.lr_schedule,\n        args.init_lr,\n        args.max_lr,\n        args.decay_end,\n        args.num_steps,\n        args.warmup_steps,\n        args.wsd_decay_steps,\n    )\n    tx = optax.adamw(\n        learning_rate=lr_schedule,\n        b1=0.9,\n        b2=0.9,\n        weight_decay=1e-4,\n        mu_dtype=args.dtype,\n    )\n    train_state = TrainState.create(apply_fn=genie.apply, params=init_params, tx=tx)\n\n    device_mesh_arr = create_device_mesh((num_devices,))\n    mesh = Mesh(devices=device_mesh_arr, axis_names=(""data"",))\n\n    replicated_sharding = NamedSharding(mesh, PartitionSpec())\n    videos_sharding = NamedSharding(mesh, PartitionSpec(""data"", None, None, None, None))\n    train_state = jax.device_put(train_state, replicated_sharding)\n\n    # --- Initialize checkpoint manager ---\n    step = 0\n    handler_registry = ocp.handlers.DefaultCheckpointHandlerRegistry()\n    handler_registry.add(\n        ""model_state"", ocp.args.StandardSave, ocp.handlers.StandardCheckpointHandler\n    )\n    handler_registry.add(\n        ""model_state"", ocp.args.StandardRestore, ocp.handlers.StandardCheckpointHandler\n    )\n    handler_registry.add(""dataloader_state"", grain.checkpoint.CheckpointSave, grain.checkpoint.CheckpointHandler)  # type: ignore\n    handler_registry.add(""dataloader_state"", grain.checkpoint.CheckpointRestore, grain.checkpoint.CheckpointHandler)  # type: ignore\n\n    checkpoint_options = ocp.CheckpointManagerOptions(\n        save_interval_steps=args.log_checkpoint_interval,\n        max_to_keep=3,\n        keep_period=args.log_checkpoint_keep_period,\n        step_format_fixed_length=6,\n        cleanup_tmp_directories=True,\n    )\n\n    checkpoint_manager = ocp.CheckpointManager(\n        args.ckpt_dir,\n        options=checkpoint_options,\n        handler_registry=handler_registry,\n    )\n\n    # --- Create DataLoaderIterator from dataloader ---\n    array_record_files = [\n        os.path.join(args.data_dir, x)\n        for x in os.listdir(args.data_dir)\n        if x.endswith("".array_record"")\n    ]\n    grain_dataloader = get_dataloader(\n        array_record_files,\n        args.seq_len,\n        # NOTE: We deliberately pass the global batch size\n        # The dataloader shards the dataset across all processes\n        args.batch_size,\n        *image_shape,\n        num_workers=8,\n        prefetch_buffer_size=1,\n        seed=args.seed,\n    )\n    initial_state = grain_dataloader._create_initial_state()\n    grain_iterator = grain.DataLoaderIterator(grain_dataloader, initial_state)\n\n    # --- Restore checkpoint ---\n    if args.restore_ckpt:\n        # Restore full dynamics model\n        abstract_train_state = jax.tree_util.tree_map(\n            ocp.utils.to_shape_dtype_struct, train_state\n        )\n        restored = checkpoint_manager.restore(\n            checkpoint_manager.latest_step(),\n            args=ocp.args.Composite(\n            model_state=ocp.args.StandardRestore(abstract_train_state),\n            dataloader_state=grain.checkpoint.CheckpointRestore(grain_iterator),\n        ),\n        )\n        train_state = restored[""model_state""]\n        grain_iterator = restored[""dataloader_state""]\n        step = checkpoint_manager.latest_step() or 0\n        print(f""Restored dataloader and model state from step {step}"")\n    else:\n        # Restore from pre-trained tokenizer (and LAM)\n        train_state = restore_genie_components(\n            train_state, replicated_sharding, dummy_inputs, rng, args\n        )\n\n    # --- TRAIN LOOP ---\n    dataloader = (jax.make_array_from_process_local_data(videos_sharding, elem) for elem in grain_iterator)  # type: ignore\n    while step < args.num_steps:\n        for videos in dataloader:\n        # for i in range(videos.shape[0]):\n        # video_i = videos[i:i+1]  # shape (1, T, H, W, C)\n        # np.save(f""overfit_dir/oai_sample_seed69_{i}.npy"", video_i)\n        # jax.debug.breakpoint()\n        # videos = np.load(""overfit_dir/oai_sample_seed69_1.npy"")  # *255.\n        # videos = videos.astype(np.uint8)\n        # videos = jax.make_array_from_process_local_data(videos_sharding, videos)\n        # while True:\n            # --- Train step ---\n            rng, _rng, _rng_dropout, _rng_mask = jax.random.split(rng, 4)\n\n            inputs = dict(\n                videos=videos,\n                rng=_rng,\n                dropout_rng=_rng_dropout,\n                mask_rng=_rng_mask,\n            )\n            train_state, loss, recon, metrics = train_step(train_state, inputs)\n            metrics[""lr""] = lr_schedule(step)\n            print(f""Step {step}, loss: {loss}"")\n            step += 1\n\n            # --- Logging ---\n            if args.log:\n                if step % args.log_interval == 0 and jax.process_index() == 0:\n                    wandb.log(\n                        {\n                            ""loss"": loss,\n                            ""step"": step,\n                            **metrics,\n                        }\n                    )\n                if step % args.log_image_interval == 0:\n                    gt_seq = inputs[""videos""][0].astype(jnp.float32) / 255.0\n                    recon_seq = recon[0].clip(0, 1)\n                    comparison_seq = jnp.concatenate((gt_seq, recon_seq), axis=1)\n                    comparison_seq = einops.rearrange(\n                        comparison_seq * 255, ""t h w c -> h (t w) c""\n                    )\n                    if jax.process_index() == 0:\n                        log_images = dict(\n                            image=wandb.Image(np.asarray(gt_seq[args.seq_len - 1])),\n                            recon=wandb.Image(np.asarray(recon_seq[args.seq_len - 1])),\n                            true_vs_recon=wandb.Image(\n                                np.asarray(comparison_seq.astype(np.uint8))\n                            ),\n                        )\n                        wandb.log(log_images)\n            # --- Checkpointing ---\n            if args.save_ckpt and step % args.log_checkpoint_interval == 0:\n                checkpoint_manager.save(\n                    step,\n                    args=ocp.args.Composite(\n                        model_state=ocp.args.StandardSave(train_state),\n                        dataloader_state=grain.checkpoint.CheckpointSave(\n                            grain_iterator\n                        ),\n                    ),\n                )\n                print(f""Saved checkpoint at step {step}"")\n            if step >= args.num_steps:\n                break\n\n    checkpoint_manager.close()\n",python,tab
+45,588014,"train_dynamics.py",2371,0,"",python,selection_mouse
+46,588162,"train_dynamics.py",2367,5,"20000",python,selection_mouse
+47,598870,"TERMINAL",0,0,"cd ..",,terminal_command
+48,599269,"TERMINAL",0,0,"ls",,terminal_command
+49,600516,"TERMINAL",0,0,"cd ..",,terminal_command
+50,600835,"TERMINAL",0,0,"ls",,terminal_command
+51,600876,"TERMINAL",0,0,"]633;E;2025-07-25 16:52:22 ls;adbf53fe-397b-40d3-9339-94ea79afad56]633;Cinteractive  train_dynamics_maskprob_fix_2_node  train_dynamics_maskprob_fix_2_node_80M  train_dynamics_maskprob_fix_8_node  train_dynamics_maskprob_fix_8_node_80M\r\n]0;tum_cte0515@hkn1990:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/checkpoints/maskgit-maskprob-fix]633;D;0",,terminal_output
+52,610052,"TERMINAL",0,0,"cd train_dynamics_maskprob_fix_2_node",,terminal_command
+53,610076,"TERMINAL",0,0,"]633;E;2025-07-25 16:52:31 cd train_dynamics_maskprob_fix_2_node;adbf53fe-397b-40d3-9339-94ea79afad56]633;C]0;tum_cte0515@hkn1990:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/checkpoints/maskgit-maskprob-fix/train_dynamics_maskprob_fix_2_node]633;D;0",,terminal_output
+54,611337,"TERMINAL",0,0,"ls",,terminal_command
+55,611384,"TERMINAL",0,0,"]633;E;2025-07-25 16:52:32 ls;adbf53fe-397b-40d3-9339-94ea79afad56]633;C3371238\r\n]0;tum_cte0515@hkn1990:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/checkpoints/maskgit-maskprob-fix/train_dynamics_maskprob_fix_2_node]633;D;0",,terminal_output
+56,613099,"TERMINAL",0,0,"cd 3371238/",,terminal_command
+57,613125,"TERMINAL",0,0,"]633;E;2025-07-25 16:52:34 cd 3371238/;adbf53fe-397b-40d3-9339-94ea79afad56]633;C]0;tum_cte0515@hkn1990:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/checkpoints/maskgit-maskprob-fix/train_dynamics_maskprob_fix_2_node/3371238]633;D;0",,terminal_output
+58,613491,"TERMINAL",0,0,"ls",,terminal_command
+59,613535,"TERMINAL",0,0,"]633;E;2025-07-25 16:52:34 ls;adbf53fe-397b-40d3-9339-94ea79afad56]633;C020000  031000  032000  033000\r\n]0;tum_cte0515@hkn1990:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/checkpoints/maskgit-maskprob-fix/train_dynamics_maskprob_fix_2_node/3371238]633;D;0",,terminal_output
+60,625849,"TERMINAL",0,0,"cd ..",,terminal_command
+61,625875,"TERMINAL",0,0,"]633;E;2025-07-25 16:52:47 cd ..;adbf53fe-397b-40d3-9339-94ea79afad56]633;C]0;tum_cte0515@hkn1990:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/checkpoints/maskgit-maskprob-fix/train_dynamics_maskprob_fix_2_node]633;D;0",,terminal_output
+62,626753,"TERMINAL",0,0,"cd ..",,terminal_command
+63,626781,"TERMINAL",0,0,"]633;E;2025-07-25 16:52:48 cd ..;adbf53fe-397b-40d3-9339-94ea79afad56]633;C]0;tum_cte0515@hkn1990:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/checkpoints/maskgit-maskprob-fix]633;D;0",,terminal_output
+64,628009,"TERMINAL",0,0,"cd ..",,terminal_command
+65,628426,"TERMINAL",0,0,"ls",,terminal_command
+66,628538,"TERMINAL",0,0,"]633;E;2025-07-25 16:52:49 ls;adbf53fe-397b-40d3-9339-94ea79afad56]633;C0000     3291405  3292335  3296571  3297606  3299069  3301026  3310436  3316022             lam_ckpt_dir                      train_dyn_causal_500M\r\n3290283  3292213  3292336  3296573  3297671  3299258  3301027  3310437  big-runs            lam_main_test                     train_dyn_new_arch-bugfixed-spatial-shift\r\n3290284  3292221  3292337  3296574  3297693  3299259  3301029  3311671  causal              maskgit-maskprob-fix              train_dyn_new_arch-bugfixed-temporal-shift\r\n3290295  3292258  3292338  3296575  3297706  3299272  3301030  3311672  checkpoints_alfred  tokenizer                         train_dyn_yolorun_new_arch\r\n3290296  3292328  3292339  3297569  3297727  3299579  3301031  3313562  coinrun             tokenizer_ckpt_dir                train_lam_minecraft_overfit_sample\r\n3290366  3292329  3294600  3297575  3299016  3300233  3306801  3313563  debug               train_dynamics_lr_schedule_const  train_tokenizer_batch_size_scaling_16_node\r\n3290367  3292330  3294601  3297576  3299062  3300290  3307618  3313564  dyn                 train_dynamics_lr_schedule_cos    train_tokenizer_minecraft_overfit_sample\r\n3290391  3292331  3294602  3297577  3299063  3300658  3307619  3313565  dynamics_ckpt_dir   train_dynamics_lr_schedule_wsd    wrap\r\n3290392  3292332  3294603  3297578  3299065  3300663  3309662  3313570  interactive         train_dyn_causal_180M\r\n3290439  3292333  3296502  3297582  3299066  3300672  3309663  3313571  lam                 train_dyn_causal_255M\r\n3290440  3292334  3296540  3297586  3299068  3301025  3309699  3313572  lam-1-action        train_dyn_causal_356M\r\n]0;tum_cte0515@hkn1990:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/checkpoints]633;D;0",,terminal_output
+67,631264,"TERMINAL",0,0,"cd causal/l",,terminal_command
+68,631312,"TERMINAL",0,0,"]633;E;2025-07-25 16:52:52 cd causal/l;adbf53fe-397b-40d3-9339-94ea79afad56]633;Cbash: cd: causal/l: No such file or directory\r\n]0;tum_cte0515@hkn1990:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/checkpoints]633;D;1",,terminal_output
+69,632746,"TERMINAL",0,0,"cd causal/",,terminal_command
+70,632787,"TERMINAL",0,0,"]633;E;2025-07-25 16:52:53 cd causal/;adbf53fe-397b-40d3-9339-94ea79afad56]633;C]0;tum_cte0515@hkn1990:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/checkpoints/causal]633;D;0",,terminal_output
+71,633679,"TERMINAL",0,0,"ls",,terminal_command
+72,633695,"TERMINAL",0,0,"]633;E;2025-07-25 16:52:54 ls;adbf53fe-397b-40d3-9339-94ea79afad56]633;Coverfit  overfit-seed69-1  overfit-seed69-1-no-noise  train_dynamics_causal_2_node  train_dynamics_causal_8_node\r\n]0;tum_cte0515@hkn1990:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/checkpoints/causal]633;D;0",,terminal_output
+73,639358,"TERMINAL",0,0,"cd train_dynamics_causal_8_node/",,terminal_command
+74,639389,"TERMINAL",0,0,"]633;E;2025-07-25 16:53:00 cd train_dynamics_causal_8_node/;adbf53fe-397b-40d3-9339-94ea79afad56]633;C]0;tum_cte0515@hkn1990:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/checkpoints/causal/train_dynamics_causal_8_node]633;D;0",,terminal_output
+75,639775,"TERMINAL",0,0,"ls",,terminal_command
+76,639812,"TERMINAL",0,0,"]633;E;2025-07-25 16:53:01 ls;adbf53fe-397b-40d3-9339-94ea79afad56]633;C3373408\r\n]0;tum_cte0515@hkn1990:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/checkpoints/causal/train_dynamics_causal_8_node]633;D;0",,terminal_output
+77,641099,"TERMINAL",0,0,"cd 3373408/",,terminal_command
+78,641109,"TERMINAL",0,0,"]633;E;2025-07-25 16:53:02 cd 3373408/;adbf53fe-397b-40d3-9339-94ea79afad56]633;C]0;tum_cte0515@hkn1990:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/checkpoints/causal/train_dynamics_causal_8_node/3373408]633;D;0",,terminal_output
+79,641668,"TERMINAL",0,0,"ls",,terminal_command
+80,641717,"TERMINAL",0,0,"]633;E;2025-07-25 16:53:02 ls;adbf53fe-397b-40d3-9339-94ea79afad56]633;C003000  004000  005000\r\n]0;tum_cte0515@hkn1990:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/checkpoints/causal/train_dynamics_causal_8_node/3373408]633;D;0",,terminal_output
+81,968762,"TERMINAL",0,0,"git branch",,terminal_command
+82,968798,"TERMINAL",0,0,"]633;E;2025-07-25 16:58:30 git branch;adbf53fe-397b-40d3-9339-94ea79afad56]633;Cfatal: not a git repository (or any parent up to mount point /hkfs)\r\nStopping at filesystem boundary (GIT_DISCOVERY_ACROSS_FILESYSTEM not set).\r\n]0;tum_cte0515@hkn1990:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/checkpoints/causal/train_dynamics_causal_8_node/3373408]633;D;128",,terminal_output
+83,971943,"TERMINAL",0,0,"bash",,terminal_focus
+84,973457,"TERMINAL",0,0,"git branch",,terminal_command
+85,973508,"TERMINAL",0,0,"]633;E;2025-07-25 16:58:34 git branch;2f980232-d92f-4231-927c-3ee17e3c6d04]633;C[?1h=\r",,terminal_output
+86,973653,"TERMINAL",0,0,"  add-wandb-name-and-tags\r\n  causal-st-transformer\r\n  causal-transformer-dynamics-model\r\n  convert-to-jax-array-in-iter\r\n  correct-batched-sampling\r\n  dev\r\n  dont-let-tf-see-gpu\r\n  feat/explicit-image-dims\r\n  fix-sampling\r\n  grad-norm-log-and-clip\r\n  grain-dataloader\r\n  logging-variants\r\n  lr-schedules\r\n  main\r\n  maskgit-different-maskprob-per-sample\r\n:",,terminal_output
+87,974834,"TERMINAL",0,0,"\r  metrics-logging-for-dynamics-model\r\n:",,terminal_output
+88,975036,"TERMINAL",0,0,"\r  monkey-patch\r\n:",,terminal_output
+89,975185,"TERMINAL",0,0,"\r* new-arch-sampling\r\n:",,terminal_output
+90,975295,"TERMINAL",0,0,"\r  preprocess_video\r\n:",,terminal_output
+91,975531,"TERMINAL",0,0,"\r  refactor-tmp\r\n:",,terminal_output
+92,975700,"TERMINAL",0,0,"\r  revised-dataloader\r\n:",,terminal_output
+93,975801,"TERMINAL",0,0,"\r  runner\r\n:",,terminal_output
+94,975984,"TERMINAL",0,0,"\r  runner-grain\r\n:",,terminal_output
+95,976103,"TERMINAL",0,0,"\r  sample-from-different-topologies\r\n:",,terminal_output
+96,976214,"TERMINAL",0,0,"\r  speedup-tfrecord-preprocessing\r\n:",,terminal_output
+97,976418,"TERMINAL",0,0,"\r  tmp\r\n:",,terminal_output
+98,976534,"TERMINAL",0,0,"\r\r(END)",,terminal_output
+99,976646,"TERMINAL",0,0,"\r\r(END)",,terminal_output
+100,976827,"TERMINAL",0,0,"[?25l\r\r(END)[?25h",,terminal_output
+101,976967,"TERMINAL",0,0,"[?25l\r\r(END)[?25h",,terminal_output
+102,977071,"TERMINAL",0,0,"\r[?1l>]0;tum_cte0515@hkn1990:~/Projects/jafar]633;D;0",,terminal_output
+103,979880,"TERMINAL",0,0,"bash",,terminal_focus
+104,1001754,"TERMINAL",0,0,"sacct",,terminal_command
+105,1001800,"TERMINAL",0,0,"]633;E;2025-07-25 16:59:03 sacct;406cfb31-2341-454a-afa8-cae7781806b2]633;CJobID           JobName  Partition    Account  AllocCPUS      State ExitCode \r\n------------ ---------- ---------- ---------- ---------- ---------- -------- \r\n3371238      train_dyn+ accelerat+ hk-projec+         48    RUNNING      0:0 \r\n3371238.bat+      batch            hk-projec+         24    RUNNING      0:0 \r\n3371238.ext+     extern            hk-projec+         48    RUNNING      0:0 \r\n3371238.0        python            hk-projec+         40    RUNNING      0:0 \r\n3372629      train_dyn+ accelerat+ hk-projec+        192  COMPLETED      0:0 \r\n3372629.bat+      batch            hk-projec+         24  COMPLETED      0:0 \r\n3372629.ext+     extern            hk-projec+        192  COMPLETED      0:0 \r\n3372629.0        python            hk-projec+        160  COMPLETED      0:0 \r\n3372631      train_dyn+ accelerat+ hk-projec+         48  COMPLETED      0:0 \r\n3372631.bat+      batch            hk-projec+         24  COMPLETED      0:0 \r\n3372631.ext+     extern            hk-projec+         48  COMPLETED      0:0 \r\n3372631.0        python            hk-projec+         40  COMPLETED      0:0 \r\n3373407      train_dyn+ accelerat+ hk-projec+         48    RUNNING      0:0 \r\n3373407.bat+      batch            hk-projec+         24    RUNNING      0:0 \r\n3373407.ext+     extern            hk-projec+         48    RUNNING      0:0 \r\n3373407.0        python            hk-projec+         40    RUNNING      0:0 \r\n3373408      train_dyn+ accelerat+ hk-projec+        192    RUNNING      0:0 \r\n3373408.bat+      batch            hk-projec+         24    RUNNING      0:0 \r\n3373408.ext+     extern            hk-projec+        192    RUNNING      0:0 \r\n3373408.0        python            hk-projec+        160    RUNNING      0:0 \r\n3371237      train_dyn+ accelerat+ hk-projec+        192    RUNNING      0:0 \r\n3371237.bat+      batch            hk-projec+         24    RUNNING      0:0 \r\n3371237.ext+     extern            hk-projec+        192    RUNNING      0:0 \r\n3371237.0        python            hk-projec+        160    RUNNING      0:0 \r\n]0;tum_cte0515@hkn1990:~/Projects/jafar]633;D;0]633;P;Cwd=/home/hk-project-p0023960/tum_cte0515/Projects/jafar",,terminal_output
+106,1011767,"TERMINAL",0,0,"alias",,terminal_command
+107,1011813,"TERMINAL",0,0,"]633;E;2025-07-25 16:59:13 alias;406cfb31-2341-454a-afa8-cae7781806b2]633;Calias egrep='egrep --color=auto'\r\nalias fgrep='fgrep --color=auto'\r\nalias fqueue='watch -n 1 ""squeue -o \""%.10i %.16P %.30j %.8u %.8T %.10M %.9l %.6D %R\""""'\r\nalias fsacct_week='sacct --format=""JobID%15,JobName%30,Partition%16,AllocCPUS%3,State%12,Elapsed%10,Timelimit%10"" --starttime $(date -d ""last week"" +%Y-%m-%d) | grep -vE ""*.batch|*.extern|*.inter|bash|python|CANCELLED|echo""'\r\nalias grep='grep --color=auto'\r\nalias idle='sinfo_t_idle'\r\nalias idling='watch -n1 sinfo_t_idle'\r\nalias l.='ls -d .* --color=auto'\r\nalias ll='ls -l --color=auto'\r\nalias ls='ls --color=auto'\r\nalias mc='. /usr/libexec/mc/mc-wrapper.sh'\r\nalias queue='watch -n1 squeue --me'\r\nalias runner='cd /home/hk-project-p0023960/tum_cte0515/Projects/jafar_jobs/'\r\nalias runner-2='cd /home/hk-project-p0023960/tum_cte0515/Projects/jafar_jobs_2/'\r\nalias salloc_cpu='salloc --time=01:00:00 --partition=dev_cpuonly --nodes=1 --cpus-per-task=128'\r\nalias salloc_node='salloc --time=01:00:00 --partition=dev_accelerated --nodes=1 --gres=gpu:1  --cpus-per-task=8'\r\nalias smi='watch -n1 nvidia-smi'\r\nalias sync-runner='sh /home/hk-project-p0023960/tum_cte0515/sync_runner.sh /home/hk-project-p0023960/tum_cte0515/Projects/jafar /home/hk-project-p0023960/tum_cte0515/Projects/jafar_jobs/'\r\nalias sync-runner-2='sh /home/hk-project-p0023960/tum_cte0515/sync_runner.sh /home/hk-project-p0023960/tum_cte0515/Projects/jafar /home/hk-project-p0023960/tum_cte0515/Projects/jafar_jobs_2/'\r\nalias xzegrep='xzegrep --color=auto'\r\nalias xzfgrep='xzfgrep --color=auto'\r\nalias xzgrep='xzgrep --color=auto'\r\nalias zegrep='zegrep --color=auto'\r\nalias zfgrep='zfgrep --color=auto'\r\nalias zgrep='zgrep --color=auto'\r\n]0;tum_cte0515@hkn1990:~/Projects/jafar]633;D;0",,terminal_output
+108,1037050,"TERMINAL",0,0,"sacct --format=""JobID%15,JobName%30,Partition%16,AllocCPUS%3,State%12,Elapsed%10,Timelimit%10"" --starttime $(date -d ""last week"" +%Y-%m-%d) | grep -vE ""*.batch|*.extern|*.inter"" | grep ""accelerate""",,terminal_command
+109,1037128,"TERMINAL",0,0,"]633;E;2025-07-25 16:59:38 sacct --format=""JobID%15,JobName%30,Partition%16,AllocCPUS%3,State%12,Elapsed%10,Timelimit%10"" --starttime $(date -d ""last week"" +%Y-%m-%d) | grep -vE ""*.batch|*.extern|*.inter"" | grep ""accelerate"";406cfb31-2341-454a-afa8-cae7781806b2]633;C        3358457     train_dyn_yolorun_new_arch      accelerated  48       FAILED   00:00:28 2-00:00:00 \r\n        3359333                           wrap      accelerated   6 CANCELLED b+   00:00:18   10:00:00 \r\n        3359334                           wrap      accelerated   6      TIMEOUT   10:00:29   10:00:00 \r\n        3359338                           wrap      accelerated   6      TIMEOUT   10:00:16   10:00:00 \r\n        3359343 train_dyn_new_arch-bugfixed-s+      accelerated  48    COMPLETED   23:19:14 2-00:00:00 \r\n        3359349 train_dyn_new_arch-bugfixed-t+      accelerated  48    COMPLETED 1-01:00:55 2-00:00:00 \r\n        3365872 train_dynamics_overfit_sample+      accelerated   0 CANCELLED b+   00:00:00 2-00:00:00 \r\n        3365873 train_dynamics_overfit_sample+      accelerated   6    COMPLETED   01:26:52 2-00:00:00 \r\n        3365876 train_dynamics_overfit_sample+      accelerated   6    COMPLETED   01:40:07 2-00:00:00 \r\n        3366843 train_dynamics_overfit_sample+      accelerated   0 CANCELLED b+   00:00:00 2-00:00:00 \r\n        3366883 train_dynamics_overfit_sample+      accelerated   6    COMPLETED   01:33:31 2-00:00:00 \r\n        3370768 train_dynamics_maskprob_fix_8+      accelerated   0 CANCELLED b+   00:00:00 2-00:00:00 \r\n        3370769 train_dynamics_maskprob_fix_8+      accelerated   0 CANCELLED b+   00:00:00 2-00:00:00 \r\n        3370787 train_dynamics_maskprob_fix_8+      accelerated   0 CANCELLED b+   00:00:00   00:15:00 \r\n        3370788 train_dynamics_maskprob_fix_8+  dev_accelerated   6 CANCELLED b+   00:05:54   00:15:00 \r\n        3370822 train_dynamics_maskprob_fix_8+      accelerated   0 CANCELLED b+   00:00:00 2-00:00:00 \r\n        3371238 train_dynamics_maskprob_fix_2+      accelerated  48      RUNNING 1-16:25:23 2-00:00:00 \r\n        3372629 train_dynamics_maskprob_fix_8+      accelerated 192    COMPLETED 1-02:29:22 2-00:00:00 \r\n        3372631 train_dynamics_maskprob_fix_2+      accelerated  48    COMPLETED 1-01:17:59 2-00:00:00 \r\n        3372929               train_dyn_causal  dev_accelerated   0 CANCELLED b+   00:00:00   00:10:00 \r\n        3372931          train_dyn_causal_180M  dev_accelerated   6       FAILED   00:00:33   00:10:00 \r\n        3372932          train_dyn_causal_255M  dev_accelerated   6       FAILED   00:00:29   00:10:00 \r\n        3372934          train_dyn_causal_356M  dev_accelerated   6       FAILED   00:00:29   00:10:00 \r\n        3372936          train_dyn_causal_500M  dev_accelerated   6       FAILED   00:00:29   00:10:00 \r\n        3372963          train_dyn_causal_180M  dev_accelerated   6 CANCELLED b+   00:00:29   00:10:00 \r\n        3372964          train_dyn_causal_255M  dev_accelerated   0 CANCELLED b+   00:00:00   00:10:00 \r\n        3372965          train_dyn_causal_356M  dev_accelerated   0 CANCELLED b+   00:00:00   00:10:00 \r\n        3372966          train_dyn_causal_500M  dev_accelerated   0 CANCELLED b+   00:00:00   00:10:00 \r\n        3372969          train_dyn_causal_180M  dev_accelerated   6       FAILED   00:02:11   00:10:00 \r\n        3372970          train_dyn_causal_255M  dev_accelerated   6       FAILED   00:02:24   00:10:00 \r\n        3372971          train_dyn_causal_356M  dev_accelerated   6       FAILED   00:02:08   00:10:00 \r\n        3372972          train_dyn_causal_500M  dev_accelerated   6       FAILED   00:02:09   00:10:00 \r\n        3373107          train_dyn_causal_180M  dev_accelerated   6    COMPLETED   00:06:15   00:10:00 \r\n        3373108          train_dyn_causal_255M  dev_accelerated   6    COMPLETED   00:07:14   00:10:00 \r\n        3373109          train_dyn_causal_356M  dev_accelerated   6       FAILED   00:04:17   00:10:00 \r\n        3373110          train_dyn_causal_500M  dev_accelerated   6       FAILED   00:04:59   00:10:00 \r\n        3373205   train_dynamics_causal_2_node      accelerated   0 CANCELLED b+   00:00:00 2-00:00:00 \r\n        3373207   train_dynamics_causal_8_node      accelerated   0 CANCELLED b+   00:00:00 2-00:00:00 \r\n        3373213   train_dynamics_causal_8_node      accelerated   0 CANCELLED b+   00:00:00 2-00:00:00 \r\n        3373276   train_dynamics_causal_2_node      accelerated   0 CANCELLED b+   00:00:00 2-00:00:00 \r\n        3373277   train_dynamics_causal_8_node      accelerated   0 CANCELLED b+   00:00:00 2-00:00:00 \r\n        3373400                           wrap      accelerated   6    COMPLETED   00:04:34   02:00:00 \r\n        3373404                           wrap      accelerated   6    COMPLETED   00:04:38   02:00:00 \r\n        3373407   train_dynamics_causal_2_node      accelerated  48      RUNNING   06:15:10 2-00:00:00 \r\n        3373408   train_dynamics_causal_8_node      accelerated 192      RUNNING   06:15:10 2-00:00:00 \r\n        3373409                           wrap      accelerated   6    COMPLETED   00:41:24   02:00:00 \r\n        3373410                           wrap      accelerated   6    COMPLETED   00:42:56   02:00:00 \r\n        3371237 train_dynamics_maskprob_fix_8+      accelerated 192      RUNNING   01:02:16 2-00:00:00 \r\n]0;tum_cte0515@hkn1990:~/Projects/jafar]633;D;0",,terminal_output
+110,5563097,"TERMINAL",0,0,"queue",,terminal_command
+111,5563185,"TERMINAL",0,0,"]633;E;2025-07-25 18:15:03 queue;406cfb31-2341-454a-afa8-cae7781806b2]633;C[?1049h(B[?7hEvery 1.0s: squeue --mehkn1990.localdomain: Fri Jul 25 18:15:03 2025JOBID PARTITION     NAME     USER ST\tTIME  NODES NODELIST(REASON)3373407 accelerat train_dy tum_cte0  R    7:30:35\t  2 hkn[0501,0506]3373408 accelerat train_dy tum_cte0  R    7:30:35\t  8 hkn[0406,0409,0411,0413,0421,0423-0424,0525]3371237 accelerat train_dy tum_cte0  R    2:17:41\t  8 hkn[0410,0429,0520,0532,0607,0610,0810,0814]3371238 accelerat train_dy tum_cte0  R 1-17:40:48\t  2 hkn[0706,0710]",,terminal_output
+112,5564311,"TERMINAL",0,0,"46629",,terminal_output
+113,5565250,"TERMINAL",0,0,"577350",,terminal_output
+114,5566469,"TERMINAL",0,0,"69952",,terminal_output
+115,5567659,"TERMINAL",0,0,"8404063",,terminal_output
+116,5568855,"TERMINAL",0,0,"91174",,terminal_output
+117,5569632,"TERMINAL",0,0,"102285",,terminal_output
+118,5570209,"TERMINAL",0,0,"13396",,terminal_output
+119,5571324,"TERMINAL",0,0,"244507",,terminal_output
+120,5572586,"TERMINAL",0,0,"35518",,terminal_output
+121,5573472,"TERMINAL",0,0,"46629",,terminal_output
+122,5574531,"TERMINAL",0,0,"57731:00",,terminal_output
+123,5575451,"TERMINAL",0,0,"68841",,terminal_output
+124,5575776,"TERMINAL",0,0,"[?1049l\r[?1l>]0;tum_cte0515@hkn1990:~/Projects/jafar]633;D;0",,terminal_output
+125,5576793,"slurm/jobs/mihir/horeka/mask_prob_fix/train_dynamics_8_nodes.sbatch",0,0,"",shellscript,tab

927a8af5474e5654810c00ce2e09fd2de87d3e5722f33fa1090d867db114e403/crowd-code-50d32311-0df8-4235-b597-7d69f06b72151752666637262-2025_07_16-13.50.55.636/source.csv

@@ -0,0 +1,143 @@
+Sequence,Time,File,RangeOffset,RangeLength,Text,Language,Type
+2,209,"extension-output-pdoom-org.crowd-code-#1-crowd-code",0,0,"",Log,tab
+3,327,"extension-output-pdoom-org.crowd-code-#1-crowd-code",0,0,"1:50:55 PM [info] Activating crowd-code\n1:50:55 PM [info] Recording started\n1:50:55 PM [info] Initializing git provider using file system watchers...\n1:50:55 PM [info] Git repository found\n1:50:55 PM [info] Git provider initialized successfully\n1:50:55 PM [info] Initial git state: [object Object]\n",Log,content
+4,3004,"TERMINAL",0,0,"/bin/python3 /hkfs/home/project/hk-project-p0023960/tum_cte0515/.cursor-server/extensions/ms-python.python-2024.12.3-linux-x64/python_files/printEnvVariablesToFile.py /hkfs/home/project/hk-project-p0023960/tum_cte0515/.cursor-server/extensions/ms-python.python-2024.12.3-linux-x64/python_files/deactivate/bash/envVars.txt",,terminal_command
+5,3050,"TERMINAL",0,0,"]633;E;2025-07-16 13:50:58 /bin/python3 /hkfs/home/project/hk-project-p0023960/tum_cte0515/.cursor-server/extensions/ms-python.python-2024.12.3-linux-x64/python_files/printEnvVariablesToFile.py /hkfs/home/project/hk-project-p0023960/tum_cte0515/.cursor-server/extensions/ms-python.python-2024.12.3-linux-x64/python_files/deactivate/bash/envVars.txt;8d1bd821-a8b0-4413-aad3-038e7b28f0ba]633;C",,terminal_output
+6,3072,"TERMINAL",0,0,"]0;tum_cte0515@hkn1991:/hkfs/home/project/hk-project-p0023960/tum_cte0515/.cursor-server/extensions/ms-python.python-2024.12.3-linux-x64/python_files/deactivate/bash]633;D;0",,terminal_output
+7,12393,"TERMINAL",0,0,"bash",,terminal_focus
+8,14724,"TERMINAL",0,0,"queue",,terminal_command
+9,14783,"TERMINAL",0,0,"]633;E;2025-07-16 13:51:10 queue;97c203bb-2de3-4bf0-b19e-fa122ab0b933]633;C[?1049h(B[?7hEvery 1.0s: squeue --mehkn1991.localdomain: Wed Jul 16 13:51:10 2025JOBID PARTITION     NAME     USER ST\tTIME  NODES NODELIST(REASON)3348592 accelerat train_dy tum_cte0  R   13:50:48\t  2 hkn[0416,0421]3349982 accelerat interact tum_cte0  R\t9:47\t  2 hkn[0407,0511]3348397 accelerat train_dy tum_cte0  R   19:12:30\t  2 hkn[0810,0815]3348399 accelerat train_dy tum_cte0  R   19:12:30\t  2 hkn[0601,0603]3348400 accelerat train_dy tum_cte0  R   19:12:30\t  2 hkn[0604,0608]3350111 accelerat interact tum_cte0  R\t9:03\t  2 hkn[0415,0422]3345116 accelerat train_dy tum_cte0  R 1-19:56:15\t  2 hkn[0503,0506]",,terminal_output
+10,15818,"TERMINAL",0,0,"19811146",,terminal_output
+11,16869,"TERMINAL",0,0,"250922257",,terminal_output
+12,17926,"TERMINAL",0,0,"315033368",,terminal_output
+13,18971,"TERMINAL",0,0,"42144479",,terminal_output
+14,20019,"TERMINAL",0,0,"532555820",,terminal_output
+15,21074,"TERMINAL",0,0,"64366691",,terminal_output
+16,22124,"TERMINAL",0,0,"754777102",,terminal_output
+17,23156,"TERMINAL",0,0,"86588813",,terminal_output
+18,24214,"TERMINAL",0,0,"97699924",,terminal_output
+19,25251,"TERMINAL",0,0,"208740404035",,terminal_output
+20,25893,"TERMINAL",0,0,"[?1049l\r[?1l>]0;tum_cte0515@hkn1991:~/Projects/jafar]633;D;0",,terminal_output
+21,27896,"TERMINAL",0,0,"idling",,terminal_command
+22,27951,"TERMINAL",0,0,"]633;E;2025-07-16 13:51:23 idling;97c203bb-2de3-4bf0-b19e-fa122ab0b933]633;C[?1049h(B[?7hEvery 1.0s: sinfo_t_idlehkn1991.localdomain: Wed Jul 16 13:51:23 2025Partition dev_cpuonly:\t 9 nodes idle\rPartition cpuonly:\t 3 nodes idle\rPartition dev_accelerated:\t 0 nodes idle\rPartition accelerated:\t 9 nodes idle\rPartition dev_accelerated-h100    :\t 0 nodes idle\rPartition accelerated-h100:\t 0 nodes idle\rPartition large:\t 7 nodes idle",,terminal_output
+23,28992,"TERMINAL",0,0,"4",,terminal_output
+24,29405,"TERMINAL",0,0,"bash",,terminal_focus
+25,30037,"TERMINAL",0,0,"5",,terminal_output
+26,31088,"TERMINAL",0,0,"6",,terminal_output
+27,32138,"TERMINAL",0,0,"7",,terminal_output
+28,32742,"TERMINAL",0,0,"salloc --time=10:00:00 --partition=accelerated --nodes=2 --ntasks-per-node=4 --gres=gpu:4 --cpus-per-task=5",,terminal_command
+29,32797,"TERMINAL",0,0,"]633;E;2025-07-16 13:51:28 salloc --time=10:00:00 --partition=accelerated --nodes=2 --ntasks-per-node=4 --gres=gpu:4 --cpus-per-task=5;dea9d5fc-91fd-447c-886d-4b0240ae057d]633;Csalloc: Granted job allocation 3350245\r\n",,terminal_output
+30,32908,"TERMINAL",0,0,"salloc: Waiting for resource configuration\r\n",,terminal_output
+31,33181,"TERMINAL",0,0,"87",,terminal_output
+32,34023,"TERMINAL",0,0,"watch",,terminal_focus
+33,34249,"TERMINAL",0,0,"9",,terminal_output
+34,35275,"TERMINAL",0,0,"30",,terminal_output
+35,35956,"TERMINAL",0,0,"[?1049l\r[?1l>]0;tum_cte0515@hkn1991:~/Projects/jafar]633;D;0",,terminal_output
+36,36736,"TERMINAL",0,0,"queue",,terminal_command
+37,36791,"TERMINAL",0,0,"]633;E;2025-07-16 13:51:32 queue;97c203bb-2de3-4bf0-b19e-fa122ab0b933]633;C[?1049h(B[?7hEvery 1.0s: squeue --mehkn1991.localdomain: Wed Jul 16 13:51:32 2025JOBID PARTITION     NAME     USER ST\tTIME  NODES NODELIST(REASON)3348592 accelerat train_dy tum_cte0  R   13:51:10\t  2 hkn[0416,0421]3349982 accelerat interact tum_cte0  R10:09\t  2 hkn[0407,0511]3348397 accelerat train_dy tum_cte0  R   19:12:52\t  2 hkn[0810,0815]3348399 accelerat train_dy tum_cte0  R   19:12:52\t  2 hkn[0601,0603]3348400 accelerat train_dy tum_cte0  R   19:12:52\t  2 hkn[0604,0608]3350111 accelerat interact tum_cte0  R\t9:25\t  2 hkn[0415,0422]3350245 accelerat interact tum_cte0  R\t0:04\t  2 hkn[0423-0424]3345116 accelerat train_dy tum_cte0  R 1-19:56:37\t  2 hkn[0503,0506]",,terminal_output
+38,37831,"TERMINAL",0,0,"3110333658",,terminal_output
+39,39190,"TERMINAL",0,0,"421444769",,terminal_output
+40,40941,"TERMINAL",0,0,"5436669841",,terminal_output
+41,41454,"TERMINAL",0,0,"[?1049l\r[?1l>]0;tum_cte0515@hkn1991:~/Projects/jafar]633;D;0",,terminal_output
+42,48631,"TERMINAL",0,0,"scancel 3350111",,terminal_command
+43,48681,"TERMINAL",0,0,"]633;E;2025-07-16 13:51:44 scancel 3350111;97c203bb-2de3-4bf0-b19e-fa122ab0b933]633;C",,terminal_output
+44,48714,"TERMINAL",0,0,"]0;tum_cte0515@hkn1991:~/Projects/jafar]633;D;0]633;P;Cwd=/home/hk-project-p0023960/tum_cte0515/Projects/jafar",,terminal_output
+45,49689,"TERMINAL",0,0,"queue",,terminal_command
+46,49762,"TERMINAL",0,0,"]633;E;2025-07-16 13:51:45 queue;97c203bb-2de3-4bf0-b19e-fa122ab0b933]633;C[?1049h(B[?7hEvery 1.0s: squeue --mehkn1991.localdomain: Wed Jul 16 13:51:45 2025JOBID PARTITION     NAME     USER ST\tTIME  NODES NODELIST(REASON)3350111 accelerat interact tum_cte0 CG\t9:37\t  2 hkn[0415,0422]3348592 accelerat train_dy tum_cte0  R   13:51:23\t  2 hkn[0416,0421]3349982 accelerat interact tum_cte0  R10:22\t  2 hkn[0407,0511]3348397 accelerat train_dy tum_cte0  R   19:13:05\t  2 hkn[0810,0815]3348399 accelerat train_dy tum_cte0  R   19:13:05\t  2 hkn[0601,0603]3348400 accelerat train_dy tum_cte0  R   19:13:05\t  2 hkn[0604,0608]3350245 accelerat interact tum_cte0  R\t0:17\t  2 hkn[0423-0424]3345116 accelerat train_dy tum_cte0  R 1-19:56:50\t  2 hkn[0503,0506]",,terminal_output
+47,50848,"TERMINAL",0,0,"64366681",,terminal_output
+48,51854,"TERMINAL",0,0,"75477792",,terminal_output
+49,52906,"TERMINAL",0,0,"865888203",,terminal_output
+50,53965,"TERMINAL",0,0,"97699914",,terminal_output
+51,55013,"TERMINAL",0,0,"508710101025",,terminal_output
+52,56066,"TERMINAL",0,0,"19811136",,terminal_output
+53,57130,"TERMINAL",0,0,"230922247",,terminal_output
+54,57167,"TERMINAL",0,0,"[?1049l\r[?1l>]0;tum_cte0515@hkn1991:~/Projects/jafar]633;D;0",,terminal_output
+55,59809,"TERMINAL",0,0,"scancel 3349982",,terminal_command
+56,59820,"TERMINAL",0,0,"]633;E;2025-07-16 13:51:55 scancel 3349982;97c203bb-2de3-4bf0-b19e-fa122ab0b933]633;C]0;tum_cte0515@hkn1991:~/Projects/jafar]633;D;0]633;P;Cwd=/home/hk-project-p0023960/tum_cte0515/Projects/jafar",,terminal_output
+57,60880,"TERMINAL",0,0,"salloc: Nodes hkn[0423-0424] are ready for job\r\n",,terminal_output
+58,61361,"TERMINAL",0,0,"queue",,terminal_command
+59,61421,"TERMINAL",0,0,"]633;E;2025-07-16 13:51:56 queue;97c203bb-2de3-4bf0-b19e-fa122ab0b933]633;C[?1049h(B[?7hEvery 1.0s: squeue --mehkn1991.localdomain: Wed Jul 16 13:51:56 2025JOBID PARTITION     NAME     USER ST\tTIME  NODES NODELIST(REASON)3349982 accelerat interact tum_cte0 CG10:32\t  2 hkn[0407,0511]3350111 accelerat interact tum_cte0 CG\t9:37\t  1 hkn04153348592 accelerat train_dy tum_cte0  R   13:51:34\t  2 hkn[0416,0421]3348397 accelerat train_dy tum_cte0  R   19:13:16\t  2 hkn[0810,0815]3348399 accelerat train_dy tum_cte0  R   19:13:16\t  2 hkn[0601,0603]3348400 accelerat train_dy tum_cte0  R   19:13:16\t  2 hkn[0604,0608]3350245 accelerat interact tum_cte0  R\t0:28\t  2 hkn[0423-0424]3345116 accelerat train_dy tum_cte0  R 1-19:57:01\t  2 hkn[0503,0506]",,terminal_output
+60,61987,"TERMINAL",0,0,"]0;tum_cte0515@hkn0423:~/Projects/jafar[?2004h[tum_cte0515@hkn0423 jafar]$ ",,terminal_output
+61,62457,"TERMINAL",0,0,"76888303",,terminal_output
+62,63517,"TERMINAL",0,0,"9799914",,terminal_output
+63,64554,"TERMINAL",0,0,"2:00820202025",,terminal_output
+64,65608,"TERMINAL",0,0,"1911136",,terminal_output
+65,66695,"TERMINAL",0,0,"24022247",,terminal_output
+66,67699,"TERMINAL",0,0,"3133358",,terminal_output
+67,68748,"TERMINAL",0,0,"4244469",,terminal_output
+68,69815,"TERMINAL",0,0,"53555710",,terminal_output
+69,70853,"TERMINAL",0,0,"61 hkn0407466681",,terminal_output
+70,71899,"TERMINAL",0,0,"7577792",,terminal_output
+71,73029,"TERMINAL",0,0,"86888403",,terminal_output
+72,74047,"TERMINAL",0,0,"9799914",,terminal_output
+73,75046,"TERMINAL",0,0,"10830303025",,terminal_output
+74,76160,"TERMINAL",0,0,"1911136",,terminal_output
+75,77184,"TERMINAL",0,0,"25022247",,terminal_output
+76,78194,"TERMINAL",0,0,"3133358",,terminal_output
+77,79248,"TERMINAL",0,0,"4244469",,terminal_output
+78,80291,"TERMINAL",0,0,"53555720",,terminal_output
+79,81385,"TERMINAL",0,0,"6466681",,terminal_output
+80,82408,"TERMINAL",0,0,"7577792",,terminal_output
+81,83532,"TERMINAL",0,0,"86888503",,terminal_output
+82,84556,"TERMINAL",0,0,"9840404025",,terminal_output
+83,85537,"TERMINAL",0,0,"21911136",,terminal_output
+84,86581,"TERMINAL",0,0,"22:0022247",,terminal_output
+85,87738,"TERMINAL",0,0,"3133358",,terminal_output
+86,88674,"TERMINAL",0,0,"4244469",,terminal_output
+87,89798,"TERMINAL",0,0,"\r548592train_dy R   13:52:03\t  2 hkn[0416,0421]3979:13:45810,081595601,060340054,060850245interact    0:57423-042445116train_dy1-19:57:30503,0506",,terminal_output
+88,90806,"TERMINAL",0,0,"6466681",,terminal_output
+89,91909,"TERMINAL",0,0,"7577792",,terminal_output
+90,92943,"TERMINAL",0,0,"868881:003",,terminal_output
+91,93977,"TERMINAL",0,0,"9799914",,terminal_output
+92,95002,"TERMINAL",0,0,"30850505025",,terminal_output
+93,96032,"TERMINAL",0,0,"1911136",,terminal_output
+94,97062,"TERMINAL",0,0,"21022247",,terminal_output
+95,98121,"TERMINAL",0,0,"3133358",,terminal_output
+96,99204,"TERMINAL",0,0,"4244469",,terminal_output
+97,100210,"TERMINAL",0,0,"53555740",,terminal_output
+98,101290,"TERMINAL",0,0,"\r6859train_dy R   13:52:14\t  2 hkn[0416,0421]3979:13:56810,081596601,060340064,060850245interact    1:08423-042445116train_dy1-19:57:41503,0506",,terminal_output
+99,102334,"TERMINAL",0,0,"7577792",,terminal_output
+100,103359,"TERMINAL",0,0,"86888103",,terminal_output
+101,104412,"TERMINAL",0,0,"9799914",,terminal_output
+102,105476,"TERMINAL",0,0,"4094:014:014:0136",,terminal_output
+103,106533,"TERMINAL",0,0,"22022247",,terminal_output
+104,107610,"TERMINAL",0,0,"3133358",,terminal_output
+105,108608,"TERMINAL",0,0,"4244469",,terminal_output
+106,109651,"TERMINAL",0,0,"53555750",,terminal_output
+107,110697,"TERMINAL",0,0,"6466681",,terminal_output
+108,111743,"TERMINAL",0,0,"7577792",,terminal_output
+109,112789,"TERMINAL",0,0,"86888203",,terminal_output
+110,113948,"TERMINAL",0,0,"9799914",,terminal_output
+111,114973,"TERMINAL",0,0,"50810101025",,terminal_output
+112,115997,"TERMINAL",0,0,"1911136",,terminal_output
+113,117011,"TERMINAL",0,0,"23022247",,terminal_output
+114,118148,"TERMINAL",0,0,"3133358",,terminal_output
+115,119171,"TERMINAL",0,0,"4244469",,terminal_output
+116,119335,"TERMINAL",0,0,"[?1049l\r[?1l>]0;tum_cte0515@hkn1991:~/Projects/jafar]633;D;0",,terminal_output
+117,120747,"TERMINAL",0,0,"smi",,terminal_command
+118,120813,"TERMINAL",0,0,"]633;E;2025-07-16 13:52:56 smi;97c203bb-2de3-4bf0-b19e-fa122ab0b933]633;C[?1049h(B[?7hEvery 1.0s: nvidia-smihkn1991.localdomain: Wed Jul 16 13:52:56 2025sh: line 1: nvidia-smi: command not found",,terminal_output
+119,121696,"TERMINAL",0,0,"[?1049l\r[?1l>]0;tum_cte0515@hkn1991:~/Projects/jafar]633;D;0",,terminal_output
+120,122902,"TERMINAL",0,0,"srun",,terminal_focus
+121,124100,"TERMINAL",0,0,"s",,terminal_output
+122,124160,"TERMINAL",0,0,"m",,terminal_output
+123,124220,"TERMINAL",0,0,"i",,terminal_output
+124,124434,"TERMINAL",0,0,"\r\n[?2004l\r",,terminal_output
+125,126034,"TERMINAL",0,0,"[?1049h(B[?7hEvery 1.0s: nvidia-smihkn0423.localdomain: Wed Jul 16 13:52:59 2025Wed Jul 16 13:53:00 2025\r+-----------------------------------------------------------------------------------------+\r| NVIDIA-SMI 570.133.20Driver Version: 570.133.20     CUDA Version: 12.8     |\r|-----------------------------------------+------------------------+----------------------+\r| GPU  NamePersistence-M | Bus-IdDisp.A | Volatile Uncorr. ECC |\r| Fan  Temp   PerfPwr:Usage/Cap |Memory-Usage | GPU-Util  Compute M. |\r|||MIG M. |\r|=========================================+========================+======================|\r|   0  NVIDIA A100-SXM4-40GBOn  |   00000000:31:00.0 Off |0 |\r| N/A   44C    P051W /  300W |\t 27MiB /  40960MiB |\t  0%\t  Default |\r|||Disabled |\r+-----------------------------------------+------------------------+----------------------+\r|   1  NVIDIA A100-SXM4-40GBOn  |   00000000:4B:00.0 Off |0 |\r| N/A   44C    P054W /  300W |\t 27MiB /  40960MiB |\t  0%\t  Default |\r|||Disabled |\r+-----------------------------------------+------------------------+----------------------+\r|   2  NVIDIA A100-SXM4-40GBOn  |   00000000:CA:00.0 Off |0 |\r| N/A   44C    P052W /  300W |\t 27MiB /  40960MiB |\t  0%\t  Default |\r|||Disabled |\r+-----------------------------------------+------------------------+----------------------+\r|   3  NVIDIA A100-SXM4-40GBOn  |   00000000:E3:00.0 Off |0 |\r| N/A   44C    P053W /  300W |\t 27MiB /  40960MiB |\t  0%\t  Default |\r|||Disabled |\r+-----------------------------------------+------------------------+----------------------+\r+-----------------------------------------------------------------------------------------+\r| Processes:|\r|  GPU   GI   CIPID   Type   Process nameGPU Memory |\r|ID   IDUsage\t  |\r|=========================================================================================|\r|    0   N/A  N/A2966G   /usr/libexec/Xorg17MiB |\r|    1   N/A  N/A2966G   /usr/libexec/Xorg17MiB |\r|    2   N/A  N/A2966G   /usr/libexec/Xorg17MiB |\r|    3   N/A  N/A2966G   /usr/libexec/Xorg17MiB |\r+-----------------------------------------------------------------------------------------+",,terminal_output
+126,126701,"TERMINAL",0,0,"[?1049l\r[?1l>]0;tum_cte0515@hkn0423:~/Projects/jafar[?2004h[tum_cte0515@hkn0423 jafar]$ ",,terminal_output
+127,127637,"TERMINAL",0,0,"s",,terminal_output
+128,127696,"TERMINAL",0,0,"[?25lo[?25h",,terminal_output
+129,127836,"TERMINAL",0,0,"[?25lu[?25h",,terminal_output
+130,127899,"TERMINAL",0,0,"[?25lr[?25h",,terminal_output
+131,128110,"TERMINAL",0,0,"[?25lc[?25h",,terminal_output
+132,128230,"TERMINAL",0,0,"[?25le[?25h",,terminal_output
+133,128289,"TERMINAL",0,0,"[?25l [?25h",,terminal_output
+134,128466,"TERMINAL",0,0,"[?25lv[?25h",,terminal_output
+135,128982,"TERMINAL",0,0,"[?25l.[?25h",,terminal_output
+136,129039,"TERMINAL",0,0,"[?25lv[?25h",,terminal_output
+137,129181,"TERMINAL",0,0,"env/",,terminal_output
+138,129467,"TERMINAL",0,0,"[?25lb[?25h",,terminal_output
+139,129530,"TERMINAL",0,0,"in/",,terminal_output
+140,130084,"TERMINAL",0,0,"[?25la[?25h",,terminal_output
+141,130844,"TERMINAL",0,0,"[?25lc[?25h",,terminal_output
+142,131268,"TERMINAL",0,0,"tivate",,terminal_output
+143,131542,"TERMINAL",0,0,"[?25l[?2004l\r[?25h]0;tum_cte0515@hkn0423:~/Projects/jafar[?2004h(jafar) [tum_cte0515@hkn0423 jafar]$ ",,terminal_output

927a8af5474e5654810c00ce2e09fd2de87d3e5722f33fa1090d867db114e403/crowd-code-527e6f79-68be-4672-b584-2b15b1cf78281754060417734-2025_08_01-17.00.27.872/source.csv

@@ -0,0 +1,1141 @@
+Sequence,Time,File,RangeOffset,RangeLength,Text,Language,Type
+2,142,"tasks",0,0,"",Log,tab
+3,3516,"extension-output-pdoom-org.crowd-code-#1-crowd-code",0,0,"5:00:27 PM [info] Activating crowd-code\n5:00:27 PM [info] Recording started\n5:00:27 PM [info] Initializing git provider using file system watchers...\n5:00:28 PM [info] Git repository found\n5:00:28 PM [info] Git provider initialized successfully\n5:00:28 PM [info] Initial git state: [object Object]\n",Log,tab
+4,5906,"TERMINAL",0,0,"undefined[tum_cte0515@hkn1990 jafar]$ queue",,terminal_command
+5,5967,"TERMINAL",0,0,"]633;E;2025-08-01 17:00:33 queue;adbf53fe-397b-40d3-9339-94ea79afad56]633;C[?1049h(B[?7hEvery 1.0s: squeue --mehkn1990.localdomain: Fri Aug  1 17:00:33 2025JOBID PARTITION     NAME     USER ST\tTIME  NODES NODELIST(REASON)3390458 accelerat train_dy tum_cte0 PD\t0:00     12 (Priority)3388151 accelerat train_to tum_cte0  R    5:57:25\t  2 hkn[0604-0605]3388153 accelerat train_to tum_cte0  R    5:57:25\t  2 hkn[0607-0608]3389801 accelerat train_dy tum_cte0  R    4:33:06\t  3 hkn[0405,0415,0417]",,terminal_output
+6,7009,"TERMINAL",0,0,"4667",,terminal_output
+7,8074,"TERMINAL",0,0,"5778",,terminal_output
+8,9126,"TERMINAL",0,0,"6889",,terminal_output
+9,10156,"TERMINAL",0,0,"79910",,terminal_output
+10,11214,"TERMINAL",0,0,"830301",,terminal_output
+11,12231,"TERMINAL",0,0,"9223",,terminal_output
+12,13305,"TERMINAL",0,0,"41334",,terminal_output
+13,14310,"TERMINAL",0,0,"2445",,terminal_output
+14,15392,"TERMINAL",0,0,"3556",,terminal_output
+15,16410,"TERMINAL",0,0,"4667",,terminal_output
+16,17465,"TERMINAL",0,0,"5778",,terminal_output
+17,18491,"TERMINAL",0,0,"6889",,terminal_output
+18,19534,"TERMINAL",0,0,"79920",,terminal_output
+19,20586,"TERMINAL",0,0,"840401",,terminal_output
+20,21648,"TERMINAL",0,0,"9112",,terminal_output
+21,22695,"TERMINAL",0,0,"50223",,terminal_output
+22,23716,"TERMINAL",0,0,"1334",,terminal_output
+23,24795,"TERMINAL",0,0,"2445",,terminal_output
+24,25808,"TERMINAL",0,0,"3556",,terminal_output
+25,26867,"TERMINAL",0,0,"4667",,terminal_output
+26,27941,"TERMINAL",0,0,"5778",,terminal_output
+27,28977,"TERMINAL",0,0,"6889",,terminal_output
+28,30015,"TERMINAL",0,0,"79930",,terminal_output
+29,31063,"TERMINAL",0,0,"850501",,terminal_output
+30,32184,"TERMINAL",0,0,"9112",,terminal_output
+31,33175,"TERMINAL",0,0,"1:00223",,terminal_output
+32,34221,"TERMINAL",0,0,"1334",,terminal_output
+33,35266,"TERMINAL",0,0,"2556",,terminal_output
+34,36298,"TERMINAL",0,0,"4667",,terminal_output
+35,37331,"TERMINAL",0,0,"5778",,terminal_output
+36,38399,"TERMINAL",0,0,"6889",,terminal_output
+37,39433,"TERMINAL",0,0,"79940",,terminal_output
+38,40497,"TERMINAL",0,0,"88:008:001",,terminal_output
+39,41540,"TERMINAL",0,0,"9112",,terminal_output
+40,42592,"TERMINAL",0,0,"10223",,terminal_output
+41,43615,"TERMINAL",0,0,"1334",,terminal_output
+42,44676,"TERMINAL",0,0,"2445",,terminal_output
+43,45741,"TERMINAL",0,0,"3556",,terminal_output
+44,46761,"TERMINAL",0,0,"4667",,terminal_output
+45,47822,"TERMINAL",0,0,"5778",,terminal_output
+46,48854,"TERMINAL",0,0,"6889",,terminal_output
+47,49895,"TERMINAL",0,0,"79950",,terminal_output
+48,50968,"TERMINAL",0,0,"810101",,terminal_output
+49,51989,"TERMINAL",0,0,"9112",,terminal_output
+50,53040,"TERMINAL",0,0,"20223",,terminal_output
+51,54111,"TERMINAL",0,0,"1334",,terminal_output
+52,55180,"TERMINAL",0,0,"2445",,terminal_output
+53,56226,"TERMINAL",0,0,"3556",,terminal_output
+54,57223,"TERMINAL",0,0,"4778",,terminal_output
+55,58282,"TERMINAL",0,0,"6889",,terminal_output
+56,59321,"TERMINAL",0,0,"7994:00",,terminal_output
+57,60374,"TERMINAL",0,0,"820201",,terminal_output
+58,61416,"TERMINAL",0,0,"9112",,terminal_output
+59,62476,"TERMINAL",0,0,"30223",,terminal_output
+60,63513,"TERMINAL",0,0,"1334",,terminal_output
+61,64582,"TERMINAL",0,0,"2445",,terminal_output
+62,65615,"TERMINAL",0,0,"3556",,terminal_output
+63,66708,"TERMINAL",0,0,"4667",,terminal_output
+64,67742,"TERMINAL",0,0,"5778",,terminal_output
+65,68754,"TERMINAL",0,0,"6889",,terminal_output
+66,69826,"TERMINAL",0,0,"79910",,terminal_output
+67,70906,"TERMINAL",0,0,"830301",,terminal_output
+68,71912,"TERMINAL",0,0,"9112",,terminal_output
+69,72979,"TERMINAL",0,0,"40223",,terminal_output
+70,74002,"TERMINAL",0,0,"1334",,terminal_output
+71,75076,"TERMINAL",0,0,"2445",,terminal_output
+72,76163,"TERMINAL",0,0,"3556",,terminal_output
+73,77175,"TERMINAL",0,0,"4667",,terminal_output
+74,78180,"TERMINAL",0,0,"5778",,terminal_output
+75,79233,"TERMINAL",0,0,"69920",,terminal_output
+76,80281,"TERMINAL",0,0,"840401",,terminal_output
+77,81337,"TERMINAL",0,0,"9112",,terminal_output
+78,82382,"TERMINAL",0,0,"50223",,terminal_output
+79,83467,"TERMINAL",0,0,"1334",,terminal_output
+80,84468,"TERMINAL",0,0,"2445",,terminal_output
+81,85520,"TERMINAL",0,0,"3556",,terminal_output
+82,86609,"TERMINAL",0,0,"4667",,terminal_output
+83,87650,"TERMINAL",0,0,"5778",,terminal_output
+84,88666,"TERMINAL",0,0,"6889",,terminal_output
+85,89710,"TERMINAL",0,0,"79930",,terminal_output
+86,90756,"TERMINAL",0,0,"850501",,terminal_output
+87,91804,"TERMINAL",0,0,"9112",,terminal_output
+88,92956,"TERMINAL",0,0,"2:00223",,terminal_output
+89,93978,"TERMINAL",0,0,"1334",,terminal_output
+90,95007,"TERMINAL",0,0,"2445",,terminal_output
+91,96005,"TERMINAL",0,0,"3556",,terminal_output
+92,97028,"TERMINAL",0,0,"4667",,terminal_output
+93,98080,"TERMINAL",0,0,"5778",,terminal_output
+94,99133,"TERMINAL",0,0,"6889",,terminal_output
+95,100183,"TERMINAL",0,0,"79940",,terminal_output
+96,101264,"TERMINAL",0,0,"89:009:001",,terminal_output
+97,102304,"TERMINAL",0,0,"10223",,terminal_output
+98,103325,"TERMINAL",0,0,"1334",,terminal_output
+99,104426,"TERMINAL",0,0,"2445",,terminal_output
+100,105450,"TERMINAL",0,0,"3556",,terminal_output
+101,106459,"TERMINAL",0,0,"4667",,terminal_output
+102,107496,"TERMINAL",0,0,"5778",,terminal_output
+103,108624,"TERMINAL",0,0,"6889",,terminal_output
+104,109647,"TERMINAL",0,0,"79950",,terminal_output
+105,110641,"TERMINAL",0,0,"810101",,terminal_output
+106,111721,"TERMINAL",0,0,"9112",,terminal_output
+107,112734,"TERMINAL",0,0,"20223",,terminal_output
+108,113785,"TERMINAL",0,0,"1334",,terminal_output
+109,114837,"TERMINAL",0,0,"2445",,terminal_output
+110,115938,"TERMINAL",0,0,"3556",,terminal_output
+111,116971,"TERMINAL",0,0,"4667",,terminal_output
+112,117974,"TERMINAL",0,0,"5778",,terminal_output
+113,119127,"TERMINAL",0,0,"6889",,terminal_output
+114,120065,"TERMINAL",0,0,"7995:00",,terminal_output
+115,121147,"TERMINAL",0,0,"820201",,terminal_output
+116,122171,"TERMINAL",0,0,"9112",,terminal_output
+117,123212,"TERMINAL",0,0,"30223",,terminal_output
+118,124287,"TERMINAL",0,0,"1445",,terminal_output
+119,125304,"TERMINAL",0,0,"3556",,terminal_output
+120,126346,"TERMINAL",0,0,"4667",,terminal_output
+121,127464,"TERMINAL",0,0,"5778",,terminal_output
+122,128489,"TERMINAL",0,0,"6889",,terminal_output
+123,129512,"TERMINAL",0,0,"79910",,terminal_output
+124,130641,"TERMINAL",0,0,"830301",,terminal_output
+125,131636,"TERMINAL",0,0,"9112",,terminal_output
+126,132688,"TERMINAL",0,0,"40223",,terminal_output
+127,133820,"TERMINAL",0,0,"1334",,terminal_output
+128,134837,"TERMINAL",0,0,"2445",,terminal_output
+129,135828,"TERMINAL",0,0,"3556",,terminal_output
+130,136878,"TERMINAL",0,0,"4667",,terminal_output
+131,137926,"TERMINAL",0,0,"5778",,terminal_output
+132,138975,"TERMINAL",0,0,"6889",,terminal_output
+133,140024,"TERMINAL",0,0,"79920",,terminal_output
+134,141074,"TERMINAL",0,0,"840401",,terminal_output
+135,142194,"TERMINAL",0,0,"9112",,terminal_output
+136,143544,"TERMINAL",0,0,"50334",,terminal_output
+137,144670,"TERMINAL",0,0,"2445",,terminal_output
+138,145628,"TERMINAL",0,0,"3556",,terminal_output
+139,146714,"TERMINAL",0,0,"4667",,terminal_output
+140,147731,"TERMINAL",0,0,"5778",,terminal_output
+141,148778,"TERMINAL",0,0,"6889",,terminal_output
+142,149888,"TERMINAL",0,0,"79930",,terminal_output
+143,151034,"TERMINAL",0,0,"850501",,terminal_output
+144,152019,"TERMINAL",0,0,"9112",,terminal_output
+145,152970,"TERMINAL",0,0,"3:00223",,terminal_output
+146,154090,"TERMINAL",0,0,"1334",,terminal_output
+147,155069,"TERMINAL",0,0,"2445",,terminal_output
+148,156144,"TERMINAL",0,0,"3556",,terminal_output
+149,157199,"TERMINAL",0,0,"4667",,terminal_output
+150,158228,"TERMINAL",0,0,"5778",,terminal_output
+151,159306,"TERMINAL",0,0,"69940",,terminal_output
+152,160306,"TERMINAL",0,0,"86:00:006:00:001",,terminal_output
+153,161365,"TERMINAL",0,0,"9112",,terminal_output
+154,162417,"TERMINAL",0,0,"10223",,terminal_output
+155,163510,"TERMINAL",0,0,"1334",,terminal_output
+156,164505,"TERMINAL",0,0,"2445",,terminal_output
+157,165559,"TERMINAL",0,0,"3556",,terminal_output
+158,166713,"TERMINAL",0,0,"4667",,terminal_output
+159,167661,"TERMINAL",0,0,"5778",,terminal_output
+160,168738,"TERMINAL",0,0,"6889",,terminal_output
+161,169755,"TERMINAL",0,0,"79950",,terminal_output
+162,170797,"TERMINAL",0,0,"810101",,terminal_output
+163,171847,"TERMINAL",0,0,"9112",,terminal_output
+164,172898,"TERMINAL",0,0,"20223",,terminal_output
+165,173942,"TERMINAL",0,0,"1334",,terminal_output
+166,175010,"TERMINAL",0,0,"2445",,terminal_output
+167,176086,"TERMINAL",0,0,"3556",,terminal_output
+168,177149,"TERMINAL",0,0,"4667",,terminal_output
+169,178154,"TERMINAL",0,0,"5778",,terminal_output
+170,179181,"TERMINAL",0,0,"6889",,terminal_output
+171,180306,"TERMINAL",0,0,"7996:00",,terminal_output
+172,181270,"TERMINAL",0,0,"921212",,terminal_output
+173,182359,"TERMINAL",0,0,"30223",,terminal_output
+174,183376,"TERMINAL",0,0,"1334",,terminal_output
+175,184423,"TERMINAL",0,0,"2445",,terminal_output
+176,185476,"TERMINAL",0,0,"3556",,terminal_output
+177,186521,"TERMINAL",0,0,"4667",,terminal_output
+178,187571,"TERMINAL",0,0,"5778",,terminal_output
+179,188626,"TERMINAL",0,0,"6889",,terminal_output
+180,189700,"TERMINAL",0,0,"79910",,terminal_output
+181,190719,"TERMINAL",0,0,"830301",,terminal_output
+182,191765,"TERMINAL",0,0,"9112",,terminal_output
+183,192865,"TERMINAL",0,0,"40223",,terminal_output
+184,193952,"TERMINAL",0,0,"1334",,terminal_output
+185,194949,"TERMINAL",0,0,"2445",,terminal_output
+186,195972,"TERMINAL",0,0,"3556",,terminal_output
+187,197096,"TERMINAL",0,0,"4667",,terminal_output
+188,198174,"TERMINAL",0,0,"5778",,terminal_output
+189,199145,"TERMINAL",0,0,"6889",,terminal_output
+190,200184,"TERMINAL",0,0,"79920",,terminal_output
+191,201218,"TERMINAL",0,0,"840401",,terminal_output
+192,202280,"TERMINAL",0,0,"9223",,terminal_output
+193,203325,"TERMINAL",0,0,"51334",,terminal_output
+194,204369,"TERMINAL",0,0,"2445",,terminal_output
+195,205412,"TERMINAL",0,0,"3556",,terminal_output
+196,206475,"TERMINAL",0,0,"4667",,terminal_output
+197,207510,"TERMINAL",0,0,"5778",,terminal_output
+198,208563,"TERMINAL",0,0,"6889",,terminal_output
+199,209651,"TERMINAL",0,0,"79930",,terminal_output
+200,210718,"TERMINAL",0,0,"850501",,terminal_output
+201,211712,"TERMINAL",0,0,"9112",,terminal_output
+202,212762,"TERMINAL",0,0,"4:00223",,terminal_output
+203,213890,"TERMINAL",0,0,"1334",,terminal_output
+204,214852,"TERMINAL",0,0,"2445",,terminal_output
+205,216042,"TERMINAL",0,0,"3556",,terminal_output
+206,216964,"TERMINAL",0,0,"4667",,terminal_output
+207,217997,"TERMINAL",0,0,"5778",,terminal_output
+208,219044,"TERMINAL",0,0,"6889",,terminal_output
+209,220184,"TERMINAL",0,0,"79940",,terminal_output
+210,221159,"TERMINAL",0,0,"81:001:001",,terminal_output
+211,222195,"TERMINAL",0,0,"9112",,terminal_output
+212,223266,"TERMINAL",0,0,"10334",,terminal_output
+213,224297,"TERMINAL",0,0,"2445",,terminal_output
+214,225345,"TERMINAL",0,0,"3556",,terminal_output
+215,226445,"TERMINAL",0,0,"4667",,terminal_output
+216,227514,"TERMINAL",0,0,"5778",,terminal_output
+217,228536,"TERMINAL",0,0,"6889",,terminal_output
+218,229538,"TERMINAL",0,0,"79950",,terminal_output
+219,230585,"TERMINAL",0,0,"810101",,terminal_output
+220,231634,"TERMINAL",0,0,"9112",,terminal_output
+221,232733,"TERMINAL",0,0,"20223",,terminal_output
+222,233761,"TERMINAL",0,0,"1334",,terminal_output
+223,234777,"TERMINAL",0,0,"2445",,terminal_output
+224,235829,"TERMINAL",0,0,"3556",,terminal_output
+225,236868,"TERMINAL",0,0,"4667",,terminal_output
+226,237919,"TERMINAL",0,0,"5778",,terminal_output
+227,238981,"TERMINAL",0,0,"6889",,terminal_output
+228,240008,"TERMINAL",0,0,"7997:00",,terminal_output
+229,241098,"TERMINAL",0,0,"820201",,terminal_output
+230,242173,"TERMINAL",0,0,"9112",,terminal_output
+231,243275,"TERMINAL",0,0,"30223",,terminal_output
+232,244209,"TERMINAL",0,0,"1334",,terminal_output
+233,245258,"TERMINAL",0,0,"2556",,terminal_output
+234,246304,"TERMINAL",0,0,"4667",,terminal_output
+235,247356,"TERMINAL",0,0,"5778",,terminal_output
+236,248507,"TERMINAL",0,0,"6889",,terminal_output
+237,249526,"TERMINAL",0,0,"79910",,terminal_output
+238,250553,"TERMINAL",0,0,"830301",,terminal_output
+239,251539,"TERMINAL",0,0,"9112",,terminal_output
+240,252580,"TERMINAL",0,0,"40223",,terminal_output
+241,253631,"TERMINAL",0,0,"1334",,terminal_output
+242,254710,"TERMINAL",0,0,"2445",,terminal_output
+243,255726,"TERMINAL",0,0,"3556",,terminal_output
+244,256772,"TERMINAL",0,0,"4667",,terminal_output
+245,257817,"TERMINAL",0,0,"5778",,terminal_output
+246,258874,"TERMINAL",0,0,"6889",,terminal_output
+247,259972,"TERMINAL",0,0,"79920",,terminal_output
+248,260963,"TERMINAL",0,0,"840401",,terminal_output
+249,262018,"TERMINAL",0,0,"9112",,terminal_output
+250,263188,"TERMINAL",0,0,"50223",,terminal_output
+251,264172,"TERMINAL",0,0,"1334",,terminal_output
+252,265201,"TERMINAL",0,0,"2445",,terminal_output
+253,266212,"TERMINAL",0,0,"3556",,terminal_output
+254,267256,"TERMINAL",0,0,"4778",,terminal_output
+255,268308,"TERMINAL",0,0,"6889",,terminal_output
+256,269362,"TERMINAL",0,0,"79930",,terminal_output
+257,270397,"TERMINAL",0,0,"850501",,terminal_output
+258,271443,"TERMINAL",0,0,"9112",,terminal_output
+259,272520,"TERMINAL",0,0,"5:00223",,terminal_output
+260,273600,"TERMINAL",0,0,"1334",,terminal_output
+261,274616,"TERMINAL",0,0,"2445",,terminal_output
+262,275615,"TERMINAL",0,0,"3556",,terminal_output
+263,276680,"TERMINAL",0,0,"4667",,terminal_output
+264,277710,"TERMINAL",0,0,"5778",,terminal_output
+265,278814,"TERMINAL",0,0,"6889",,terminal_output
+266,279815,"TERMINAL",0,0,"79940",,terminal_output
+267,280964,"TERMINAL",0,0,"82:002:001",,terminal_output
+268,281899,"TERMINAL",0,0,"9112",,terminal_output
+269,282951,"TERMINAL",0,0,"10223",,terminal_output
+270,283999,"TERMINAL",0,0,"1334",,terminal_output
+271,285042,"TERMINAL",0,0,"2445",,terminal_output
+272,286098,"TERMINAL",0,0,"3556",,terminal_output
+273,287199,"TERMINAL",0,0,"4667",,terminal_output
+274,288263,"TERMINAL",0,0,"5778",,terminal_output
+275,289346,"TERMINAL",0,0,"6889",,terminal_output
+276,290277,"TERMINAL",0,0,"7101051",,terminal_output
+277,291331,"TERMINAL",0,0,"9112",,terminal_output
+278,292378,"TERMINAL",0,0,"20223",,terminal_output
+279,293429,"TERMINAL",0,0,"1334",,terminal_output
+280,294479,"TERMINAL",0,0,"2445",,terminal_output
+281,295604,"TERMINAL",0,0,"3556",,terminal_output
+282,296572,"TERMINAL",0,0,"4667",,terminal_output
+283,297654,"TERMINAL",0,0,"5778",,terminal_output
+284,298657,"TERMINAL",0,0,"6889",,terminal_output
+285,299705,"TERMINAL",0,0,"7998:00",,terminal_output
+286,300806,"TERMINAL",0,0,"820201",,terminal_output
+287,302674,"TERMINAL",0,0,"9223",,terminal_output
+288,303720,"TERMINAL",0,0,"31334",,terminal_output
+289,304674,"TERMINAL",0,0,"2445",,terminal_output
+290,305722,"TERMINAL",0,0,"3556",,terminal_output
+291,306770,"TERMINAL",0,0,"4667",,terminal_output
+292,307812,"TERMINAL",0,0,"5778",,terminal_output
+293,308907,"TERMINAL",0,0,"6889",,terminal_output
+294,309930,"TERMINAL",0,0,"79910",,terminal_output
+295,311005,"TERMINAL",0,0,"830301",,terminal_output
+296,312106,"TERMINAL",0,0,"9112",,terminal_output
+297,313121,"TERMINAL",0,0,"40223",,terminal_output
+298,314146,"TERMINAL",0,0,"1334",,terminal_output
+299,315167,"TERMINAL",0,0,"2445",,terminal_output
+300,316314,"TERMINAL",0,0,"3556",,terminal_output
+301,317243,"TERMINAL",0,0,"4778",,terminal_output
+302,318302,"TERMINAL",0,0,"6889",,terminal_output
+303,319349,"TERMINAL",0,0,"79920",,terminal_output
+304,320396,"TERMINAL",0,0,"840401",,terminal_output
+305,321456,"TERMINAL",0,0,"9112",,terminal_output
+306,322521,"TERMINAL",0,0,"50223",,terminal_output
+307,323554,"TERMINAL",0,0,"1334",,terminal_output
+308,324604,"TERMINAL",0,0,"2445",,terminal_output
+309,325699,"TERMINAL",0,0,"3556",,terminal_output
+310,326720,"TERMINAL",0,0,"4667",,terminal_output
+311,327768,"TERMINAL",0,0,"5778",,terminal_output
+312,328889,"TERMINAL",0,0,"6889",,terminal_output
+313,329915,"TERMINAL",0,0,"79930",,terminal_output
+314,330940,"TERMINAL",0,0,"850501",,terminal_output
+315,332064,"TERMINAL",0,0,"9112",,terminal_output
+316,333093,"TERMINAL",0,0,"6:00223",,terminal_output
+317,334216,"TERMINAL",0,0,"1334",,terminal_output
+318,335134,"TERMINAL",0,0,"2445",,terminal_output
+319,336191,"TERMINAL",0,0,"3556",,terminal_output
+320,337224,"TERMINAL",0,0,"4667",,terminal_output
+321,338270,"TERMINAL",0,0,"6889",,terminal_output
+322,339326,"TERMINAL",0,0,"79940",,terminal_output
+323,340372,"TERMINAL",0,0,"83:003:001",,terminal_output
+324,341425,"TERMINAL",0,0,"9112",,terminal_output
+325,342508,"TERMINAL",0,0,"10223",,terminal_output
+326,343532,"TERMINAL",0,0,"1334",,terminal_output
+327,344658,"TERMINAL",0,0,"2445",,terminal_output
+328,345786,"TERMINAL",0,0,"3556",,terminal_output
+329,346679,"TERMINAL",0,0,"4667",,terminal_output
+330,347713,"TERMINAL",0,0,"5778",,terminal_output
+331,348857,"TERMINAL",0,0,"6889",,terminal_output
+332,349814,"TERMINAL",0,0,"79950",,terminal_output
+333,350865,"TERMINAL",0,0,"810101",,terminal_output
+334,351918,"TERMINAL",0,0,"9112",,terminal_output
+335,352968,"TERMINAL",0,0,"20223",,terminal_output
+336,354021,"TERMINAL",0,0,"1334",,terminal_output
+337,355076,"TERMINAL",0,0,"2445",,terminal_output
+338,356128,"TERMINAL",0,0,"3556",,terminal_output
+339,357212,"TERMINAL",0,0,"4667",,terminal_output
+340,358224,"TERMINAL",0,0,"5778",,terminal_output
+341,359302,"TERMINAL",0,0,"7999:00",,terminal_output
+342,360318,"TERMINAL",0,0,"820201",,terminal_output
+343,361357,"TERMINAL",0,0,"9112",,terminal_output
+344,362582,"TERMINAL",0,0,"30223",,terminal_output
+345,363606,"TERMINAL",0,0,"1334",,terminal_output
+346,364515,"TERMINAL",0,0,"2445",,terminal_output
+347,365562,"TERMINAL",0,0,"3556",,terminal_output
+348,366612,"TERMINAL",0,0,"4667",,terminal_output
+349,367681,"TERMINAL",0,0,"5778",,terminal_output
+350,368710,"TERMINAL",0,0,"6889",,terminal_output
+351,369777,"TERMINAL",0,0,"79910",,terminal_output
+352,370862,"TERMINAL",0,0,"830301",,terminal_output
+353,371896,"TERMINAL",0,0,"9112",,terminal_output
+354,372907,"TERMINAL",0,0,"40223",,terminal_output
+355,374099,"TERMINAL",0,0,"1334",,terminal_output
+356,375072,"TERMINAL",0,0,"2445",,terminal_output
+357,376096,"TERMINAL",0,0,"3556",,terminal_output
+358,377124,"TERMINAL",0,0,"4667",,terminal_output
+359,378166,"TERMINAL",0,0,"5778",,terminal_output
+360,379269,"TERMINAL",0,0,"6889",,terminal_output
+361,380295,"TERMINAL",0,0,"79920",,terminal_output
+362,381273,"TERMINAL",0,0,"841412",,terminal_output
+363,382315,"TERMINAL",0,0,"50223",,terminal_output
+364,383394,"TERMINAL",0,0,"1334",,terminal_output
+365,384516,"TERMINAL",0,0,"2445",,terminal_output
+366,385565,"TERMINAL",0,0,"3556",,terminal_output
+367,386644,"TERMINAL",0,0,"4667",,terminal_output
+368,387656,"TERMINAL",0,0,"5778",,terminal_output
+369,388757,"TERMINAL",0,0,"6889",,terminal_output
+370,389816,"TERMINAL",0,0,"79930",,terminal_output
+371,390843,"TERMINAL",0,0,"850501",,terminal_output
+372,391888,"TERMINAL",0,0,"9112",,terminal_output
+373,392993,"TERMINAL",0,0,"7:00223",,terminal_output
+374,394146,"TERMINAL",0,0,"1334",,terminal_output
+375,395039,"TERMINAL",0,0,"2445",,terminal_output
+376,396170,"TERMINAL",0,0,"3556",,terminal_output
+377,397116,"TERMINAL",0,0,"4667",,terminal_output
+378,398156,"TERMINAL",0,0,"5778",,terminal_output
+379,399207,"TERMINAL",0,0,"6889",,terminal_output
+380,400278,"TERMINAL",0,0,"74:004:0041",,terminal_output
+381,401302,"TERMINAL",0,0,"9112",,terminal_output
+382,402337,"TERMINAL",0,0,"10223",,terminal_output
+383,403378,"TERMINAL",0,0,"1334",,terminal_output
+384,404428,"TERMINAL",0,0,"2445",,terminal_output
+385,405485,"TERMINAL",0,0,"3556",,terminal_output
+386,406611,"TERMINAL",0,0,"4667",,terminal_output
+387,407637,"TERMINAL",0,0,"5778",,terminal_output
+388,408659,"TERMINAL",0,0,"6889",,terminal_output
+389,409683,"TERMINAL",0,0,"79950",,terminal_output
+390,410713,"TERMINAL",0,0,"810101",,terminal_output
+391,411759,"TERMINAL",0,0,"9112",,terminal_output
+392,412806,"TERMINAL",0,0,"20223",,terminal_output
+393,413872,"TERMINAL",0,0,"1334",,terminal_output
+394,414935,"TERMINAL",0,0,"2445",,terminal_output
+395,416055,"TERMINAL",0,0,"3556",,terminal_output
+396,417057,"TERMINAL",0,0,"4667",,terminal_output
+397,418042,"TERMINAL",0,0,"5778",,terminal_output
+398,419085,"TERMINAL",0,0,"6889",,terminal_output
+399,420429,"TERMINAL",0,0,"79940:00",,terminal_output
+400,421257,"TERMINAL",0,0,"820201",,terminal_output
+401,422278,"TERMINAL",0,0,"9223",,terminal_output
+402,423303,"TERMINAL",0,0,"31334",,terminal_output
+403,424326,"TERMINAL",0,0,"2445",,terminal_output
+404,425375,"TERMINAL",0,0,"3556",,terminal_output
+405,426419,"TERMINAL",0,0,"4667",,terminal_output
+406,427474,"TERMINAL",0,0,"5778",,terminal_output
+407,428519,"TERMINAL",0,0,"6889",,terminal_output
+408,429595,"TERMINAL",0,0,"79910",,terminal_output
+409,430607,"TERMINAL",0,0,"830301",,terminal_output
+410,431656,"TERMINAL",0,0,"9112",,terminal_output
+411,432765,"TERMINAL",0,0,"40223",,terminal_output
+412,433890,"TERMINAL",0,0,"1334",,terminal_output
+413,434874,"TERMINAL",0,0,"2445",,terminal_output
+414,435851,"TERMINAL",0,0,"3556",,terminal_output
+415,437167,"TERMINAL",0,0,"4667",,terminal_output
+416,438053,"TERMINAL",0,0,"5778",,terminal_output
+417,438995,"TERMINAL",0,0,"6889",,terminal_output
+418,440051,"TERMINAL",0,0,"79920",,terminal_output
+419,441090,"TERMINAL",0,0,"840401",,terminal_output
+420,442147,"TERMINAL",0,0,"9112",,terminal_output
+421,443182,"TERMINAL",0,0,"50223",,terminal_output
+422,444278,"TERMINAL",0,0,"1445",,terminal_output
+423,445358,"TERMINAL",0,0,"3556",,terminal_output
+424,446323,"TERMINAL",0,0,"4667",,terminal_output
+425,447366,"TERMINAL",0,0,"5778",,terminal_output
+426,448404,"TERMINAL",0,0,"6889",,terminal_output
+427,449524,"TERMINAL",0,0,"79930",,terminal_output
+428,450542,"TERMINAL",0,0,"850501",,terminal_output
+429,451565,"TERMINAL",0,0,"9112",,terminal_output
+430,452592,"TERMINAL",0,0,"8:00223",,terminal_output
+431,453653,"TERMINAL",0,0,"1334",,terminal_output
+432,454690,"TERMINAL",0,0,"2445",,terminal_output
+433,455771,"TERMINAL",0,0,"3556",,terminal_output
+434,456789,"TERMINAL",0,0,"4667",,terminal_output
+435,457834,"TERMINAL",0,0,"5778",,terminal_output
+436,458936,"TERMINAL",0,0,"6889",,terminal_output
+437,459929,"TERMINAL",0,0,"79940",,terminal_output
+438,460980,"TERMINAL",0,0,"85:005:001",,terminal_output
+439,462219,"TERMINAL",0,0,"9112",,terminal_output
+440,463136,"TERMINAL",0,0,"10223",,terminal_output
+441,464131,"TERMINAL",0,0,"1334",,terminal_output
+442,465286,"TERMINAL",0,0,"2445",,terminal_output
+443,466317,"TERMINAL",0,0,"3667",,terminal_output
+444,467286,"TERMINAL",0,0,"5778",,terminal_output
+445,468344,"TERMINAL",0,0,"6889",,terminal_output
+446,469394,"TERMINAL",0,0,"79950",,terminal_output
+447,470424,"TERMINAL",0,0,"810101",,terminal_output
+448,471477,"TERMINAL",0,0,"9112",,terminal_output
+449,472555,"TERMINAL",0,0,"20223",,terminal_output
+450,473674,"TERMINAL",0,0,"1334",,terminal_output
+451,474723,"TERMINAL",0,0,"2445",,terminal_output
+452,475844,"TERMINAL",0,0,"3556",,terminal_output
+453,476723,"TERMINAL",0,0,"4667",,terminal_output
+454,477776,"TERMINAL",0,0,"5778",,terminal_output
+455,478821,"TERMINAL",0,0,"6889",,terminal_output
+456,479874,"TERMINAL",0,0,"7991:00",,terminal_output
+457,480913,"TERMINAL",0,0,"820201",,terminal_output
+458,481983,"TERMINAL",0,0,"9112",,terminal_output
+459,483009,"TERMINAL",0,0,"30223",,terminal_output
+460,484065,"TERMINAL",0,0,"1334",,terminal_output
+461,485152,"TERMINAL",0,0,"2445",,terminal_output
+462,486177,"TERMINAL",0,0,"3556",,terminal_output
+463,487231,"TERMINAL",0,0,"4667",,terminal_output
+464,488235,"TERMINAL",0,0,"5889",,terminal_output
+465,489291,"TERMINAL",0,0,"79910",,terminal_output
+466,490338,"TERMINAL",0,0,"830301",,terminal_output
+467,491388,"TERMINAL",0,0,"9112",,terminal_output
+468,492440,"TERMINAL",0,0,"40223",,terminal_output
+469,493478,"TERMINAL",0,0,"1334",,terminal_output
+470,494532,"TERMINAL",0,0,"2445",,terminal_output
+471,495580,"TERMINAL",0,0,"3556",,terminal_output
+472,496647,"TERMINAL",0,0,"4667",,terminal_output
+473,497672,"TERMINAL",0,0,"5778",,terminal_output
+474,498733,"TERMINAL",0,0,"6889",,terminal_output
+475,499816,"TERMINAL",0,0,"79920",,terminal_output
+476,500817,"TERMINAL",0,0,"840401",,terminal_output
+477,501947,"TERMINAL",0,0,"9112",,terminal_output
+478,502952,"TERMINAL",0,0,"50223",,terminal_output
+479,503999,"TERMINAL",0,0,"1334",,terminal_output
+480,505122,"TERMINAL",0,0,"2445",,terminal_output
+481,506251,"TERMINAL",0,0,"3556",,terminal_output
+482,507191,"TERMINAL",0,0,"4667",,terminal_output
+483,508300,"TERMINAL",0,0,"5889",,terminal_output
+484,509332,"TERMINAL",0,0,"79930",,terminal_output
+485,510337,"TERMINAL",0,0,"850501",,terminal_output
+486,511384,"TERMINAL",0,0,"9112",,terminal_output
+487,512420,"TERMINAL",0,0,"9:00223",,terminal_output
+488,513512,"TERMINAL",0,0,"1334",,terminal_output
+489,514507,"TERMINAL",0,0,"2445",,terminal_output
+490,515567,"TERMINAL",0,0,"3556",,terminal_output
+491,516603,"TERMINAL",0,0,"4667",,terminal_output
+492,517716,"TERMINAL",0,0,"5778",,terminal_output
+493,518834,"TERMINAL",0,0,"6889",,terminal_output
+494,519766,"TERMINAL",0,0,"79940",,terminal_output
+495,520891,"TERMINAL",0,0,"86:006:001",,terminal_output
+496,521916,"TERMINAL",0,0,"9112",,terminal_output
+497,522899,"TERMINAL",0,0,"10223",,terminal_output
+498,523952,"TERMINAL",0,0,"1334",,terminal_output
+499,524999,"TERMINAL",0,0,"2445",,terminal_output
+500,526046,"TERMINAL",0,0,"3556",,terminal_output
+501,527122,"TERMINAL",0,0,"4667",,terminal_output
+502,528146,"TERMINAL",0,0,"5778",,terminal_output
+503,529223,"TERMINAL",0,0,"6889",,terminal_output
+504,530304,"TERMINAL",0,0,"7101051",,terminal_output
+505,531283,"TERMINAL",0,0,"9112",,terminal_output
+506,532328,"TERMINAL",0,0,"20223",,terminal_output
+507,533383,"TERMINAL",0,0,"1334",,terminal_output
+508,534434,"TERMINAL",0,0,"2445",,terminal_output
+509,535535,"TERMINAL",0,0,"3556",,terminal_output
+510,536557,"TERMINAL",0,0,"4667",,terminal_output
+511,537587,"TERMINAL",0,0,"5778",,terminal_output
+512,538710,"TERMINAL",0,0,"6889",,terminal_output
+513,539730,"TERMINAL",0,0,"7992:00",,terminal_output
+514,540786,"TERMINAL",0,0,"820201",,terminal_output
+515,541837,"TERMINAL",0,0,"9112",,terminal_output
+516,542875,"TERMINAL",0,0,"30223",,terminal_output
+517,543929,"TERMINAL",0,0,"1334",,terminal_output
+518,544965,"TERMINAL",0,0,"2445",,terminal_output
+519,546041,"TERMINAL",0,0,"3556",,terminal_output
+520,547065,"TERMINAL",0,0,"4667",,terminal_output
+521,548159,"TERMINAL",0,0,"5778",,terminal_output
+522,549279,"TERMINAL",0,0,"6889",,terminal_output
+523,550283,"TERMINAL",0,0,"79910",,terminal_output
+524,551244,"TERMINAL",0,0,"831312",,terminal_output
+525,552328,"TERMINAL",0,0,"40223",,terminal_output
+526,553355,"TERMINAL",0,0,"1334",,terminal_output
+527,554554,"TERMINAL",0,0,"2445",,terminal_output
+528,555447,"TERMINAL",0,0,"3556",,terminal_output
+529,556486,"TERMINAL",0,0,"4667",,terminal_output
+530,557529,"TERMINAL",0,0,"5778",,terminal_output
+531,558570,"TERMINAL",0,0,"6889",,terminal_output
+532,559643,"TERMINAL",0,0,"79920",,terminal_output
+533,560661,"TERMINAL",0,0,"840401",,terminal_output
+534,561707,"TERMINAL",0,0,"9112",,terminal_output
+535,562771,"TERMINAL",0,0,"50223",,terminal_output
+536,563903,"TERMINAL",0,0,"1334",,terminal_output
+537,564922,"TERMINAL",0,0,"2445",,terminal_output
+538,565948,"TERMINAL",0,0,"3556",,terminal_output
+539,566970,"TERMINAL",0,0,"4667",,terminal_output
+540,568012,"TERMINAL",0,0,"5778",,terminal_output
+541,569120,"TERMINAL",0,0,"6889",,terminal_output
+542,570144,"TERMINAL",0,0,"79930",,terminal_output
+543,571174,"TERMINAL",0,0,"850501",,terminal_output
+544,572255,"TERMINAL",0,0,"9112",,terminal_output
+545,573266,"TERMINAL",0,0,"10:00334",,terminal_output
+546,574314,"TERMINAL",0,0,"2445",,terminal_output
+547,575450,"TERMINAL",0,0,"3556",,terminal_output
+548,576459,"TERMINAL",0,0,"4667",,terminal_output
+549,577517,"TERMINAL",0,0,"5778",,terminal_output
+550,578627,"TERMINAL",0,0,"6889",,terminal_output
+551,579672,"TERMINAL",0,0,"79940",,terminal_output
+552,580693,"TERMINAL",0,0,"87:007:001",,terminal_output
+553,581704,"TERMINAL",0,0,"9112",,terminal_output
+554,582783,"TERMINAL",0,0,"10223",,terminal_output
+555,583866,"TERMINAL",0,0,"1334",,terminal_output
+556,584847,"TERMINAL",0,0,"2445",,terminal_output
+557,585925,"TERMINAL",0,0,"3556",,terminal_output
+558,587042,"TERMINAL",0,0,"4667",,terminal_output
+559,588065,"TERMINAL",0,0,"5778",,terminal_output
+560,589034,"TERMINAL",0,0,"6889",,terminal_output
+561,590107,"TERMINAL",0,0,"79950",,terminal_output
+562,591135,"TERMINAL",0,0,"810101",,terminal_output
+563,592205,"TERMINAL",0,0,"9112",,terminal_output
+564,593265,"TERMINAL",0,0,"20334",,terminal_output
+565,594265,"TERMINAL",0,0,"2445",,terminal_output
+566,595339,"TERMINAL",0,0,"3556",,terminal_output
+567,596709,"TERMINAL",0,0,"4667",,terminal_output
+568,597779,"TERMINAL",0,0,"5778",,terminal_output
+569,598805,"TERMINAL",0,0,"6889",,terminal_output
+570,599871,"TERMINAL",0,0,"7993:00",,terminal_output
+571,600909,"TERMINAL",0,0,"820201",,terminal_output
+572,601953,"TERMINAL",0,0,"9112",,terminal_output
+573,602996,"TERMINAL",0,0,"30223",,terminal_output
+574,604056,"TERMINAL",0,0,"1334",,terminal_output
+575,605103,"TERMINAL",0,0,"2445",,terminal_output
+576,606158,"TERMINAL",0,0,"3556",,terminal_output
+577,607239,"TERMINAL",0,0,"4667",,terminal_output
+578,608362,"TERMINAL",0,0,"5889",,terminal_output
+579,609453,"TERMINAL",0,0,"79910",,terminal_output
+580,610337,"TERMINAL",0,0,"830301",,terminal_output
+581,611386,"TERMINAL",0,0,"9112",,terminal_output
+582,612443,"TERMINAL",0,0,"40223",,terminal_output
+583,613481,"TERMINAL",0,0,"1334",,terminal_output
+584,614532,"TERMINAL",0,0,"2445",,terminal_output
+585,615583,"TERMINAL",0,0,"3556",,terminal_output
+586,616660,"TERMINAL",0,0,"4667",,terminal_output
+587,617683,"TERMINAL",0,0,"5778",,terminal_output
+588,618742,"TERMINAL",0,0,"6889",,terminal_output
+589,619768,"TERMINAL",0,0,"79920",,terminal_output
+590,620833,"TERMINAL",0,0,"840401",,terminal_output
+591,621897,"TERMINAL",0,0,"9112",,terminal_output
+592,623008,"TERMINAL",0,0,"50223",,terminal_output
+593,624010,"TERMINAL",0,0,"1334",,terminal_output
+594,625033,"TERMINAL",0,0,"2445",,terminal_output
+595,626072,"TERMINAL",0,0,"3556",,terminal_output
+596,627445,"TERMINAL",0,0,"4778",,terminal_output
+597,628621,"TERMINAL",0,0,"6889",,terminal_output
+598,629641,"TERMINAL",0,0,"79930",,terminal_output
+599,630668,"TERMINAL",0,0,"850501",,terminal_output
+600,631677,"TERMINAL",0,0,"9112",,terminal_output
+601,632684,"TERMINAL",0,0,"1:00223",,terminal_output
+602,633737,"TERMINAL",0,0,"1334",,terminal_output
+603,634790,"TERMINAL",0,0,"2445",,terminal_output
+604,635853,"TERMINAL",0,0,"3556",,terminal_output
+605,636886,"TERMINAL",0,0,"4667",,terminal_output
+606,637946,"TERMINAL",0,0,"5778",,terminal_output
+607,639023,"TERMINAL",0,0,"6889",,terminal_output
+608,640021,"TERMINAL",0,0,"79940",,terminal_output
+609,641115,"TERMINAL",0,0,"88:008:001",,terminal_output
+610,642131,"TERMINAL",0,0,"9112",,terminal_output
+611,643299,"TERMINAL",0,0,"10223",,terminal_output
+612,644316,"TERMINAL",0,0,"1334",,terminal_output
+613,645306,"TERMINAL",0,0,"2556",,terminal_output
+614,646294,"TERMINAL",0,0,"4667",,terminal_output
+615,647333,"TERMINAL",0,0,"5778",,terminal_output
+616,648384,"TERMINAL",0,0,"6889",,terminal_output
+617,649468,"TERMINAL",0,0,"79950",,terminal_output
+618,650518,"TERMINAL",0,0,"810101",,terminal_output
+619,651590,"TERMINAL",0,0,"9112",,terminal_output
+620,652618,"TERMINAL",0,0,"20223",,terminal_output
+621,653677,"TERMINAL",0,0,"1334",,terminal_output
+622,654724,"TERMINAL",0,0,"2445",,terminal_output
+623,655763,"TERMINAL",0,0,"3556",,terminal_output
+624,656815,"TERMINAL",0,0,"4667",,terminal_output
+625,657869,"TERMINAL",0,0,"5778",,terminal_output
+626,658925,"TERMINAL",0,0,"6889",,terminal_output
+627,659966,"TERMINAL",0,0,"7994:00",,terminal_output
+628,661080,"TERMINAL",0,0,"820201",,terminal_output
+629,662100,"TERMINAL",0,0,"9112",,terminal_output
+630,663128,"TERMINAL",0,0,"30223",,terminal_output
+631,664256,"TERMINAL",0,0,"1334",,terminal_output
+632,665231,"TERMINAL",0,0,"2445",,terminal_output
+633,666250,"TERMINAL",0,0,"3667",,terminal_output
+634,667324,"TERMINAL",0,0,"5778",,terminal_output
+635,668359,"TERMINAL",0,0,"6889",,terminal_output
+636,669397,"TERMINAL",0,0,"79910",,terminal_output
+637,670447,"TERMINAL",0,0,"830301",,terminal_output
+638,671523,"TERMINAL",0,0,"9112",,terminal_output
+639,672652,"TERMINAL",0,0,"40223",,terminal_output
+640,673593,"TERMINAL",0,0,"1334",,terminal_output
+641,674695,"TERMINAL",0,0,"2445",,terminal_output
+642,675722,"TERMINAL",0,0,"3556",,terminal_output
+643,676734,"TERMINAL",0,0,"4667",,terminal_output
+644,677787,"TERMINAL",0,0,"5778",,terminal_output
+645,678837,"TERMINAL",0,0,"6889",,terminal_output
+646,679887,"TERMINAL",0,0,"79920",,terminal_output
+647,680948,"TERMINAL",0,0,"840401",,terminal_output
+648,682027,"TERMINAL",0,0,"9112",,terminal_output
+649,683036,"TERMINAL",0,0,"50223",,terminal_output
+650,684113,"TERMINAL",0,0,"1334",,terminal_output
+651,685139,"TERMINAL",0,0,"2445",,terminal_output
+652,686271,"TERMINAL",0,0,"3556",,terminal_output
+653,687323,"TERMINAL",0,0,"4778",,terminal_output
+654,688316,"TERMINAL",0,0,"6889",,terminal_output
+655,689340,"TERMINAL",0,0,"79930",,terminal_output
+656,690376,"TERMINAL",0,0,"850501",,terminal_output
+657,691417,"TERMINAL",0,0,"9112",,terminal_output
+658,692468,"TERMINAL",0,0,"2:00223",,terminal_output
+659,693511,"TERMINAL",0,0,"1334",,terminal_output
+660,694570,"TERMINAL",0,0,"2445",,terminal_output
+661,695604,"TERMINAL",0,0,"3556",,terminal_output
+662,696678,"TERMINAL",0,0,"4667",,terminal_output
+663,697735,"TERMINAL",0,0,"5778",,terminal_output
+664,698835,"TERMINAL",0,0,"6889",,terminal_output
+665,699917,"TERMINAL",0,0,"79940",,terminal_output
+666,700910,"TERMINAL",0,0,"89:009:001",,terminal_output
+667,701908,"TERMINAL",0,0,"9112",,terminal_output
+668,703060,"TERMINAL",0,0,"10223",,terminal_output
+669,704089,"TERMINAL",0,0,"1334",,terminal_output
+670,705108,"TERMINAL",0,0,"2445",,terminal_output
+671,706132,"TERMINAL",0,0,"3556",,terminal_output
+672,707162,"TERMINAL",0,0,"4667",,terminal_output
+673,708323,"TERMINAL",0,0,"5778",,terminal_output
+674,709339,"TERMINAL",0,0,"69950",,terminal_output
+675,710349,"TERMINAL",0,0,"810101",,terminal_output
+676,711369,"TERMINAL",0,0,"9112",,terminal_output
+677,712435,"TERMINAL",0,0,"20223",,terminal_output
+678,713465,"TERMINAL",0,0,"1334",,terminal_output
+679,714518,"TERMINAL",0,0,"2445",,terminal_output
+680,715602,"TERMINAL",0,0,"3556",,terminal_output
+681,716682,"TERMINAL",0,0,"4667",,terminal_output
+682,717704,"TERMINAL",0,0,"5778",,terminal_output
+683,718708,"TERMINAL",0,0,"6889",,terminal_output
+684,719855,"TERMINAL",0,0,"7995:00",,terminal_output
+685,720879,"TERMINAL",0,0,"820201",,terminal_output
+686,721853,"TERMINAL",0,0,"9112",,terminal_output
+687,722928,"TERMINAL",0,0,"30223",,terminal_output
+688,724156,"TERMINAL",0,0,"1334",,terminal_output
+689,725001,"TERMINAL",0,0,"2445",,terminal_output
+690,726049,"TERMINAL",0,0,"3556",,terminal_output
+691,727104,"TERMINAL",0,0,"4667",,terminal_output
+692,728148,"TERMINAL",0,0,"5778",,terminal_output
+693,729201,"TERMINAL",0,0,"6889",,terminal_output
+694,730261,"TERMINAL",0,0,"7303011",,terminal_output
+695,731330,"TERMINAL",0,0,"9112",,terminal_output
+696,732355,"TERMINAL",0,0,"40223",,terminal_output
+697,733406,"TERMINAL",0,0,"1334",,terminal_output
+698,734457,"TERMINAL",0,0,"2445",,terminal_output
+699,735507,"TERMINAL",0,0,"3556",,terminal_output
+700,736560,"TERMINAL",0,0,"4667",,terminal_output
+701,737609,"TERMINAL",0,0,"5778",,terminal_output
+702,738703,"TERMINAL",0,0,"6889",,terminal_output
+703,739709,"TERMINAL",0,0,"79920",,terminal_output
+704,740753,"TERMINAL",0,0,"840401",,terminal_output
+705,741803,"TERMINAL",0,0,"9112",,terminal_output
+706,742850,"TERMINAL",0,0,"50223",,terminal_output
+707,743904,"TERMINAL",0,0,"1334",,terminal_output
+708,744990,"TERMINAL",0,0,"2445",,terminal_output
+709,746458,"TERMINAL",0,0,"3667",,terminal_output
+710,747499,"TERMINAL",0,0,"5778",,terminal_output
+711,748630,"TERMINAL",0,0,"6889",,terminal_output
+712,749611,"TERMINAL",0,0,"79930",,terminal_output
+713,750679,"TERMINAL",0,0,"850501",,terminal_output
+714,751685,"TERMINAL",0,0,"9112",,terminal_output
+715,752727,"TERMINAL",0,0,"3:00223",,terminal_output
+716,753968,"TERMINAL",0,0,"1334",,terminal_output
+717,754898,"TERMINAL",0,0,"2445",,terminal_output
+718,755927,"TERMINAL",0,0,"3556",,terminal_output
+719,756928,"TERMINAL",0,0,"4667",,terminal_output
+720,758050,"TERMINAL",0,0,"5778",,terminal_output
+721,759075,"TERMINAL",0,0,"6889",,terminal_output
+722,760102,"TERMINAL",0,0,"79940",,terminal_output
+723,761124,"TERMINAL",0,0,"810:0010:001",,terminal_output
+724,762196,"TERMINAL",0,0,"9112",,terminal_output
+725,763302,"TERMINAL",0,0,"10223",,terminal_output
+726,764269,"TERMINAL",0,0,"2445",,terminal_output
+727,765352,"TERMINAL",0,0,"3556",,terminal_output
+728,766447,"TERMINAL",0,0,"4667",,terminal_output
+729,767411,"TERMINAL",0,0,"5778",,terminal_output
+730,768461,"TERMINAL",0,0,"6889",,terminal_output
+731,769621,"TERMINAL",0,0,"79950",,terminal_output
+732,770647,"TERMINAL",0,0,"810101",,terminal_output
+733,771611,"TERMINAL",0,0,"9112",,terminal_output
+734,772651,"TERMINAL",0,0,"20223",,terminal_output
+735,773702,"TERMINAL",0,0,"1334",,terminal_output
+736,774779,"TERMINAL",0,0,"2445",,terminal_output
+737,775806,"TERMINAL",0,0,"3556",,terminal_output
+738,776857,"TERMINAL",0,0,"4667",,terminal_output
+739,777923,"TERMINAL",0,0,"5778",,terminal_output
+740,779019,"TERMINAL",0,0,"6889",,terminal_output
+741,780102,"TERMINAL",0,0,"7996:00",,terminal_output
+742,781052,"TERMINAL",0,0,"820201",,terminal_output
+743,782114,"TERMINAL",0,0,"9112",,terminal_output
+744,783244,"TERMINAL",0,0,"30223",,terminal_output
+745,784382,"TERMINAL",0,0,"1334",,terminal_output
+746,785290,"TERMINAL",0,0,"2556",,terminal_output
+747,786416,"TERMINAL",0,0,"4667",,terminal_output
+748,787373,"TERMINAL",0,0,"5778",,terminal_output
+749,788373,"TERMINAL",0,0,"6889",,terminal_output
+750,789425,"TERMINAL",0,0,"79910",,terminal_output
+751,790463,"TERMINAL",0,0,"830301",,terminal_output
+752,791527,"TERMINAL",0,0,"9112",,terminal_output
+753,792582,"TERMINAL",0,0,"40223",,terminal_output
+754,793659,"TERMINAL",0,0,"1334",,terminal_output
+755,794736,"TERMINAL",0,0,"2445",,terminal_output
+756,795738,"TERMINAL",0,0,"3556",,terminal_output
+757,796758,"TERMINAL",0,0,"4667",,terminal_output
+758,797885,"TERMINAL",0,0,"5778",,terminal_output
+759,798909,"TERMINAL",0,0,"6889",,terminal_output
+760,799932,"TERMINAL",0,0,"79920",,terminal_output
+761,800960,"TERMINAL",0,0,"840401",,terminal_output
+762,802086,"TERMINAL",0,0,"9112",,terminal_output
+763,803031,"TERMINAL",0,0,"50223",,terminal_output
+764,804076,"TERMINAL",0,0,"1334",,terminal_output
+765,805125,"TERMINAL",0,0,"2445",,terminal_output
+766,806169,"TERMINAL",0,0,"3556",,terminal_output
+767,807271,"TERMINAL",0,0,"4667",,terminal_output
+768,808370,"TERMINAL",0,0,"5889",,terminal_output
+769,809392,"TERMINAL",0,0,"79930",,terminal_output
+770,810381,"TERMINAL",0,0,"850501",,terminal_output
+771,811428,"TERMINAL",0,0,"9112",,terminal_output
+772,812477,"TERMINAL",0,0,"4:00223",,terminal_output
+773,813530,"TERMINAL",0,0,"1334",,terminal_output
+774,814579,"TERMINAL",0,0,"2445",,terminal_output
+775,815625,"TERMINAL",0,0,"3556",,terminal_output
+776,816677,"TERMINAL",0,0,"4667",,terminal_output
+777,817755,"TERMINAL",0,0,"5778",,terminal_output
+778,818780,"TERMINAL",0,0,"6889",,terminal_output
+779,819814,"TERMINAL",0,0,"79940",,terminal_output
+780,820929,"TERMINAL",0,0,"81:001:001",,terminal_output
+781,821949,"TERMINAL",0,0,"9112",,terminal_output
+782,822977,"TERMINAL",0,0,"10223",,terminal_output
+783,824104,"TERMINAL",0,0,"1334",,terminal_output
+784,825047,"TERMINAL",0,0,"2445",,terminal_output
+785,826148,"TERMINAL",0,0,"3556",,terminal_output
+786,827170,"TERMINAL",0,0,"4667",,terminal_output
+787,828206,"TERMINAL",0,0,"5778",,terminal_output
+788,829252,"TERMINAL",0,0,"69950",,terminal_output
+789,830296,"TERMINAL",0,0,"810101",,terminal_output
+790,831352,"TERMINAL",0,0,"9112",,terminal_output
+791,832402,"TERMINAL",0,0,"20223",,terminal_output
+792,833441,"TERMINAL",0,0,"1334",,terminal_output
+793,834584,"TERMINAL",0,0,"2445",,terminal_output
+794,835542,"TERMINAL",0,0,"3556",,terminal_output
+795,836679,"TERMINAL",0,0,"4667",,terminal_output
+796,837717,"TERMINAL",0,0,"5778",,terminal_output
+797,838774,"TERMINAL",0,0,"6889",,terminal_output
+798,839804,"TERMINAL",0,0,"7997:00",,terminal_output
+799,840791,"TERMINAL",0,0,"820201",,terminal_output
+800,841836,"TERMINAL",0,0,"9112",,terminal_output
+801,843075,"TERMINAL",0,0,"30223",,terminal_output
+802,843964,"TERMINAL",0,0,"1334",,terminal_output
+803,844996,"TERMINAL",0,0,"2445",,terminal_output
+804,846034,"TERMINAL",0,0,"3556",,terminal_output
+805,847141,"TERMINAL",0,0,"4667",,terminal_output
+806,848114,"TERMINAL",0,0,"5778",,terminal_output
+807,849166,"TERMINAL",0,0,"6889",,terminal_output
+808,850219,"TERMINAL",0,0,"79910",,terminal_output
+809,851256,"TERMINAL",0,0,"831312",,terminal_output
+810,852308,"TERMINAL",0,0,"40223",,terminal_output
+811,853418,"TERMINAL",0,0,"1334",,terminal_output
+812,854412,"TERMINAL",0,0,"2445",,terminal_output
+813,855448,"TERMINAL",0,0,"3556",,terminal_output
+814,856498,"TERMINAL",0,0,"4667",,terminal_output
+815,857547,"TERMINAL",0,0,"5778",,terminal_output
+816,858607,"TERMINAL",0,0,"6889",,terminal_output
+817,859842,"TERMINAL",0,0,"79920",,terminal_output
+818,860866,"TERMINAL",0,0,"840401",,terminal_output
+819,861762,"TERMINAL",0,0,"9112",,terminal_output
+820,862789,"TERMINAL",0,0,"50223",,terminal_output
+821,863836,"TERMINAL",0,0,"1334",,terminal_output
+822,864884,"TERMINAL",0,0,"2445",,terminal_output
+823,865958,"TERMINAL",0,0,"3556",,terminal_output
+824,867211,"TERMINAL",0,0,"4667",,terminal_output
+825,868233,"TERMINAL",0,0,"5778",,terminal_output
+826,869232,"TERMINAL",0,0,"69930",,terminal_output
+827,870272,"TERMINAL",0,0,"850501",,terminal_output
+828,871324,"TERMINAL",0,0,"9112",,terminal_output
+829,872372,"TERMINAL",0,0,"5:00223",,terminal_output
+830,873417,"TERMINAL",0,0,"1334",,terminal_output
+831,874479,"TERMINAL",0,0,"2445",,terminal_output
+832,875510,"TERMINAL",0,0,"3556",,terminal_output
+833,876546,"TERMINAL",0,0,"4667",,terminal_output
+834,877655,"TERMINAL",0,0,"5778",,terminal_output
+835,878647,"TERMINAL",0,0,"6889",,terminal_output
+836,879709,"TERMINAL",0,0,"79940",,terminal_output
+837,880734,"TERMINAL",0,0,"82:002:001",,terminal_output
+838,881789,"TERMINAL",0,0,"9112",,terminal_output
+839,882837,"TERMINAL",0,0,"10223",,terminal_output
+840,883884,"TERMINAL",0,0,"1334",,terminal_output
+841,884955,"TERMINAL",0,0,"2445",,terminal_output
+842,886004,"TERMINAL",0,0,"3556",,terminal_output
+843,887107,"TERMINAL",0,0,"4667",,terminal_output
+844,888080,"TERMINAL",0,0,"5778",,terminal_output
+845,889123,"TERMINAL",0,0,"6889",,terminal_output
+846,890174,"TERMINAL",0,0,"79950",,terminal_output
+847,891454,"TERMINAL",0,0,"810101",,terminal_output
+848,892253,"TERMINAL",0,0,"9223",,terminal_output
+849,893298,"TERMINAL",0,0,"21334",,terminal_output
+850,894348,"TERMINAL",0,0,"2445",,terminal_output
+851,895444,"TERMINAL",0,0,"3556",,terminal_output
+852,896453,"TERMINAL",0,0,"4667",,terminal_output
+853,897486,"TERMINAL",0,0,"5778",,terminal_output
+854,898540,"TERMINAL",0,0,"6889",,terminal_output
+855,899589,"TERMINAL",0,0,"7998:00",,terminal_output
+856,900628,"TERMINAL",0,0,"820201",,terminal_output
+857,901687,"TERMINAL",0,0,"9112",,terminal_output
+858,902745,"TERMINAL",0,0,"30223",,terminal_output
+859,903774,"TERMINAL",0,0,"1334",,terminal_output
+860,904998,"TERMINAL",0,0,"2445",,terminal_output
+861,906020,"TERMINAL",0,0,"3556",,terminal_output
+862,907045,"TERMINAL",0,0,"4667",,terminal_output
+863,908176,"TERMINAL",0,0,"5778",,terminal_output
+864,909101,"TERMINAL",0,0,"6889",,terminal_output
+865,910149,"TERMINAL",0,0,"79910",,terminal_output
+866,911200,"TERMINAL",0,0,"830301",,terminal_output
+867,912270,"TERMINAL",0,0,"9223",,terminal_output
+868,913357,"TERMINAL",0,0,"41334",,terminal_output
+869,914449,"TERMINAL",0,0,"2445",,terminal_output
+870,915444,"TERMINAL",0,0,"3556",,terminal_output
+871,916465,"TERMINAL",0,0,"4667",,terminal_output
+872,917495,"TERMINAL",0,0,"5778",,terminal_output
+873,918513,"TERMINAL",0,0,"6889",,terminal_output
+874,919561,"TERMINAL",0,0,"79920",,terminal_output
+875,920608,"TERMINAL",0,0,"840401",,terminal_output
+876,921669,"TERMINAL",0,0,"9112",,terminal_output
+877,922700,"TERMINAL",0,0,"50223",,terminal_output
+878,923766,"TERMINAL",0,0,"1334",,terminal_output
+879,924792,"TERMINAL",0,0,"2445",,terminal_output
+880,925872,"TERMINAL",0,0,"3556",,terminal_output
+881,926963,"TERMINAL",0,0,"4667",,terminal_output
+882,927974,"TERMINAL",0,0,"5778",,terminal_output
+883,929061,"TERMINAL",0,0,"6889",,terminal_output
+884,930086,"TERMINAL",0,0,"79930",,terminal_output
+885,931123,"TERMINAL",0,0,"850501",,terminal_output
+886,932236,"TERMINAL",0,0,"9112",,terminal_output
+887,933260,"TERMINAL",0,0,"6:00223",,terminal_output
+888,934284,"TERMINAL",0,0,"1445",,terminal_output
+889,935455,"TERMINAL",0,0,"3556",,terminal_output
+890,936434,"TERMINAL",0,0,"4667",,terminal_output
+891,937446,"TERMINAL",0,0,"5778",,terminal_output
+892,938457,"TERMINAL",0,0,"6889",,terminal_output
+893,939501,"TERMINAL",0,0,"79940",,terminal_output
+894,940551,"TERMINAL",0,0,"83:003:001",,terminal_output
+895,941605,"TERMINAL",0,0,"9112",,terminal_output
+896,942659,"TERMINAL",0,0,"10223",,terminal_output
+897,943705,"TERMINAL",0,0,"1334",,terminal_output
+898,944867,"TERMINAL",0,0,"2445",,terminal_output
+899,945860,"TERMINAL",0,0,"3556",,terminal_output
+900,946835,"TERMINAL",0,0,"4667",,terminal_output
+901,947904,"TERMINAL",0,0,"5778",,terminal_output
+902,949029,"TERMINAL",0,0,"6889",,terminal_output
+903,950057,"TERMINAL",0,0,"79950",,terminal_output
+904,951075,"TERMINAL",0,0,"810101",,terminal_output
+905,952080,"TERMINAL",0,0,"9112",,terminal_output
+906,953128,"TERMINAL",0,0,"20223",,terminal_output
+907,954174,"TERMINAL",0,0,"1334",,terminal_output
+908,955228,"TERMINAL",0,0,"2445",,terminal_output
+909,956274,"TERMINAL",0,0,"3667",,terminal_output
+910,957352,"TERMINAL",0,0,"5778",,terminal_output
+911,958369,"TERMINAL",0,0,"6889",,terminal_output
+912,959474,"TERMINAL",0,0,"7999:00",,terminal_output
+913,960496,"TERMINAL",0,0,"820201",,terminal_output
+914,961524,"TERMINAL",0,0,"9112",,terminal_output
+915,962546,"TERMINAL",0,0,"30223",,terminal_output
+916,963776,"TERMINAL",0,0,"1334",,terminal_output
+917,964695,"TERMINAL",0,0,"2445",,terminal_output
+918,965682,"TERMINAL",0,0,"3556",,terminal_output
+919,966750,"TERMINAL",0,0,"4667",,terminal_output
+920,967797,"TERMINAL",0,0,"5778",,terminal_output
+921,968813,"TERMINAL",0,0,"6889",,terminal_output
+922,969892,"TERMINAL",0,0,"79910",,terminal_output
+923,970978,"TERMINAL",0,0,"830301",,terminal_output
+924,971964,"TERMINAL",0,0,"9112",,terminal_output
+925,973097,"TERMINAL",0,0,"40223",,terminal_output
+926,974116,"TERMINAL",0,0,"1334",,terminal_output
+927,975142,"TERMINAL",0,0,"2445",,terminal_output
+928,976166,"TERMINAL",0,0,"3556",,terminal_output
+929,977293,"TERMINAL",0,0,"4667",,terminal_output
+930,978236,"TERMINAL",0,0,"5889",,terminal_output
+931,979280,"TERMINAL",0,0,"79920",,terminal_output
+932,980330,"TERMINAL",0,0,"840401",,terminal_output
+933,981369,"TERMINAL",0,0,"9112",,terminal_output
+934,982441,"TERMINAL",0,0,"50223",,terminal_output
+935,983490,"TERMINAL",0,0,"1334",,terminal_output
+936,984571,"TERMINAL",0,0,"2445",,terminal_output
+937,985622,"TERMINAL",0,0,"3556",,terminal_output
+938,986616,"TERMINAL",0,0,"4667",,terminal_output
+939,987671,"TERMINAL",0,0,"5778",,terminal_output
+940,988759,"TERMINAL",0,0,"6889",,terminal_output
+941,989886,"TERMINAL",0,0,"79930",,terminal_output
+942,990912,"TERMINAL",0,0,"850501",,terminal_output
+943,991871,"TERMINAL",0,0,"9112",,terminal_output
+944,992976,"TERMINAL",0,0,"7:00223",,terminal_output
+945,993968,"TERMINAL",0,0,"1334",,terminal_output
+946,995018,"TERMINAL",0,0,"2445",,terminal_output
+947,996072,"TERMINAL",0,0,"3556",,terminal_output
+948,997135,"TERMINAL",0,0,"4667",,terminal_output
+949,998169,"TERMINAL",0,0,"5778",,terminal_output
+950,999229,"TERMINAL",0,0,"6889",,terminal_output
+951,1000318,"TERMINAL",0,0,"74:004:0041",,terminal_output
+952,1001384,"TERMINAL",0,0,"9112",,terminal_output
+953,1002382,"TERMINAL",0,0,"10223",,terminal_output
+954,1003455,"TERMINAL",0,0,"1334",,terminal_output
+955,1004531,"TERMINAL",0,0,"2445",,terminal_output
+956,1005553,"TERMINAL",0,0,"3556",,terminal_output
+957,1006575,"TERMINAL",0,0,"4667",,terminal_output
+958,1007626,"TERMINAL",0,0,"5778",,terminal_output
+959,1008731,"TERMINAL",0,0,"6889",,terminal_output
+960,1009754,"TERMINAL",0,0,"79950",,terminal_output
+961,1010779,"TERMINAL",0,0,"810101",,terminal_output
+962,1011818,"TERMINAL",0,0,"9112",,terminal_output
+963,1012874,"TERMINAL",0,0,"20223",,terminal_output
+964,1013921,"TERMINAL",0,0,"1334",,terminal_output
+965,1014977,"TERMINAL",0,0,"2445",,terminal_output
+966,1016009,"TERMINAL",0,0,"3556",,terminal_output
+967,1017129,"TERMINAL",0,0,"4667",,terminal_output
+968,1018161,"TERMINAL",0,0,"5778",,terminal_output
+969,1019173,"TERMINAL",0,0,"6889",,terminal_output
+970,1020297,"TERMINAL",0,0,"79950:00",,terminal_output
+971,1021334,"TERMINAL",0,0,"821212",,terminal_output
+972,1022349,"TERMINAL",0,0,"30223",,terminal_output
+973,1023362,"TERMINAL",0,0,"1334",,terminal_output
+974,1024415,"TERMINAL",0,0,"2445",,terminal_output
+975,1025462,"TERMINAL",0,0,"3556",,terminal_output
+976,1026516,"TERMINAL",0,0,"4667",,terminal_output
+977,1027575,"TERMINAL",0,0,"5778",,terminal_output
+978,1028622,"TERMINAL",0,0,"6889",,terminal_output
+979,1029700,"TERMINAL",0,0,"79910",,terminal_output
+980,1030716,"TERMINAL",0,0,"830301",,terminal_output
+981,1031773,"TERMINAL",0,0,"9112",,terminal_output
+982,1032837,"TERMINAL",0,0,"40223",,terminal_output
+983,1033918,"TERMINAL",0,0,"1334",,terminal_output
+984,1034942,"TERMINAL",0,0,"2445",,terminal_output
+985,1035965,"TERMINAL",0,0,"3556",,terminal_output
+986,1036990,"TERMINAL",0,0,"4667",,terminal_output
+987,1038119,"TERMINAL",0,0,"5778",,terminal_output
+988,1039091,"TERMINAL",0,0,"6889",,terminal_output
+989,1040145,"TERMINAL",0,0,"79920",,terminal_output
+990,1041190,"TERMINAL",0,0,"840401",,terminal_output
+991,1042247,"TERMINAL",0,0,"9223",,terminal_output
+992,1043306,"TERMINAL",0,0,"51334",,terminal_output
+993,1044364,"TERMINAL",0,0,"2445",,terminal_output
+994,1045389,"TERMINAL",0,0,"3556",,terminal_output
+995,1046550,"TERMINAL",0,0,"4667",,terminal_output
+996,1047490,"TERMINAL",0,0,"5778",,terminal_output
+997,1048562,"TERMINAL",0,0,"6889",,terminal_output
+998,1049561,"TERMINAL",0,0,"79930",,terminal_output
+999,1050604,"TERMINAL",0,0,"850501",,terminal_output
+1000,1051644,"TERMINAL",0,0,"9112",,terminal_output
+1001,1052763,"TERMINAL",0,0,"8:00223",,terminal_output
+1002,1053729,"TERMINAL",0,0,"1334",,terminal_output
+1003,1055117,"TERMINAL",0,0,"2445",,terminal_output
+1004,1056256,"TERMINAL",0,0,"3556",,terminal_output
+1005,1057334,"TERMINAL",0,0,"4667",,terminal_output
+1006,1058292,"TERMINAL",0,0,"5889",,terminal_output
+1007,1059322,"TERMINAL",0,0,"79940",,terminal_output
+1008,1060347,"TERMINAL",0,0,"85:005:001",,terminal_output
+1009,1061394,"TERMINAL",0,0,"9112",,terminal_output
+1010,1062431,"TERMINAL",0,0,"10223",,terminal_output
+1011,1063538,"TERMINAL",0,0,"1334",,terminal_output
+1012,1064546,"TERMINAL",0,0,"2445",,terminal_output
+1013,1065590,"TERMINAL",0,0,"3556",,terminal_output
+1014,1066673,"TERMINAL",0,0,"4667",,terminal_output
+1015,1067688,"TERMINAL",0,0,"5778",,terminal_output
+1016,1068740,"TERMINAL",0,0,"6889",,terminal_output
+1017,1069790,"TERMINAL",0,0,"79950",,terminal_output
+1018,1070910,"TERMINAL",0,0,"810101",,terminal_output
+1019,1071873,"TERMINAL",0,0,"9112",,terminal_output
+1020,1072915,"TERMINAL",0,0,"20223",,terminal_output
+1021,1073977,"TERMINAL",0,0,"1334",,terminal_output
+1022,1075022,"TERMINAL",0,0,"2445",,terminal_output
+1023,1076063,"TERMINAL",0,0,"3556",,terminal_output
+1024,1077170,"TERMINAL",0,0,"4667",,terminal_output
+1025,1078164,"TERMINAL",0,0,"5778",,terminal_output
+1026,1079292,"TERMINAL",0,0,"6889",,terminal_output
+1027,1080307,"TERMINAL",0,0,"720201:01",,terminal_output
+1028,1081329,"TERMINAL",0,0,"9112",,terminal_output
+1029,1082364,"TERMINAL",0,0,"30223",,terminal_output
+1030,1083412,"TERMINAL",0,0,"1334",,terminal_output
+1031,1084614,"TERMINAL",0,0,"2445",,terminal_output
+1032,1085534,"TERMINAL",0,0,"3556",,terminal_output
+1033,1086553,"TERMINAL",0,0,"4667",,terminal_output
+1034,1087633,"TERMINAL",0,0,"5778",,terminal_output
+1035,1088648,"TERMINAL",0,0,"6889",,terminal_output
+1036,1089697,"TERMINAL",0,0,"79910",,terminal_output
+1037,1090756,"TERMINAL",0,0,"830301",,terminal_output
+1038,1091788,"TERMINAL",0,0,"9112",,terminal_output
+1039,1092843,"TERMINAL",0,0,"40223",,terminal_output
+1040,1093927,"TERMINAL",0,0,"1334",,terminal_output
+1041,1094950,"TERMINAL",0,0,"2445",,terminal_output
+1042,1095972,"TERMINAL",0,0,"3556",,terminal_output
+1043,1097106,"TERMINAL",0,0,"4667",,terminal_output
+1044,1098124,"TERMINAL",0,0,"5778",,terminal_output
+1045,1099132,"TERMINAL",0,0,"6889",,terminal_output
+1046,1100165,"TERMINAL",0,0,"79920",,terminal_output
+1047,1101215,"TERMINAL",0,0,"840401",,terminal_output
+1048,1102275,"TERMINAL",0,0,"9223",,terminal_output
+1049,1103346,"TERMINAL",0,0,"51334",,terminal_output
+1050,1104374,"TERMINAL",0,0,"2445",,terminal_output
+1051,1105396,"TERMINAL",0,0,"3556",,terminal_output
+1052,1106522,"TERMINAL",0,0,"4667",,terminal_output
+1053,1107593,"TERMINAL",0,0,"5778",,terminal_output
+1054,1108571,"TERMINAL",0,0,"6889",,terminal_output
+1055,1109595,"TERMINAL",0,0,"79930",,terminal_output
+1056,1110611,"TERMINAL",0,0,"850501",,terminal_output
+1057,1111673,"TERMINAL",0,0,"9112",,terminal_output
+1058,1112713,"TERMINAL",0,0,"9:00223",,terminal_output
+1059,1113775,"TERMINAL",0,0,"1334",,terminal_output
+1060,1114816,"TERMINAL",0,0,"2445",,terminal_output
+1061,1115867,"TERMINAL",0,0,"3556",,terminal_output
+1062,1116895,"TERMINAL",0,0,"4667",,terminal_output
+1063,1117942,"TERMINAL",0,0,"5778",,terminal_output
+1064,1119016,"TERMINAL",0,0,"6889",,terminal_output
+1065,1120048,"TERMINAL",0,0,"79940",,terminal_output
+1066,1121089,"TERMINAL",0,0,"86:006:001",,terminal_output
+1067,1122186,"TERMINAL",0,0,"9112",,terminal_output
+1068,1123213,"TERMINAL",0,0,"10223",,terminal_output
+1069,1124240,"TERMINAL",0,0,"1445",,terminal_output
+1070,1125293,"TERMINAL",0,0,"3556",,terminal_output
+1071,1126345,"TERMINAL",0,0,"4667",,terminal_output
+1072,1127385,"TERMINAL",0,0,"5778",,terminal_output
+1073,1128454,"TERMINAL",0,0,"6889",,terminal_output
+1074,1129535,"TERMINAL",0,0,"79950",,terminal_output
+1075,1130581,"TERMINAL",0,0,"810101",,terminal_output
+1076,1131627,"TERMINAL",0,0,"9112",,terminal_output
+1077,1132651,"TERMINAL",0,0,"20223",,terminal_output
+1078,1133707,"TERMINAL",0,0,"1334",,terminal_output
+1079,1134756,"TERMINAL",0,0,"2445",,terminal_output
+1080,1135912,"TERMINAL",0,0,"3556",,terminal_output
+1081,1136855,"TERMINAL",0,0,"4667",,terminal_output
+1082,1137928,"TERMINAL",0,0,"5778",,terminal_output
+1083,1138948,"TERMINAL",0,0,"6889",,terminal_output
+1084,1140025,"TERMINAL",0,0,"7992:00",,terminal_output
+1085,1141047,"TERMINAL",0,0,"820201",,terminal_output
+1086,1142093,"TERMINAL",0,0,"9112",,terminal_output
+1087,1143152,"TERMINAL",0,0,"30223",,terminal_output
+1088,1144187,"TERMINAL",0,0,"1334",,terminal_output
+1089,1145259,"TERMINAL",0,0,"2556",,terminal_output
+1090,1146282,"TERMINAL",0,0,"4667",,terminal_output
+1091,1147324,"TERMINAL",0,0,"5778",,terminal_output
+1092,1148405,"TERMINAL",0,0,"6889",,terminal_output
+1093,1149436,"TERMINAL",0,0,"79910",,terminal_output
+1094,1150553,"TERMINAL",0,0,"830301",,terminal_output
+1095,1151578,"TERMINAL",0,0,"9112",,terminal_output
+1096,1152706,"TERMINAL",0,0,"40223",,terminal_output
+1097,1153607,"TERMINAL",0,0,"1334",,terminal_output
+1098,1154656,"TERMINAL",0,0,"2445",,terminal_output
+1099,1155701,"TERMINAL",0,0,"3556",,terminal_output
+1100,1156753,"TERMINAL",0,0,"4667",,terminal_output
+1101,1157836,"TERMINAL",0,0,"5778",,terminal_output
+1102,1158849,"TERMINAL",0,0,"6889",,terminal_output
+1103,1159934,"TERMINAL",0,0,"79920",,terminal_output
+1104,1161006,"TERMINAL",0,0,"840401",,terminal_output
+1105,1162020,"TERMINAL",0,0,"9112",,terminal_output
+1106,1163148,"TERMINAL",0,0,"50223",,terminal_output
+1107,1164175,"TERMINAL",0,0,"1334",,terminal_output
+1108,1165194,"TERMINAL",0,0,"2445",,terminal_output
+1109,1166224,"TERMINAL",0,0,"3556",,terminal_output
+1110,1167245,"TERMINAL",0,0,"4778",,terminal_output
+1111,1168386,"TERMINAL",0,0,"6889",,terminal_output
+1112,1169398,"TERMINAL",0,0,"79930",,terminal_output
+1113,1170379,"TERMINAL",0,0,"850501",,terminal_output
+1114,1171439,"TERMINAL",0,0,"9112",,terminal_output
+1115,1172480,"TERMINAL",0,0,"20:00223",,terminal_output
+1116,1173550,"TERMINAL",0,0,"1334",,terminal_output
+1117,1174642,"TERMINAL",0,0,"2445",,terminal_output
+1118,1175630,"TERMINAL",0,0,"3556",,terminal_output
+1119,1176673,"TERMINAL",0,0,"4667",,terminal_output
+1120,1177761,"TERMINAL",0,0,"5778",,terminal_output
+1121,1178762,"TERMINAL",0,0,"6889",,terminal_output
+1122,1179840,"TERMINAL",0,0,"79940",,terminal_output
+1123,1181070,"TERMINAL",0,0,"87:007:001",,terminal_output
+1124,1182096,"TERMINAL",0,0,"9112",,terminal_output
+1125,1183120,"TERMINAL",0,0,"10223",,terminal_output
+1126,1184149,"TERMINAL",0,0,"1334",,terminal_output
+1127,1185194,"TERMINAL",0,0,"2445",,terminal_output
+1128,1186246,"TERMINAL",0,0,"3667",,terminal_output
+1129,1187295,"TERMINAL",0,0,"5778",,terminal_output
+1130,1188354,"TERMINAL",0,0,"6889",,terminal_output
+1131,1189394,"TERMINAL",0,0,"79950",,terminal_output
+1132,1190538,"TERMINAL",0,0,"810101",,terminal_output
+1133,1191495,"TERMINAL",0,0,"9112",,terminal_output
+1134,1192643,"TERMINAL",0,0,"20223",,terminal_output
+1135,1193669,"TERMINAL",0,0,"1334",,terminal_output
+1136,1194707,"TERMINAL",0,0,"2445",,terminal_output
+1137,1195719,"TERMINAL",0,0,"3556",,terminal_output
+1138,1196797,"TERMINAL",0,0,"4667",,terminal_output
+1139,1197823,"TERMINAL",0,0,"5778",,terminal_output
+1140,1198873,"TERMINAL",0,0,"6889",,terminal_output
+1141,1199938,"TERMINAL",0,0,"7993:00",,terminal_output

927a8af5474e5654810c00ce2e09fd2de87d3e5722f33fa1090d867db114e403/crowd-code-614215de-0f98-45e4-a49e-9e7d7e57cff91757422920029-2025_09_09-15.02.34.03/source.csv

@@ -0,0 +1,198 @@
+Sequence,Time,File,RangeOffset,RangeLength,Text,Language,Type
+1,5,"train_lam.py",0,0,"from dataclasses import dataclass, field\nimport os\nfrom typing import cast\n\nimport einops\nfrom jax.sharding import Mesh, PartitionSpec, NamedSharding\nfrom jax.experimental.mesh_utils import create_device_mesh\nimport optax\nimport orbax.checkpoint as ocp\nimport numpy as np\nimport dm_pix as pix\nimport jax\nimport jax.numpy as jnp\nimport tyro\nimport wandb\nimport grain\nimport flax.nnx as nnx\n\nfrom models.lam import LatentActionModel\nfrom utils.dataloader import get_dataloader\nfrom utils.lr_utils import get_lr_schedule\nfrom utils.parameter_utils import count_parameters_by_component\n\n\n@dataclass\nclass Args:\n    # Experiment\n    num_steps: int = 200_000\n    seed: int = 0\n    seq_len: int = 16\n    image_channels: int = 3\n    image_height: int = 90\n    image_width: int = 160\n    data_dir: str = """"\n    save_ckpt: bool = False\n    restore_ckpt: bool = False\n    # Optimization\n    batch_size: int = 36\n    vq_beta: float = 0.25\n    init_lr: float = 0.0\n    max_lr: float = 3e-5\n    decay_end: float = 0.0\n    wsd_decay_steps: int = (\n        10000  # NOTE: wsd_decay_steps will only be used when using a wsd-schedule\n    )\n    warmup_steps: int = 5000\n    lr_schedule: str = ""wsd""  # supported options: wsd, cos\n    vq_reset_thresh: int = 50\n    # LAM\n    model_dim: int = 512\n    ffn_dim: int = 2048\n    latent_dim: int = 32\n    num_latents: int = 6\n    patch_size: int = 16\n    num_blocks: int = 4\n    num_heads: int = 8\n    dropout: float = 0.0\n    codebook_dropout: float = 0.0\n    param_dtype = jnp.float32\n    dtype = jnp.bfloat16\n    # Logging\n    log: bool = False\n    entity: str = """"\n    project: str = """"\n    name: str = ""train_lam""\n    tags: list[str] = field(default_factory=lambda: [""lam""])\n    log_interval: int = 5\n    log_image_interval: int = 250\n    ckpt_dir: str = """"\n    log_checkpoint_interval: int = 10000\n    log_checkpoint_keep_period: int = 20000\n    wandb_id: str = """"\n    use_flash_attention: bool = True\n\n\nargs = tyro.cli(Args)\n\n\ndef lam_loss_fn(\n    model: LatentActionModel, inputs: dict\n) -> tuple[jax.Array, tuple[jax.Array, jax.Array, dict]]:\n    # --- Compute loss ---\n    gt = jnp.asarray(inputs[""videos""], dtype=jnp.float32) / 255.0\n    inputs[""videos""] = gt.astype(args.dtype)\n    model.train()\n    outputs = model(inputs, training=True)\n    outputs[""recon""] = outputs[""recon""].astype(jnp.float32)\n    gt_future_frames = gt[:, 1:]\n    mse = jnp.square(gt_future_frames - outputs[""recon""]).mean()\n    q_loss = jnp.square(jax.lax.stop_gradient(outputs[""emb""]) - outputs[""z""]).mean()\n    commitment_loss = jnp.square(\n        outputs[""emb""] - jax.lax.stop_gradient(outputs[""z""])\n    ).mean()\n    loss = mse + q_loss + args.vq_beta * commitment_loss\n\n    # --- Compute validation metrics ---\n    gt = gt_future_frames.clip(0, 1).reshape(-1, *gt_future_frames.shape[2:])\n    recon = outputs[""recon""].clip(0, 1).reshape(-1, *outputs[""recon""].shape[2:])\n    psnr = jnp.asarray(pix.psnr(gt, recon)).mean()\n    ssim = jnp.asarray(pix.ssim(gt, recon)).mean()\n    count_fn = jax.vmap(lambda i: (outputs[""indices""] == i).sum())\n    index_counts = count_fn(jnp.arange(args.num_latents))\n    metrics = dict(\n        loss=loss,\n        mse=mse,\n        q_loss=q_loss,\n        commitment_loss=commitment_loss,\n        psnr=psnr,\n        ssim=ssim,\n        codebook_usage=(index_counts != 0).mean(),\n    )\n    return loss, (outputs[""recon""], index_counts, metrics)\n\n\n@nnx.jit\ndef train_step(\n    lam: LatentActionModel,\n    optimizer: nnx.Optimizer,\n    inputs: dict,\n    action_last_active: jax.Array,\n    rng: jax.Array,\n) -> tuple[jax.Array, jax.Array, jax.Array, dict]:\n    def loss_fn(\n        model: LatentActionModel,\n    ) -> tuple[jax.Array, tuple[jax.Array, jax.Array, dict]]:\n        return lam_loss_fn(model, inputs)\n\n    # --- Update model ---\n    (loss, (recon, idx_counts, metrics)), grads = nnx.value_and_grad(\n        loss_fn, has_aux=True\n    )(lam)\n    optimizer.update(grads)\n\n    # --- Reset inactive latent actions ---\n    codebook = lam.vq.codebook\n    num_codes = len(codebook)\n    active_codes = idx_counts != 0.0\n    action_last_active = jnp.where(active_codes, 0, action_last_active + 1)\n    p_code = active_codes / active_codes.sum()\n    reset_idxs = jax.random.choice(rng, num_codes, shape=(num_codes,), p=p_code)\n    do_reset = action_last_active >= args.vq_reset_thresh\n    new_codebook = jnp.where(\n        jnp.expand_dims(do_reset, -1), codebook[reset_idxs], codebook.value\n    )\n    lam.vq.codebook.value = new_codebook\n    action_last_active = jnp.where(do_reset, 0, action_last_active)\n    return loss, recon, action_last_active, metrics\n\n\nif __name__ == ""__main__"":\n    jax.distributed.initialize()\n    num_devices = jax.device_count()\n    if num_devices == 0:\n        raise ValueError(""No JAX devices found."")\n    print(f""Running on {num_devices} devices."")\n\n    if args.batch_size % num_devices != 0:\n        raise ValueError(\n            f""Global batch size {args.batch_size} must be divisible by ""\n            f""number of devices {num_devices}.""\n        )\n\n    per_device_batch_size_for_init = args.batch_size // num_devices\n\n    rng = jax.random.key(args.seed)\n\n    # --- Initialize model ---\n    rng, _rng = jax.random.split(rng)\n    rngs = nnx.Rngs(_rng)\n    lam = LatentActionModel(\n        in_dim=args.image_channels,\n        model_dim=args.model_dim,\n        ffn_dim=args.ffn_dim,\n        latent_dim=args.latent_dim,\n        num_latents=args.num_latents,\n        patch_size=args.patch_size,\n        num_blocks=args.num_blocks,\n        num_heads=args.num_heads,\n        dropout=args.dropout,\n        codebook_dropout=args.codebook_dropout,\n        param_dtype=args.param_dtype,\n        dtype=args.dtype,\n        use_flash_attention=args.use_flash_attention,\n        rngs=rngs,\n    )\n\n    # Count parameters\n    _, params, _ = nnx.split(lam, nnx.Param, ...)\n    param_counts = count_parameters_by_component(params)\n\n    if args.log and jax.process_index() == 0:\n        wandb_init_kwargs = {\n            ""entity"": args.entity,\n            ""project"": args.project,\n            ""name"": args.name,\n            ""tags"": args.tags,\n            ""group"": ""debug"",\n            ""config"": args,\n        }\n\n        if args.wandb_id:\n            wandb_init_kwargs.update(\n                {\n                    ""id"": args.wandb_id,\n                    ""resume"": ""allow"",\n                }\n            )\n        wandb.init(**wandb_init_kwargs)\n\n        wandb.config.update({""model_param_count"": param_counts})\n\n    print(""Parameter counts:"")\n    print(param_counts)\n\n    # --- Initialize optimizer ---\n    lr_schedule = get_lr_schedule(\n        args.lr_schedule,\n        args.init_lr,\n        args.max_lr,\n        args.decay_end,\n        args.num_steps,\n        args.warmup_steps,\n        args.wsd_decay_steps,\n    )\n    tx = optax.adamw(\n        learning_rate=lr_schedule,\n        b1=0.9,\n        b2=0.9,\n        weight_decay=1e-4,\n        mu_dtype=args.dtype,\n    )\n    optimizer = nnx.Optimizer(lam, tx)\n\n    # FIXME: switch to create_hybrid_device_mesh for runs spanning multiple nodes\n    device_mesh_arr = create_device_mesh((num_devices,))\n    mesh = Mesh(devices=device_mesh_arr, axis_names=(""data"",))\n\n    replicated_sharding = NamedSharding(mesh, PartitionSpec())\n    videos_sharding = NamedSharding(mesh, PartitionSpec(""data"", None, None, None, None))\n\n    model_state = nnx.state(optimizer.model)\n    model_sharded_state = jax.lax.with_sharding_constraint(\n        model_state, replicated_sharding\n    )\n    nnx.update(optimizer.model, model_sharded_state)\n    optimizer_state = nnx.state(optimizer, nnx.optimizer.OptState)\n    optimizer_sharded_state = jax.lax.with_sharding_constraint(\n        optimizer_state, replicated_sharding\n    )\n    nnx.update(optimizer, optimizer_sharded_state)\n\n    # --- Initialize checkpoint manager ---\n    step = 0\n    handler_registry = ocp.handlers.DefaultCheckpointHandlerRegistry()\n    handler_registry.add(\n        ""model_state"", ocp.args.PyTreeSave, ocp.handlers.PyTreeCheckpointHandler\n    )\n    handler_registry.add(\n        ""model_state"", ocp.args.PyTreeRestore, ocp.handlers.PyTreeCheckpointHandler\n    )\n    handler_registry.add(\n        ""dataloader_state"",\n        grain.checkpoint.CheckpointSave,\n        cast(ocp.handlers.CheckpointHandler, grain.checkpoint.CheckpointHandler),\n    )\n    handler_registry.add(\n        ""dataloader_state"",\n        grain.checkpoint.CheckpointRestore,\n        cast(ocp.handlers.CheckpointHandler, grain.checkpoint.CheckpointHandler),\n    )\n\n    checkpoint_options = ocp.CheckpointManagerOptions(\n        save_interval_steps=args.log_checkpoint_interval,\n        max_to_keep=3,\n        keep_period=args.log_checkpoint_keep_period,\n        step_format_fixed_length=6,\n        cleanup_tmp_directories=True,\n    )\n\n    checkpoint_manager = ocp.CheckpointManager(\n        args.ckpt_dir,\n        options=checkpoint_options,\n        handler_registry=handler_registry,\n    )\n\n    # --- Create DataLoaderIterator from dataloader ---\n    image_shape = (args.image_height, args.image_width, args.image_channels)\n    array_record_files = [\n        os.path.join(args.data_dir, x)\n        for x in os.listdir(args.data_dir)\n        if x.endswith("".array_record"")\n    ]\n    grain_dataloader = get_dataloader(\n        array_record_files,\n        args.seq_len,\n        # NOTE: We deliberately pass the global batch size\n        # The dataloader shards the dataset across all processes\n        args.batch_size,\n        *image_shape,\n        num_workers=8,\n        prefetch_buffer_size=1,\n        seed=args.seed,\n    )\n    initial_state = grain_dataloader._create_initial_state()\n    grain_iterator = grain.DataLoaderIterator(grain_dataloader, initial_state)\n\n    # --- Restore checkpoint ---\n    if args.restore_ckpt:\n        abstract_optimizer = nnx.eval_shape(lambda: optimizer)\n        abstract_optimizer_state = nnx.state(abstract_optimizer)\n        restored = checkpoint_manager.restore(\n            checkpoint_manager.latest_step(),\n            args=ocp.args.Composite(\n                model_state=ocp.args.PyTreeRestore(abstract_optimizer_state),  # type: ignore\n                dataloader_state=grain.checkpoint.CheckpointRestore(grain_iterator),  # type: ignore\n            ),\n        )\n        restored_optimizer_state = restored[""model_state""]\n        nnx.update(optimizer, restored_optimizer_state)\n        grain_iterator = restored[""dataloader_state""]\n        step = checkpoint_manager.latest_step() or 0\n        print(f""Restored dataloader and model state from step {step}"")\n\n    # --- TRAIN LOOP ---\n    dataloader = (\n        jax.make_array_from_process_local_data(videos_sharding, elem)\n        for elem in grain_iterator\n    )\n    print(f""Starting training from step {step}..."")\n    action_last_active = jnp.zeros(args.num_latents, dtype=jnp.int32)\n    while step < args.num_steps:\n        for videos in dataloader:\n            # --- Train step ---\n            rng, _rng = jax.random.split(rng)\n\n            inputs = dict(videos=videos, rng=_rng)\n            rng, _rng = jax.random.split(rng)\n            loss, recon, action_last_active, metrics = train_step(\n                lam, optimizer, inputs, action_last_active, _rng\n            )\n            metrics[""lr""] = lr_schedule(step)\n            print(f""Step {step}, loss: {loss}"")\n            step += 1\n\n            # --- Logging ---\n            if args.log:\n                if step % args.log_interval == 0 and jax.process_index() == 0:\n                    wandb.log(\n                        {\n                            ""loss"": loss,\n                            ""step"": step,\n                            **metrics,\n                        }\n                    )\n                if step % args.log_image_interval == 0:\n                    gt_seq = inputs[""videos""][0, 1:].astype(jnp.float32) / 255.0\n                    recon_seq = recon[0].clip(0, 1)\n                    comparison_seq = jnp.concatenate((gt_seq, recon_seq), axis=1)\n                    comparison_seq = einops.rearrange(\n                        comparison_seq * 255, ""t h w c -> h (t w) c""\n                    )\n                    # NOTE: Process-dependent control flow deliberately happens\n                    # after indexing operation since it must not contain code\n                    # sections that lead to cross-accelerator communication.\n                    if jax.process_index() == 0:\n                        log_images = dict(\n                            image=wandb.Image(np.asarray(gt_seq[0])),\n                            recon=wandb.Image(np.asarray(recon_seq[0])),\n                            true_vs_recon=wandb.Image(\n                                np.asarray(comparison_seq.astype(np.uint8))\n                            ),\n                        )\n                        wandb.log(log_images)\n            # --- Checkpointing ---\n            if args.save_ckpt and step % args.log_checkpoint_interval == 0:\n                optimizer_state = nnx.state(optimizer)\n                checkpoint_manager.save(\n                    step,\n                    args=ocp.args.Composite(\n                        model_state=ocp.args.PyTreeSave(optimizer_state),  # type: ignore\n                        dataloader_state=grain.checkpoint.CheckpointSave(  # type: ignore\n                            grain_iterator  # type: ignore\n                        ),\n                    ),\n                )\n                print(f""Saved checkpoint at step {step}"")\n            if step >= args.num_steps:\n                break\n\n    checkpoint_manager.close()\n",python,tab
+2,832,"extension-output-pdoom-org.crowd-code-#1-crowd-code",0,0,"",Log,tab
+3,1155,"extension-output-pdoom-org.crowd-code-#1-crowd-code",0,0,"3:02:33 PM [info] Activating crowd-code\n3:02:34 PM [info] Recording started\n3:02:34 PM [info] Initializing git provider using file system watchers...\n3:02:34 PM [info] Git repository found\n3:02:34 PM [info] Git provider initialized successfully\n3:02:34 PM [info] Initial git state: [object Object]\n",Log,content
+4,3766,"train_lam.py",0,0,"",python,tab
+5,3771,"TERMINAL",0,0,"",,terminal_focus
+6,14658,"TERMINAL",0,0,"source /home/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/bin/activate",,terminal_command
+7,136281,"TERMINAL",0,0,"bash",,terminal_focus
+8,138228,"TERMINAL",0,0,"idling",,terminal_command
+9,138301,"TERMINAL",0,0,"]633;C[?1049h(B[?7hEvery 1.0s: sinfo_t_idlehkn1990.localdomain: Tue Sep  9 15:04:52 2025Partition dev_cpuonly:     10 nodes idle\rPartition cpuonly:     26 nodes idle\rPartition dev_accelerated:\t 0 nodes idle\rPartition accelerated:     29 nodes idle\rPartition dev_accelerated-h100    :\t 0 nodes idle\rPartition accelerated-h100:\t 0 nodes idle\rPartition large:\t 0 nodes idle\rPartition accelerated-h200:\t 0 nodes idle",,terminal_output
+10,139401,"TERMINAL",0,0,"3\t",,terminal_output
+11,140419,"TERMINAL",0,0,"4\t",,terminal_output
+12,140811,"TERMINAL",0,0,"[?1049l\r[?1l>]0;tum_cte0515@hkn1990:~/Projects/jasmine",,terminal_output
+13,146318,"TERMINAL",0,0,"_node",,terminal_command
+14,146350,"TERMINAL",0,0,"]633;C]0;tum_cte0515@hkn1990:~/Projects/jasmine",,terminal_output
+15,150256,"TERMINAL",0,0,"salloc --time=05:00:00 --partition=accelerated --nodes=1 --gres=gpu:1  --cpus-per-task=8",,terminal_command
+16,150320,"TERMINAL",0,0,"]633;Csalloc: Granted job allocation 3480467\r\n",,terminal_output
+17,150455,"TERMINAL",0,0,"salloc: Waiting for resource configuration\r\n",,terminal_output
+18,177507,"TERMINAL",0,0,"salloc: Nodes hkn0818 are ready for job\r\n",,terminal_output
+19,178680,"TERMINAL",0,0,"]0;tum_cte0515@hkn0818:~/Projects/jasmine[?2004h[tum_cte0515@hkn0818 jasmine]$ ",,terminal_output
+20,184841,"TERMINAL",0,0,"s",,terminal_output
+21,185268,"TERMINAL",0,0,"",,terminal_output
+22,185837,"TERMINAL",0,0,"cd",,terminal_output
+23,186104,"TERMINAL",0,0,".",,terminal_output
+24,186235,"TERMINAL",0,0,".",,terminal_output
+25,186821,"TERMINAL",0,0,"",,terminal_output
+26,187026,"TERMINAL",0,0,"",,terminal_output
+27,187154,"TERMINAL",0,0," ",,terminal_output
+28,187287,"TERMINAL",0,0,".",,terminal_output
+29,187418,"TERMINAL",0,0,".",,terminal_output
+30,187533,"TERMINAL",0,0,"\r\n[?2004l\r]0;tum_cte0515@hkn0818:~/Projects[?2004h[tum_cte0515@hkn0818 Projects]$ ",,terminal_output
+31,187833,"TERMINAL",0,0,"c",,terminal_output
+32,187906,"TERMINAL",0,0,"d",,terminal_output
+33,188023,"TERMINAL",0,0," ",,terminal_output
+34,188446,"TERMINAL",0,0,"j",,terminal_output
+35,188581,"TERMINAL",0,0,"a",,terminal_output
+36,189081,"TERMINAL",0,0,"",,terminal_output
+37,189470,"TERMINAL",0,0,"f",,terminal_output
+38,189596,"TERMINAL",0,0,"ar/",,terminal_output
+39,190052,"TERMINAL",0,0,"\r\n[?2004l\r]0;tum_cte0515@hkn0818:~/Projects/jafar[?2004h[tum_cte0515@hkn0818 jafar]$ ",,terminal_output
+40,192215,"TERMINAL",0,0,"s",,terminal_output
+41,192295,"TERMINAL",0,0,"o",,terminal_output
+42,192360,"TERMINAL",0,0,"u",,terminal_output
+43,192442,"TERMINAL",0,0,"r",,terminal_output
+44,192712,"TERMINAL",0,0,"c",,terminal_output
+45,192823,"TERMINAL",0,0,"e ",,terminal_output
+46,193263,"TERMINAL",0,0,".",,terminal_output
+47,193438,"TERMINAL",0,0,"v",,terminal_output
+48,193680,"TERMINAL",0,0,"env/",,terminal_output
+49,193926,"TERMINAL",0,0,"b",,terminal_output
+50,193991,"TERMINAL",0,0,"in/",,terminal_output
+51,194269,"TERMINAL",0,0,"ac",,terminal_output
+52,194456,"TERMINAL",0,0,"tivate",,terminal_output
+53,194719,"TERMINAL",0,0,"\r\n[?2004l\r]0;tum_cte0515@hkn0818:~/Projects/jafar[?2004h(jafar) [tum_cte0515@hkn0818 jafar]$ ",,terminal_output
+54,199639,"TERMINAL",0,0,"c",,terminal_output
+55,200167,"TERMINAL",0,0,"",,terminal_output
+56,200333,"TERMINAL",0,0,"c",,terminal_output
+57,200457,"TERMINAL",0,0,"u",,terminal_output
+58,200651,"TERMINAL",0,0,"r",,terminal_output
+59,200855,"TERMINAL",0,0,"s",,terminal_output
+60,200929,"TERMINAL",0,0,"o",,terminal_output
+61,201054,"TERMINAL",0,0,"r",,terminal_output
+62,201137,"TERMINAL",0,0," ",,terminal_output
+63,201325,"TERMINAL",0,0,".",,terminal_output
+64,203292,"TERMINAL",0,0,"",,terminal_output
+65,203449,"TERMINAL",0,0,"",,terminal_output
+66,204518,"TERMINAL",0,0,"bash",,terminal_focus
+67,209503,"TERMINAL",0,0,"cursor ../jafar/",,terminal_command
+68,209549,"TERMINAL",0,0,"]633;C",,terminal_output
+69,209647,"TERMINAL",0,0,"]0;tum_cte0515@hkn1990:~/Projects/jasmine",,terminal_output
+70,296145,"TERMINAL",0,0,"srun",,terminal_focus
+71,297158,"TERMINAL",0,0,"sh",,terminal_output
+72,297269,"TERMINAL",0,0," ",,terminal_output
+73,297457,"TERMINAL",0,0,"s",,terminal_output
+74,297642,"TERMINAL",0,0,"a",,terminal_output
+75,297756,"TERMINAL",0,0,"m",,terminal_output
+76,298119,"TERMINAL",0,0,"",,terminal_output
+77,298266,"TERMINAL",0,0,"",,terminal_output
+78,298379,"TERMINAL",0,0,"",,terminal_output
+79,298493,"TERMINAL",0,0,"",,terminal_output
+80,298649,"TERMINAL",0,0,"",,terminal_output
+81,298774,"TERMINAL",0,0,"",,terminal_output
+82,298990,"TERMINAL",0,0,"p",,terminal_output
+83,299131,"TERMINAL",0,0,"y",,terminal_output
+84,299351,"TERMINAL",0,0,"t",,terminal_output
+85,299403,"TERMINAL",0,0,"h",,terminal_output
+86,299606,"TERMINAL",0,0,"o",,terminal_output
+87,299785,"TERMINAL",0,0,"n",,terminal_output
+88,299839,"TERMINAL",0,0," ",,terminal_output
+89,300340,"TERMINAL",0,0,"",,terminal_output
+90,300490,"TERMINAL",0,0,"",,terminal_output
+91,300640,"TERMINAL",0,0,"",,terminal_output
+92,300793,"TERMINAL",0,0,"",,terminal_output
+93,300915,"TERMINAL",0,0,"",,terminal_output
+94,301091,"TERMINAL",0,0,"",,terminal_output
+95,301270,"TERMINAL",0,0,"",,terminal_output
+96,304897,"slurm/jobs/mihir/horeka/preprocessing/coinrun_chunked.sh",0,0,"#!/usr/bin/env bash\n\n#SBATCH --nodes=1\n#SBATCH --ntasks-per-node=1\n#SBATCH --time=10:00:00\n#SBATCH --partition=large\n#SBATCH --cpus-per-task=16\n#SBATCH --output=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/logs/logs_mihir/preprocess/coinrun/%x_%j.log\n#SBATCH --error=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/logs/logs_mihir/preprocess/coinrun/%x_%j.log\n#SBATCH --job-name=preprocess_coinrun_chunked\n\nsource .venv/bin/activate\n\npython generate_dataset.py \\n    --min_episode_length 1000 \\n    --output_dir /hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/data_coinrun/coinrun_episodes_test \\n    --num_episodes 50",shellscript,tab
+97,315511,"slurm/jobs/mihir/horeka/maskgit_big_runs/train_dynamics_8_nodes.sbatch",0,0,"#!/usr/bin/env bash\n\n#SBATCH --nodes=8\n#SBATCH --ntasks-per-node=4\n#SBATCH --time=48:00:00\n#SBATCH --partition=accelerated\n#SBATCH --cpus-per-task=5\n#SBATCH --gres=gpu:4\n#SBATCH --output=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/logs/logs_mihir/maskgit/dynamics-cotraining/%x_%j.log\n#SBATCH --error=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/logs/logs_mihir/maskgit/dynamics-cotraining/%x_%j.log\n#SBATCH --job-name=train_dynamics_maskgit_8_node\n#SBATCH --requeue\n#SBATCH --signal=b:usr1@300 # 5 min before timeout\n\n# --- signal trap to requeue job before timeout ---\nrequeue_job() {\n    echo ""[$(date)] caught sigusr1 (timeout warning), requeueing slurm job $SLURM_JOB_ID...""\n    # optional: trigger checkpoint saving here\n    # e.g., touch $checkpoint_dir/requeue_trigger\n    scontrol requeue $SLURM_JOB_ID\n    exit 0\n}\n\ntrap requeue_job sigusr1\n\n# set checkpoint flag based on restart count\nrestart_count=$(scontrol show job $SLURM_JOB_ID | grep -o 'Restarts=[0-9]*' | cut -d'=' -f2)\n\nif [ $restart_count -eq 0 ]; then\n    restore_ckpt_flag=""--no-restore-ckpt""\nelse\n    restore_ckpt_flag=""--restore-ckpt""\nfi\n\n\n\n# Log the sbatch script\ncat $0\n\nmodule unload mpi/openmpi/5.0\nmodule unload devel/cuda/12.4\nsource .venv/bin/activate\n\narray_records_dir=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/data_new/open_ai_minecraft_arrayrecords_chunked\n\njob_name=$SLURM_JOB_NAME\nslurm_job_id=$SLURM_JOB_ID\n\nCHECKPOINT_DIR=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/checkpoints/mihir/holiday/maskgit/$job_name/$slurm_job_id\nmkdir -p $CHECKPOINT_DIR\n\ntokenizer_ckpt_dir=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/checkpoints/big-runs/tokenizer/train_tokenizer_1e-4/3412401\n\nenv | grep SLURM\n\nsrun python train_dynamics.py \\n    --save_ckpt \\n    $restore_ckpt_flag \\n    --wandb_id $SLURM_JOB_ID \\n    --ckpt_dir $CHECKPOINT_DIR \\n    --batch_size=160 \\n    --init_lr=0 \\n    --dyna_type=maskgit \\n    --num_latent_actions=100 \\n    --max_lr=8e-5 \\n    --log_image_interval=1000 \\n    --log \\n    --log_checkpoint_interval=1000 \\n    --name=dynamics-maskgit-8-node-$slurm_job_id \\n    --tags dynamics maskgit 8-node post-launch-main \\n    --entity instant-uv \\n    --project jafar \\n    --dyna_dim=1024 \\n    --dyna_num_blocks=16 \\n    --dyna_num_heads=16 \\n    --dyna_ffn_dim=4096 \\n    --tokenizer_checkpoint=$tokenizer_ckpt_dir \\n    --data_dir $array_records_dir &\n\nchild_pid=$!\n\nwait $child_pid\n",shellscript,tab
+98,317803,"slurm/jobs/mihir/horeka/maskgit_big_runs/train_dynamics_8_nodes.sbatch",1257,0,"",shellscript,selection_mouse
+99,317823,"slurm/jobs/mihir/horeka/maskgit_big_runs/train_dynamics_8_nodes.sbatch",1256,0,"",shellscript,selection_command
+100,318630,"slurm/jobs/mihir/horeka/maskgit_big_runs/train_dynamics_8_nodes.sbatch",1231,0,"",shellscript,selection_mouse
+101,318645,"slurm/jobs/mihir/horeka/maskgit_big_runs/train_dynamics_8_nodes.sbatch",1230,0,"",shellscript,selection_command
+102,318788,"slurm/jobs/mihir/horeka/maskgit_big_runs/train_dynamics_8_nodes.sbatch",1230,1,"4",shellscript,selection_mouse
+103,318793,"slurm/jobs/mihir/horeka/maskgit_big_runs/train_dynamics_8_nodes.sbatch",1231,0,"",shellscript,selection_command
+104,318985,"slurm/jobs/mihir/horeka/maskgit_big_runs/train_dynamics_8_nodes.sbatch",1229,2,".4",shellscript,selection_mouse
+105,318986,"slurm/jobs/mihir/horeka/maskgit_big_runs/train_dynamics_8_nodes.sbatch",1224,7,"da/12.4",shellscript,selection_mouse
+106,318987,"slurm/jobs/mihir/horeka/maskgit_big_runs/train_dynamics_8_nodes.sbatch",1215,16," devel/cuda/12.4",shellscript,selection_mouse
+107,318987,"slurm/jobs/mihir/horeka/maskgit_big_runs/train_dynamics_8_nodes.sbatch",1212,19,"oad devel/cuda/12.4",shellscript,selection_mouse
+108,318988,"slurm/jobs/mihir/horeka/maskgit_big_runs/train_dynamics_8_nodes.sbatch",1210,21,"nload devel/cuda/12.4",shellscript,selection_mouse
+109,318988,"slurm/jobs/mihir/horeka/maskgit_big_runs/train_dynamics_8_nodes.sbatch",1207,24,"e unload devel/cuda/12.4",shellscript,selection_mouse
+110,318989,"slurm/jobs/mihir/horeka/maskgit_big_runs/train_dynamics_8_nodes.sbatch",1205,26,"ule unload devel/cuda/12.4",shellscript,selection_mouse
+111,318990,"slurm/jobs/mihir/horeka/maskgit_big_runs/train_dynamics_8_nodes.sbatch",1173,58,"odule unload mpi/openmpi/5.0\nmodule unload devel/cuda/12.4",shellscript,selection_mouse
+112,318990,"slurm/jobs/mihir/horeka/maskgit_big_runs/train_dynamics_8_nodes.sbatch",1172,59,"module unload mpi/openmpi/5.0\nmodule unload devel/cuda/12.4",shellscript,selection_mouse
+113,323244,"TERMINAL",0,0,"module unload mpi/openmpi/5.0\r\n\rmodule unload devel/cuda/12.4",,terminal_output
+114,323792,"TERMINAL",0,0,"module unload mpi/openmpi/5.0\r\n\rmodule unload devel/cuda/12.4\r\n[?2004l\r",,terminal_output
+115,323980,"TERMINAL",0,0,"]0;tum_cte0515@hkn0818:~/Projects/jafar[?2004h(jafar) [tum_cte0515@hkn0818 jafar]$ ",,terminal_output
+116,325638,"TERMINAL",0,0,"p",,terminal_output
+117,325835,"TERMINAL",0,0,"y",,terminal_output
+118,326049,"TERMINAL",0,0,"th",,terminal_output
+119,326267,"TERMINAL",0,0,"o",,terminal_output
+120,326445,"TERMINAL",0,0,"n ",,terminal_output
+121,326587,"TERMINAL",0,0,"s",,terminal_output
+122,326984,"TERMINAL",0,0,"",,terminal_output
+123,327730,"TERMINAL",0,0,"a",,terminal_output
+124,327897,"TERMINAL",0,0,"m",,terminal_output
+125,327962,"TERMINAL",0,0,"ple.py ",,terminal_output
+126,479201,"TERMINAL",0,0,"",,terminal_output
+127,479313,"TERMINAL",0,0,"",,terminal_output
+128,479457,"TERMINAL",0,0,"",,terminal_output
+129,479719,"TERMINAL",0,0,"s",,terminal_output
+130,479798,"TERMINAL",0,0,"h",,terminal_output
+131,479946,"TERMINAL",0,0," ",,terminal_output
+132,480208,"TERMINAL",0,0,"slurm/jobs/mihir/horeka/jafar_og_reproduction/og_coinrun_dynamics_reproduction-sample.sh",,terminal_output
+133,480569,"TERMINAL",0,0,"\rslurm/jobs/mihir/horeka/jafar_og_reproduction/og_coinrun_dynamics_reproduction-sample.sh\r\n[?2004l\r\r\nNote: the module ""mpi/openmpi/5.0"" cannot be unloaded because it was not loaded.\r\n\r\n",,terminal_output
+134,480642,"TERMINAL",0,0,"\r\nNote: the module ""devel/cuda/12.4"" cannot be unloaded because it was not loaded.\r\n\r\n",,terminal_output
+135,495073,"TERMINAL",0,0,"╭─ Parsing error ────────────────────────────╮\r\n│ Argument --checkpoint: expected 1 argument │\r\n│ ────────────────────────────────────────── │\r\n│ For full helptext, run sample.py --help    │\r\n╰────────────────────────────────────────────╯\r\n",,terminal_output
+136,495881,"TERMINAL",0,0,"]0;tum_cte0515@hkn0818:~/Projects/jafar[?2004h(jafar) [tum_cte0515@hkn0818 jafar]$ ",,terminal_output
+137,515622,"TERMINAL",0,0,"bash",,terminal_focus
+138,519030,"TERMINAL",0,0,"cd $ws_dir",,terminal_command
+139,519130,"TERMINAL",0,0,"]633;C]0;tum_cte0515@hkn1990:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared",,terminal_output
+140,519620,"TERMINAL",0,0,"ls",,terminal_command
+141,519672,"TERMINAL",0,0,"]633;C",,terminal_output
+142,519736,"TERMINAL",0,0,"checkpoints  count_items.sh  data  data_atari  data_coinrun  data_minecraft  data_new  huggingface  logs  possibly_corrupt_files_in_this_workspace.txt  scripts\r\n]0;tum_cte0515@hkn1990:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared",,terminal_output
+143,521863,"TERMINAL",0,0,"cd checkpoints/",,terminal_command
+144,590615,"TERMINAL",0,0,"srun",,terminal_focus
+145,591123,"TERMINAL",0,0,"sh slurm/jobs/mihir/horeka/jafar_og_reproduction/og_coinrun_dynamics_reproduction-sample.sh",,terminal_output
+146,591321,"TERMINAL",0,0,"\r\n[?2004l\r\r\nNote: the module ""mpi/openmpi/5.0"" cannot be unloaded because it was not loaded.\r\n\r\n\r\nNote: the module ""devel/cuda/12.4"" cannot be unloaded because it was not loaded.\r\n\r\n",,terminal_output
+147,605029,"TERMINAL",0,0,"2025-09-09 15:12:38.956720: W external/xla/xla/service/gpu/autotuning/dot_search_space.cc:200] All configs were filtered out because none of them sufficiently match the hints. Maybe the hints set does not contain a good representative set of valid configs?Working around this by using the full hints set instead.\r\n",,terminal_output
+148,606875,"TERMINAL",0,0,"2025-09-09 15:12:40.801380: W external/xla/xla/service/gpu/autotuning/dot_search_space.cc:200] All configs were filtered out because none of them sufficiently match the hints. Maybe the hints set does not contain a good representative set of valid configs?Working around this by using the full hints set instead.\r\n",,terminal_output
+149,613803,"TERMINAL",0,0,"2025-09-09 15:12:47.615654: W external/xla/xla/service/gpu/autotuning/dot_search_space.cc:200] All configs were filtered out because none of them sufficiently match the hints. Maybe the hints set does not contain a good representative set of valid configs?Working around this by using the full hints set instead.\r\n",,terminal_output
+150,620537,"TERMINAL",0,0,"2025-09-09 15:12:54.464597: W external/xla/xla/service/gpu/autotuning/dot_search_space.cc:200] All configs were filtered out because none of them sufficiently match the hints. Maybe the hints set does not contain a good representative set of valid configs?Working around this by using the full hints set instead.\r\n",,terminal_output
+151,623582,"TERMINAL",0,0,"/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jafar/.venv/lib/python3.10/site-packages/orbax/checkpoint/_src/serialization/type_handlers.py:1269: UserWarning: Sharding info not provided when restoring. Populating sharding info from sharding file. Please note restoration time will be slightly increased due to reading from file. Note also that this option is unsafe when restoring on a different topology than the checkpoint was saved with.\r\n  warnings.warn(\r\n",,terminal_output
+152,632767,"TERMINAL",0,0,"Frame 1\r\n",,terminal_output
+153,636609,"TERMINAL",0,0,"2025-09-09 15:13:10.533420: W external/xla/xla/service/gpu/autotuning/dot_search_space.cc:200] All configs were filtered out because none of them sufficiently match the hints. Maybe the hints set does not contain a good representative set of valid configs?Working around this by using the full hints set instead.\r\n",,terminal_output
+154,646296,"TERMINAL",0,0,"Frame 2\r\n",,terminal_output
+155,650947,"TERMINAL",0,0,"2025-09-09 15:13:24.859501: W external/xla/xla/service/gpu/autotuning/dot_search_space.cc:200] All configs were filtered out because none of them sufficiently match the hints. Maybe the hints set does not contain a good representative set of valid configs?Working around this by using the full hints set instead.\r\n",,terminal_output
+156,652638,"TERMINAL",0,0,"2025-09-09 15:13:26.514152: W external/xla/xla/service/gpu/autotuning/dot_search_space.cc:200] All configs were filtered out because none of them sufficiently match the hints. Maybe the hints set does not contain a good representative set of valid configs?Working around this by using the full hints set instead.\r\n",,terminal_output
+157,660390,"TERMINAL",0,0,"Frame 3\r\n",,terminal_output
+158,665267,"TERMINAL",0,0,"2025-09-09 15:13:39.121931: W external/xla/xla/service/gpu/autotuning/dot_search_space.cc:200] All configs were filtered out because none of them sufficiently match the hints. Maybe the hints set does not contain a good representative set of valid configs?Working around this by using the full hints set instead.\r\n",,terminal_output
+159,666778,"TERMINAL",0,0,"2025-09-09 15:13:40.677335: W external/xla/xla/service/gpu/autotuning/dot_search_space.cc:200] All configs were filtered out because none of them sufficiently match the hints. Maybe the hints set does not contain a good representative set of valid configs?Working around this by using the full hints set instead.\r\n",,terminal_output
+160,674375,"TERMINAL",0,0,"Frame 4\r\n",,terminal_output
+161,679274,"TERMINAL",0,0,"2025-09-09 15:13:53.202302: W external/xla/xla/service/gpu/autotuning/dot_search_space.cc:200] All configs were filtered out because none of them sufficiently match the hints. Maybe the hints set does not contain a good representative set of valid configs?Working around this by using the full hints set instead.\r\n",,terminal_output
+162,680926,"TERMINAL",0,0,"2025-09-09 15:13:54.848446: W external/xla/xla/service/gpu/autotuning/dot_search_space.cc:200] All configs were filtered out because none of them sufficiently match the hints. Maybe the hints set does not contain a good representative set of valid configs?Working around this by using the full hints set instead.\r\n",,terminal_output
+163,688985,"TERMINAL",0,0,"Frame 5\r\n",,terminal_output
+164,694378,"TERMINAL",0,0,"2025-09-09 15:14:08.272770: W external/xla/xla/service/gpu/autotuning/dot_search_space.cc:200] All configs were filtered out because none of them sufficiently match the hints. Maybe the hints set does not contain a good representative set of valid configs?Working around this by using the full hints set instead.\r\n",,terminal_output
+165,695951,"TERMINAL",0,0,"2025-09-09 15:14:09.862514: W external/xla/xla/service/gpu/autotuning/dot_search_space.cc:200] All configs were filtered out because none of them sufficiently match the hints. Maybe the hints set does not contain a good representative set of valid configs?Working around this by using the full hints set instead.\r\n",,terminal_output
+166,703627,"TERMINAL",0,0,"Frame 6\r\n",,terminal_output
+167,709262,"TERMINAL",0,0,"2025-09-09 15:14:23.141537: W external/xla/xla/service/gpu/autotuning/dot_search_space.cc:200] All configs were filtered out because none of them sufficiently match the hints. Maybe the hints set does not contain a good representative set of valid configs?Working around this by using the full hints set instead.\r\n",,terminal_output
+168,710977,"TERMINAL",0,0,"2025-09-09 15:14:24.889095: W external/xla/xla/service/gpu/autotuning/dot_search_space.cc:200] All configs were filtered out because none of them sufficiently match the hints. Maybe the hints set does not contain a good representative set of valid configs?Working around this by using the full hints set instead.\r\n",,terminal_output
+169,718590,"TERMINAL",0,0,"Frame 7\r\n",,terminal_output
+170,723917,"TERMINAL",0,0,"2025-09-09 15:14:37.836699: W external/xla/xla/service/gpu/autotuning/dot_search_space.cc:200] All configs were filtered out because none of them sufficiently match the hints. Maybe the hints set does not contain a good representative set of valid configs?Working around this by using the full hints set instead.\r\n",,terminal_output
+171,725556,"TERMINAL",0,0,"2025-09-09 15:14:39.463510: W external/xla/xla/service/gpu/autotuning/dot_search_space.cc:200] All configs were filtered out because none of them sufficiently match the hints. Maybe the hints set does not contain a good representative set of valid configs?Working around this by using the full hints set instead.\r\n",,terminal_output
+172,733384,"TERMINAL",0,0,"Frame 8\r\n",,terminal_output
+173,738800,"TERMINAL",0,0,"2025-09-09 15:14:52.727758: W external/xla/xla/service/gpu/autotuning/dot_search_space.cc:200] All configs were filtered out because none of them sufficiently match the hints. Maybe the hints set does not contain a good representative set of valid configs?Working around this by using the full hints set instead.\r\n",,terminal_output
+174,740402,"TERMINAL",0,0,"2025-09-09 15:14:54.299840: W external/xla/xla/service/gpu/autotuning/dot_search_space.cc:200] All configs were filtered out because none of them sufficiently match the hints. Maybe the hints set does not contain a good representative set of valid configs?Working around this by using the full hints set instead.\r\n",,terminal_output
+175,749092,"TERMINAL",0,0,"Frame 9\r\n",,terminal_output
+176,754734,"TERMINAL",0,0,"2025-09-09 15:15:08.632875: W external/xla/xla/service/gpu/autotuning/dot_search_space.cc:200] All configs were filtered out because none of them sufficiently match the hints. Maybe the hints set does not contain a good representative set of valid configs?Working around this by using the full hints set instead.\r\n",,terminal_output
+177,756531,"TERMINAL",0,0,"2025-09-09 15:15:10.383487: W external/xla/xla/service/gpu/autotuning/dot_search_space.cc:200] All configs were filtered out because none of them sufficiently match the hints. Maybe the hints set does not contain a good representative set of valid configs?Working around this by using the full hints set instead.\r\n",,terminal_output
+178,764967,"TERMINAL",0,0,"Frame 10\r\n",,terminal_output
+179,770735,"TERMINAL",0,0,"2025-09-09 15:15:24.664649: W external/xla/xla/service/gpu/autotuning/dot_search_space.cc:200] All configs were filtered out because none of them sufficiently match the hints. Maybe the hints set does not contain a good representative set of valid configs?Working around this by using the full hints set instead.\r\n",,terminal_output
+180,772393,"TERMINAL",0,0,"2025-09-09 15:15:26.310567: W external/xla/xla/service/gpu/autotuning/dot_search_space.cc:200] All configs were filtered out because none of them sufficiently match the hints. Maybe the hints set does not contain a good representative set of valid configs?Working around this by using the full hints set instead.\r\n",,terminal_output
+181,781170,"TERMINAL",0,0,"Frame 11\r\n",,terminal_output
+182,787078,"TERMINAL",0,0,"2025-09-09 15:15:40.965308: W external/xla/xla/service/gpu/autotuning/dot_search_space.cc:200] All configs were filtered out because none of them sufficiently match the hints. Maybe the hints set does not contain a good representative set of valid configs?Working around this by using the full hints set instead.\r\n",,terminal_output
+183,788711,"TERMINAL",0,0,"2025-09-09 15:15:42.606920: W external/xla/xla/service/gpu/autotuning/dot_search_space.cc:200] All configs were filtered out because none of them sufficiently match the hints. Maybe the hints set does not contain a good representative set of valid configs?Working around this by using the full hints set instead.\r\n",,terminal_output
+184,797254,"TERMINAL",0,0,"Frame 12\r\n",,terminal_output
+185,803163,"TERMINAL",0,0,"2025-09-09 15:15:56.991416: W external/xla/xla/service/gpu/autotuning/dot_search_space.cc:200] All configs were filtered out because none of them sufficiently match the hints. Maybe the hints set does not contain a good representative set of valid configs?Working around this by using the full hints set instead.\r\n",,terminal_output
+186,804697,"TERMINAL",0,0,"2025-09-09 15:15:58.566806: W external/xla/xla/service/gpu/autotuning/dot_search_space.cc:200] All configs were filtered out because none of them sufficiently match the hints. Maybe the hints set does not contain a good representative set of valid configs?Working around this by using the full hints set instead.\r\n",,terminal_output
+187,813965,"TERMINAL",0,0,"Frame 13\r\n",,terminal_output
+188,819878,"TERMINAL",0,0,"2025-09-09 15:16:13.806078: W external/xla/xla/service/gpu/autotuning/dot_search_space.cc:200] All configs were filtered out because none of them sufficiently match the hints. Maybe the hints set does not contain a good representative set of valid configs?Working around this by using the full hints set instead.\r\n",,terminal_output
+189,821645,"TERMINAL",0,0,"2025-09-09 15:16:15.573480: W external/xla/xla/service/gpu/autotuning/dot_search_space.cc:200] All configs were filtered out because none of them sufficiently match the hints. Maybe the hints set does not contain a good representative set of valid configs?Working around this by using the full hints set instead.\r\n",,terminal_output
+190,830300,"TERMINAL",0,0,"Frame 14\r\n",,terminal_output
+191,835874,"TERMINAL",0,0,"2025-09-09 15:16:29.796506: W external/xla/xla/service/gpu/autotuning/dot_search_space.cc:200] All configs were filtered out because none of them sufficiently match the hints. Maybe the hints set does not contain a good representative set of valid configs?Working around this by using the full hints set instead.\r\n",,terminal_output
+192,837625,"TERMINAL",0,0,"2025-09-09 15:16:31.539961: W external/xla/xla/service/gpu/autotuning/dot_search_space.cc:200] All configs were filtered out because none of them sufficiently match the hints. Maybe the hints set does not contain a good representative set of valid configs?Working around this by using the full hints set instead.\r\n",,terminal_output
+193,846471,"TERMINAL",0,0,"Frame 15\r\n",,terminal_output
+194,852495,"TERMINAL",0,0,"2025-09-09 15:16:46.418156: W external/xla/xla/service/gpu/autotuning/dot_search_space.cc:200] All configs were filtered out because none of them sufficiently match the hints. Maybe the hints set does not contain a good representative set of valid configs?Working around this by using the full hints set instead.\r\n",,terminal_output
+195,854207,"TERMINAL",0,0,"2025-09-09 15:16:48.135477: W external/xla/xla/service/gpu/autotuning/dot_search_space.cc:200] All configs were filtered out because none of them sufficiently match the hints. Maybe the hints set does not contain a good representative set of valid configs?Working around this by using the full hints set instead.\r\n",,terminal_output
+196,865530,"TERMINAL",0,0,"SSIM: 0.6239567399024963\r\n",,terminal_output
+197,869800,"TERMINAL",0,0,"]0;tum_cte0515@hkn0818:~/Projects/jafar[?2004h(jafar) [tum_cte0515@hkn0818 jafar]$ ",,terminal_output

927a8af5474e5654810c00ce2e09fd2de87d3e5722f33fa1090d867db114e403/crowd-code-629e4b25-2201-4663-a7f2-936116295b151757499356136-2025_09_10-12.16.59.652/source.csv

@@ -0,0 +1,3 @@
+Sequence,Time,File,RangeOffset,RangeLength,Text,Language,Type
+1,5,"train_dynamics.py",0,0,"from dataclasses import dataclass, field\nimport os\nimport time\n\nimport einops\nfrom flax.training import orbax_utils\nfrom flax.training.train_state import TrainState\nimport optax\nimport orbax\nimport numpy as np\nimport jax\nimport jax.numpy as jnp\nimport tyro\nimport wandb\n\nfrom genie import Genie, restore_genie_components\nfrom utils.dataloader import get_dataloader\n\nts = int(time.time())\n\n\n@dataclass\nclass Args:\n    # Experiment\n    num_steps: int = 200_000\n    seed: int = 0\n    seq_len: int = 16\n    image_channels: int = 3\n    image_resolution: int = 64\n    data_dir: str = ""data/coinrun_episodes""\n    # Optimization\n    batch_size: int = 36\n    min_lr: float = 3e-6\n    max_lr: float = 3e-5\n    warmup_steps: int = 5000\n    # Tokenizer\n    tokenizer_dim: int = 512\n    latent_patch_dim: int = 32\n    num_patch_latents: int = 1024\n    patch_size: int = 4\n    tokenizer_num_blocks: int = 8\n    tokenizer_num_heads: int = 8\n    tokenizer_checkpoint: str = """"\n    # LAM\n    lam_dim: int = 512\n    latent_action_dim: int = 32\n    num_latent_actions: int = 6\n    lam_patch_size: int = 16\n    lam_num_blocks: int = 8\n    lam_num_heads: int = 8\n    lam_checkpoint: str = """"\n    # Dynamics\n    dyna_dim: int = 512\n    dyna_num_blocks: int = 12\n    dyna_num_heads: int = 8\n    dropout: float = 0.0\n    mask_limit: float = 0.5\n    # Logging\n    log: bool = False\n    entity: str = """"\n    project: str = """"\n    log_interval: int = 5\n    log_image_interval: int = 250\n    ckpt_dir: str = """"\n    log_checkpoint_interval: int = 25000\n    log_gradients: bool = False\n    name: str = """"\n    tags: list[str] = field(default_factory=lambda: [""dynamics""])\n    wandb_id: str = """"\n\nargs = tyro.cli(Args)\n\n\ndef dynamics_loss_fn(params, state, inputs):\n    """"""Compute masked dynamics loss""""""\n    outputs = state.apply_fn(\n        params, inputs, training=True, rngs={""dropout"": inputs[""dropout_rng""]}\n    )\n    mask = outputs[""mask""]\n    ce_loss = optax.softmax_cross_entropy_with_integer_labels(\n        outputs[""token_logits""], outputs[""video_tokens""]\n    )\n    ce_loss = (mask * ce_loss).sum() / mask.sum()\n    acc = outputs[""token_logits""].argmax(-1) == outputs[""video_tokens""]\n    acc = (mask * acc).sum() / mask.sum()\n    select_probs = jax.nn.softmax(outputs[""token_logits""])\n    metrics = dict(\n        cross_entropy_loss=ce_loss,\n        masked_token_accuracy=acc,\n        select_logit=outputs[""token_logits""].max(-1).mean(),\n        select_p=select_probs.max(-1).mean(),\n        entropy=jax.scipy.special.entr(select_probs).sum(-1).mean(),\n    )\n    return ce_loss, (outputs[""recon""], metrics)\n\n\n@jax.jit\ndef train_step(state, inputs):\n    """"""Update state and compute metrics""""""\n    grad_fn = jax.value_and_grad(dynamics_loss_fn, has_aux=True, allow_int=True)\n    (loss, (recon, metrics)), grads = grad_fn(state.params, state, inputs)\n    state = state.apply_gradients(grads=grads)\n    if args.log_gradients:\n        metrics[""gradients_std/""] = jax.tree.map(\n            lambda x: x.std(), grads[""params""][""dynamics""]\n        )\n    return state, loss, recon, metrics\n\n\nif __name__ == ""__main__"":\n    rng = jax.random.PRNGKey(args.seed)\n    if args.log:\n        wandb_init_kwargs = {\n            ""entity"": args.entity,\n            ""project"": args.project,\n            ""name"": args.name,\n            ""tags"": args.tags,\n            ""group"": ""debug"",\n            ""config"": args,\n        }\n        if args.wandb_id:\n            wandb_init_kwargs.update(\n                {\n                    ""id"": args.wandb_id,\n                    ""resume"": ""allow"",\n                }\n            )\n        wandb.init(**wandb_init_kwargs)\n\n    # --- Initialize model ---\n    genie = Genie(\n        # Tokenizer\n        in_dim=args.image_channels,\n        tokenizer_dim=args.tokenizer_dim,\n        latent_patch_dim=args.latent_patch_dim,\n        num_patch_latents=args.num_patch_latents,\n        patch_size=args.patch_size,\n        tokenizer_num_blocks=args.tokenizer_num_blocks,\n        tokenizer_num_heads=args.tokenizer_num_heads,\n        # LAM\n        lam_dim=args.lam_dim,\n        latent_action_dim=args.latent_action_dim,\n        num_latent_actions=args.num_latent_actions,\n        lam_patch_size=args.lam_patch_size,\n        lam_num_blocks=args.lam_num_blocks,\n        lam_num_heads=args.lam_num_heads,\n        # Dynamics\n        dyna_dim=args.dyna_dim,\n        dyna_num_blocks=args.dyna_num_blocks,\n        dyna_num_heads=args.dyna_num_heads,\n        dropout=args.dropout,\n        mask_limit=args.mask_limit,\n    )\n    rng, _rng = jax.random.split(rng)\n    image_shape = (args.image_resolution, args.image_resolution, args.image_channels)\n    dummy_inputs = dict(\n        videos=jnp.zeros(\n            (args.batch_size, args.seq_len, *image_shape), dtype=jnp.float32\n        ),\n        mask_rng=_rng,\n    )\n    rng, _rng = jax.random.split(rng)\n    init_params = genie.init(_rng, dummy_inputs)\n    init_params = restore_genie_components(\n        init_params, args.tokenizer_checkpoint, args.lam_checkpoint\n    )\n\n    # --- Initialize optimizer ---\n    lr_schedule = optax.warmup_cosine_decay_schedule(\n        args.min_lr, args.max_lr, args.warmup_steps, args.num_steps\n    )\n    tx = optax.adamw(learning_rate=lr_schedule, b1=0.9, b2=0.9, weight_decay=1e-4)\n    train_state = TrainState.create(apply_fn=genie.apply, params=init_params, tx=tx)\n\n    # --- TRAIN LOOP ---\n    dataloader = get_dataloader(args.data_dir, args.seq_len, args.batch_size)\n    step = 0\n    while step < args.num_steps:\n        for videos in dataloader:\n            # --- Train step ---\n            rng, _rng, _mask_rng = jax.random.split(rng, 3)\n            inputs = dict(\n                videos=videos,\n                action=jnp.zeros((args.batch_size, args.seq_len), dtype=jnp.float32),\n                dropout_rng=_rng,\n                mask_rng=_mask_rng,\n            )\n            train_state, loss, recon, metrics = train_step(train_state, inputs)\n            print(f""Step {step}, loss: {loss}"")\n            step += 1\n\n            # --- Logging ---\n            if args.log:\n                if step % args.log_interval == 0:\n                    wandb.log({""loss"": loss, ""step"": step, **metrics})\n                if step % args.log_image_interval == 0:\n                    gt_seq = inputs[""videos""][0]\n                    recon_seq = recon[0].clip(0, 1)\n                    comparison_seq = jnp.concatenate((gt_seq, recon_seq), axis=1)\n                    comparison_seq = einops.rearrange(\n                        comparison_seq * 255, ""t h w c -> h (t w) c""\n                    )\n                    log_images = dict(\n                        image=wandb.Image(np.asarray(gt_seq[15])),\n                        recon=wandb.Image(np.asarray(recon_seq[15])),\n                        true_vs_recon=wandb.Image(\n                            np.asarray(comparison_seq.astype(np.uint8))\n                        ),\n                    )\n                    wandb.log(log_images)\n                if step % args.log_checkpoint_interval == 0:\n                    ckpt = {""model"": train_state}\n                    orbax_checkpointer = orbax.checkpoint.PyTreeCheckpointer()\n                    save_args = orbax_utils.save_args_from_target(ckpt)\n                    orbax_checkpointer.save(\n                        os.path.join(os.getcwd(), args.ckpt_dir, f""genie_{ts}_{step}""),\n                        ckpt,\n                        save_args=save_args,\n                    )\n            if step >= args.num_steps:\n                break\n",python,tab
+2,2706,"extension-output-pdoom-org.crowd-code-#1-crowd-code",0,0,"12:16:59 PM [info] Activating crowd-code\n12:16:59 PM [info] Recording started\n12:16:59 PM [info] Initializing git provider using file system watchers...\n12:17:00 PM [info] Git repository found\n12:17:00 PM [info] Git provider initialized successfully\n12:17:00 PM [info] Initial git state: [object Object]\n",Log,tab

927a8af5474e5654810c00ce2e09fd2de87d3e5722f33fa1090d867db114e403/crowd-code-6aec90c8-8d95-4bad-afcb-92c28c6ff5241753889052956-2025_07_30-17.24.41.914/source.csv

@@ -0,0 +1,33 @@
+Sequence,Time,File,RangeOffset,RangeLength,Text,Language,Type
+2,825,"extension-output-pdoom-org.crowd-code-#1-crowd-code",0,0,"5:24:41 PM [info] Activating crowd-code\n5:24:41 PM [info] Recording started\n5:24:41 PM [info] Initializing git provider using file system watchers...\n",Log,tab
+3,828,"extension-output-pdoom-org.crowd-code-#1-crowd-code",150,0,"5:24:42 PM [info] Git repository found\n5:24:42 PM [info] Git provider initialized successfully\n",Log,content
+4,1007,"extension-output-pdoom-org.crowd-code-#1-crowd-code",245,0,"5:24:42 PM [info] Initial git state: [object Object]\n",Log,content
+5,3381,"TERMINAL",0,0,"git branch",,terminal_command
+6,3453,"TERMINAL",0,0,"]633;E;2025-07-30 17:24:45 git branch;adbf53fe-397b-40d3-9339-94ea79afad56]633;C[?1h=\r  add-wandb-name-and-tags\r\n  before-nnx\r\n  causal-st-transformer\r\n  causal-transformer-dynamics-model\r\n  convert-to-jax-array-in-iter\r\n  correct-batched-sampling\r\n  dev\r\n  dont-let-tf-see-gpu\r\n  feat/explicit-image-dims\r\n  fix-action-padding-lam-future-information-access\r\n  fix-sampling\r\n  fix-transformer-forwardpass\r\n  fix/spatiotemporal-pe-once-in-STTransformer\r\n  grad-norm-log-and-clip\r\n  grain-dataloader\r\n  logging-variants\r\n  lr-schedules\r\n  main\r\n  maskgit-different-maskprob-per-sample\r\n  metrics-logging-for-dynamics-model\r\n  monkey-patch\r\n* new-arch-sampling\r\n  preprocess_video\r\n  refactor-tmp\r\n  revised-dataloader\r\n  runner\r\n  runner-grain\r\n  sample-from-different-topologies\r\n  speedup-tfrecord-preprocessing\r\n  tmp\r\n:",,terminal_output
+7,8802,"TERMINAL",0,0,"  add-wandb-name-and-tags\r\n  before-nnx\r\n  causal-st-transformer\r\n  causal-transformer-dynamics-model\r\n  convert-to-jax-array-in-iter\r\n  correct-batched-sampling\r\n  dev\r\n  dont-let-tf-see-gpu\r\n  feat/explicit-image-dims\r\n  fix-action-padding-lam-future-information-access\r\n  fix-sampling\r\n  fix-transformer-forwardpass\r\n  fix/spatiotemporal-pe-once-in-STTransformer\r\n  grad-norm-log-and-clip\r\n  grain-dataloader\r\n  logging-variants\r\n  lr-schedules\r\n  main\r\n  maskgit-different-maskprob-per-sample\r\n  metrics-logging-for-dynamics-model\r\n  monkey-patch\r\n* new-arch-sampling\r\n  preprocess_video\r\n  refactor-tmp\r\n  revised-dataloader\r\n  runner\r\n  runner-grain\r\n  sample-from-different-topologies\r\n  speedup-tfrecord-preprocessing\r\n  tmp\r\n:",,terminal_output
+8,9676,"TERMINAL",0,0,"\r[?1l>\r\n]0;tum_cte0515@hkn1991:~/Projects/jafar]633;D;130",,terminal_output
+9,12549,"TERMINAL",0,0,"git checkout fix/spatiotemporal-pe-once-in-STTransformer",,terminal_command
+10,12581,"TERMINAL",0,0,"]633;E;2025-07-30 17:24:54 git checkout fix/spatiotemporal-pe-once-in-STTransformer;adbf53fe-397b-40d3-9339-94ea79afad56]633;C",,terminal_output
+11,12867,"TERMINAL",0,0,"g",,terminal_output
+12,12874,"TERMINAL",0,0,"[?25li[?25h",,terminal_output
+13,12941,"TERMINAL",0,0,"[?25lSwitched to branch 'fix/spatiotemporal-pe-once-in-STTransformer'\r\n]0;tum_cte0515@hkn1991:~/Projects/jafar]633;D;0[?25h",,terminal_output
+14,13779,"TERMINAL",0,0,"git diff",,terminal_command
+15,15343,"extension-output-pdoom-org.crowd-code-#1-crowd-code",0,0,"Switched from branch 'new-arch-sampling' to 'fix/spatiotemporal-pe-once-in-STTransformer'",Log,git_branch_checkout
+16,15499,"extension-output-pdoom-org.crowd-code-#1-crowd-code",298,0,"5:24:57 PM [info] Branch checkout detected: new-arch-sampling -> fix/spatiotemporal-pe-once-in-STTransformer\n5:24:57 PM [info] Recording git checkout: Switched from branch 'new-arch-sampling' to 'fix/spatiotemporal-pe-once-in-STTransformer'\n5:24:57 PM [info] Resetting file cache due to branch checkout\n",Log,content
+17,16601,"TERMINAL",0,0,"git checkout main",,terminal_command
+18,16637,"TERMINAL",0,0,"]633;E;2025-07-30 17:24:58 git checkout main;adbf53fe-397b-40d3-9339-94ea79afad56]633;CSwitched to branch 'main'\r\nYour branch is up to date with 'origin/main'.\r\n]0;tum_cte0515@hkn1991:~/Projects/jafar]633;D;0",,terminal_output
+19,20343,"extension-output-pdoom-org.crowd-code-#1-crowd-code",0,0,"Switched from branch 'fix/spatiotemporal-pe-once-in-STTransformer' to 'main'",Log,git_branch_checkout
+20,20470,"extension-output-pdoom-org.crowd-code-#1-crowd-code",601,0,"5:25:02 PM [info] Branch checkout detected: fix/spatiotemporal-pe-once-in-STTransformer -> main\n5:25:02 PM [info] Recording git checkout: Switched from branch 'fix/spatiotemporal-pe-once-in-STTransformer' to 'main'\n5:25:02 PM [info] Resetting file cache due to branch checkout\n",Log,content
+21,78754,"TERMINAL",0,0,"git checkout fix/spatiotemporal-pe-once-in-STTransformer",,terminal_command
+22,80243,"TERMINAL",0,0,"git push",,terminal_command
+23,80344,"extension-output-pdoom-org.crowd-code-#1-crowd-code",0,0,"Switched from branch 'main' to 'fix/spatiotemporal-pe-once-in-STTransformer'",Log,git_branch_checkout
+24,80522,"extension-output-pdoom-org.crowd-code-#1-crowd-code",878,0,"5:26:02 PM [info] Branch checkout detected: main -> fix/spatiotemporal-pe-once-in-STTransformer\n5:26:02 PM [info] Recording git checkout: Switched from branch 'main' to 'fix/spatiotemporal-pe-once-in-STTransformer'\n5:26:02 PM [info] Resetting file cache due to branch checkout\n",Log,content
+25,83258,"TERMINAL",0,0,"git push --set-upstream origin fix/spatiotemporal-pe-once-in-STTransformer",,terminal_command
+26,83333,"TERMINAL",0,0,"]633;E;2025-07-30 17:26:05 git push --set-upstream origin fix/spatiotemporal-pe-once-in-STTransformer;adbf53fe-397b-40d3-9339-94ea79afad56]633;C",,terminal_output
+27,84677,"TERMINAL",0,0,"Enumerating objects: 7, done.\r\nCounting objects:  14% (1/7)\rCounting objects:  28% (2/7)\rCounting objects:  42% (3/7)\rCounting objects:  57% (4/7)\rCounting objects:  71% (5/7)\rCounting objects:  85% (6/7)\rCounting objects: 100% (7/7)\rCounting objects: 100% (7/7), done.\r\nDelta compression using up to 152 threads\r\nCompressing objects:  25% (1/4)\rCompressing objects:  50% (2/4)\rCompressing objects:  75% (3/4)\rCompressing objects: 100% (4/4)\rCompressing objects: 100% (4/4), done.\r\nWriting objects:  25% (1/4)\rWriting objects:  50% (2/4)\rWriting objects:  75% (3/4)\rWriting objects: 100% (4/4)\rWriting objects: 100% (4/4), 870 bytes | 870.00 KiB/s, done.\r\nTotal 4 (delta 3), reused 0 (delta 0), pack-reused 0\r\n",,terminal_output
+28,84825,"TERMINAL",0,0,"remote: Resolving deltas:   0% (0/3)\rremote: Resolving deltas:  33% (1/3)\rremote: Resolving deltas:  66% (2/3)\rremote: Resolving deltas: 100% (3/3)\rremote: Resolving deltas: 100% (3/3), completed with 3 local objects.\r\nremote: This repository moved. Please use the new location:\r\nremote:   git@github.com:p-doom/jasmine.git\r\n",,terminal_output
+29,84974,"TERMINAL",0,0,"remote: \r\nremote: Create a pull request for 'fix/spatiotemporal-pe-once-in-STTransformer' on GitHub by visiting:\r\nremote:      https://github.com/p-doom/jasmine/pull/new/fix/spatiotemporal-pe-once-in-STTransformer\r\nremote: \r\n",,terminal_output
+30,85061,"TERMINAL",0,0,"To github.com:p-doom/jafar.git\r\n * [new branch]      fix/spatiotemporal-pe-once-in-STTransformer -> fix/spatiotemporal-pe-once-in-STTransformer\r\nbranch 'fix/spatiotemporal-pe-once-in-STTransformer' set up to track 'origin/fix/spatiotemporal-pe-once-in-STTransformer'.\r\n",,terminal_output
+31,85082,"TERMINAL",0,0,"]0;tum_cte0515@hkn1991:~/Projects/jafar]633;D;0]633;P;Cwd=/home/hk-project-p0023960/tum_cte0515/Projects/jafar",,terminal_output
+32,167394,"genie.py",0,0,"from typing import Dict\n\nimport optax\nimport jax\nimport jax.numpy as jnp\nimport flax.nnx as nnx\nimport orbax.checkpoint as ocp\n\nfrom models.dynamics import DynamicsMaskGIT\nfrom models.lam import LatentActionModel\nfrom models.tokenizer import TokenizerVQVAE\n\n\nclass Genie(nnx.Module):\n    """"""Genie model""""""\n\n    def __init__(\n        self,\n        in_dim: int,\n        tokenizer_dim: int,\n        tokenizer_ffn_dim: int,\n        latent_patch_dim: int,\n        num_patch_latents: int,\n        patch_size: int,\n        tokenizer_num_blocks: int,\n        tokenizer_num_heads: int,\n        lam_dim: int,\n        lam_ffn_dim: int,\n        latent_action_dim: int,\n        num_latent_actions: int,\n        lam_patch_size: int,\n        lam_num_blocks: int,\n        lam_num_heads: int,\n        lam_co_train: bool,\n        dyna_dim: int,\n        dyna_ffn_dim: int,\n        dyna_num_blocks: int,\n        dyna_num_heads: int,\n        param_dtype: jnp.dtype,\n        dtype: jnp.dtype,\n        use_flash_attention: bool,\n        rngs: nnx.Rngs,\n        dropout: float = 0.0,\n        mask_limit: float = 0.0,\n    ):\n        # --- Tokenizer ---\n        self.in_dim = in_dim\n        self.tokenizer_dim = tokenizer_dim\n        self.tokenizer_ffn_dim = tokenizer_ffn_dim\n        self.latent_patch_dim = latent_patch_dim\n        self.num_patch_latents = num_patch_latents\n        self.patch_size = patch_size\n        self.tokenizer_num_blocks = tokenizer_num_blocks\n        self.tokenizer_num_heads = tokenizer_num_heads\n        # --- LAM ---\n        self.lam_dim = lam_dim\n        self.lam_ffn_dim = lam_ffn_dim\n        self.latent_action_dim = latent_action_dim\n        self.num_latent_actions = num_latent_actions\n        self.lam_patch_size = lam_patch_size\n        self.lam_num_blocks = lam_num_blocks\n        self.lam_num_heads = lam_num_heads\n        self.lam_co_train = lam_co_train\n        # --- Dynamics ---\n        self.dyna_dim = dyna_dim\n        self.dyna_ffn_dim = dyna_ffn_dim\n        self.dyna_num_blocks = dyna_num_blocks\n        self.dyna_num_heads = dyna_num_heads\n        self.param_dtype = param_dtype\n        self.dtype = dtype\n        self.use_flash_attention = use_flash_attention\n        self.dropout = dropout\n        self.mask_limit = mask_limit\n\n        self.tokenizer = TokenizerVQVAE(\n            in_dim=self.in_dim,\n            model_dim=self.tokenizer_dim,\n            ffn_dim=self.tokenizer_ffn_dim,\n            latent_dim=self.latent_patch_dim,\n            num_latents=self.num_patch_latents,\n            patch_size=self.patch_size,\n            num_blocks=self.tokenizer_num_blocks,\n            num_heads=self.tokenizer_num_heads,\n            dropout=0.0,\n            codebook_dropout=0.0,\n            param_dtype=self.param_dtype,\n            dtype=self.dtype,\n            use_flash_attention=self.use_flash_attention,\n            rngs=rngs,\n        )\n        self.lam = LatentActionModel(\n            in_dim=self.in_dim,\n            model_dim=self.lam_dim,\n            ffn_dim=self.lam_ffn_dim,\n            latent_dim=self.latent_patch_dim,\n            num_latents=self.num_latent_actions,\n            patch_size=self.lam_patch_size,\n            num_blocks=self.lam_num_blocks,\n            num_heads=self.lam_num_heads,\n            dropout=0.0,\n            codebook_dropout=0.0,\n            param_dtype=self.param_dtype,\n            dtype=self.dtype,\n            use_flash_attention=self.use_flash_attention,\n            rngs=rngs,\n        )\n        self.dynamics = DynamicsMaskGIT(\n            model_dim=self.dyna_dim,\n            ffn_dim=self.dyna_ffn_dim,\n            num_latents=self.num_patch_latents,\n            latent_action_dim=self.latent_action_dim,\n            num_blocks=self.dyna_num_blocks,\n            num_heads=self.dyna_num_heads,\n            dropout=self.dropout,\n            mask_limit=self.mask_limit,\n            param_dtype=self.param_dtype,\n            dtype=self.dtype,\n            use_flash_attention=self.use_flash_attention,\n            rngs=rngs,\n        )\n\n    def __call__(\n        self, batch: Dict[str, jax.Array], training: bool = True\n    ) -> Dict[str, jax.Array]:\n        videos_BTHWC = batch[""videos""]\n        tokenizer_outputs = self.tokenizer.vq_encode(videos_BTHWC, training=False)\n        token_indices_BTN = tokenizer_outputs[""indices""]\n        lam_outputs = self.lam.vq_encode(videos_BTHWC, training=False)\n        z_q_BTm11L = lam_outputs[""z_q""]\n        action_indices_E = lam_outputs[""indices""]\n        latent_actions_BTm11L = jax.lax.cond(\n            self.lam_co_train,\n            lambda: z_q_BTm11L,\n            lambda: jax.lax.stop_gradient(z_q_BTm11L),\n        )\n        outputs = dict(\n            video_tokens=jax.lax.stop_gradient(token_indices_BTN),\n            latent_actions=latent_actions_BTm11L,\n        )\n        outputs[""mask_rng""] = batch[""mask_rng""]\n        dyna_logits_BTNV, dyna_mask = self.dynamics(outputs, training)\n        outputs[""token_logits""] = dyna_logits_BTNV\n        if dyna_mask is not None:\n            outputs[""mask""] = dyna_mask\n        mle_indices_BTN = jnp.argmax(outputs[""token_logits""], axis=-1)\n        H, W = batch[""videos""].shape[2:4]\n        outputs[""recon""] = self.tokenizer.decode(mle_indices_BTN, (H, W))\n        outputs[""lam_indices""] = action_indices_E\n        return outputs\n\n    def sample(\n        self,\n        batch: Dict[str, jax.Array],\n        seq_len: int,\n        steps: int = 25,\n        temperature: float = 1,\n        sample_argmax: bool = False,\n    ) -> jax.Array:\n        """"""\n        Autoregressively samples up to `seq_len` future frames, following Figure 8 of the paper.\n\n        - Input frames are tokenized once.\n        - Future frames are generated autoregressively in token space.\n        - All frames are detokenized in a single pass.\n\n        Note:\n        - For interactive or step-wise sampling, detokenization should occur after each action.\n        - To maintain consistent tensor shapes across timesteps, all current and future frames are decoded at every step.\n        - Temporal causal structure is preserved by\n            a) reapplying the mask before each decoding step.\n            b) a temporal causal mask is applied within each ST-transformer block.\n\n        Dimension keys:\n            B: batch size\n            T: number of input (conditioning) frames\n            N: number of patches per frame\n            M: model dimension\n            S: sequence length\n            H: height\n            W: width\n            D: B * T * N\n            E: B * (T - 1)\n        """"""\n        # --- Encode videos and actions ---\n        videos_BTHWC = batch[""videos""]\n        latent_actions_E = batch[""latent_actions""]\n        tokenizer_out = self.tokenizer.vq_encode(videos_BTHWC, training=False)\n        token_idxs_BTN = tokenizer_out[""indices""]\n        B, T, N = token_idxs_BTN.shape\n        pad_shape = (B, seq_len - T, N)\n        pad = jnp.zeros(pad_shape, dtype=token_idxs_BTN.dtype)\n        token_idxs_BSN = jnp.concatenate([token_idxs_BTN, pad], axis=1)\n        action_tokens_EL = self.lam.vq.get_codes(latent_actions_E)\n\n        def maskgit_step_fn(\n            carry: tuple[jax.Array, jax.Array, jax.Array, jax.Array], step: jax.Array\n        ) -> tuple[tuple[jax.Array, jax.Array, jax.Array, jax.Array], None]:\n            rng, token_idxs_BSN, mask_BSN, action_tokens_EL = carry\n            S, N = token_idxs_BSN.shape[1:]\n\n            # --- Construct + encode video ---\n            vid_embed_BSNM = self.dynamics.patch_embed(token_idxs_BSN)\n            mask_token_111M = self.dynamics.mask_token.value\n            mask_expanded_BSN1 = mask_BSN[..., None]\n            vid_embed_BSNM = jnp.where(mask_expanded_BSN1, mask_token_111M, vid_embed_BSNM)\n\n            # --- Predict transition ---\n            action_tokens_BSm1L = jnp.reshape(action_tokens_EL, (B, S - 1, 1))\n            act_embed_BSm1M = self.dynamics.action_up(action_tokens_BSm1L)\n            # FIXME (f.srambical): We must not pad the actions, but remove the last frame (https://github.com/p-doom/jasmine/issues/122)\n            vid_embed_BSNM += jnp.pad(act_embed_BSm1M, ((0, 0), (1, 0), (0, 0), (0, 0)))\n            unmasked_ratio = jnp.cos(jnp.pi * (step + 1) / (steps * 2))\n            step_temp = temperature * (1.0 - unmasked_ratio)\n            final_logits_BSNV = self.dynamics.transformer(vid_embed_BSNM) / step_temp\n\n            # --- Sample new tokens for final frame ---\n            if sample_argmax:\n                sampled_token_idxs_BSN = jnp.argmax(final_logits_BSNV, axis=-1)\n            else:\n                rng, _rng = jax.random.split(rng)\n                sampled_token_idxs_BSN = jax.random.categorical(_rng, final_logits_BSNV)\n            gather_fn = jax.vmap(jax.vmap(jax.vmap(lambda x, y: x[y])))\n            final_token_probs_BSN = gather_fn(\n                jax.nn.softmax(final_logits_BSNV), sampled_token_idxs_BSN\n            )\n            final_token_probs_BSN += ~mask_BSN\n            # Update masked tokens only\n            token_idxs_BSN = jnp.where(mask_BSN, sampled_token_idxs_BSN, token_idxs_BSN)\n\n            # --- Update mask ---\n            num_unmasked_tokens = jnp.round(N * (1.0 - unmasked_ratio)).astype(int)\n            idx_mask_N = jnp.arange(final_token_probs_BSN.shape[-1]) > num_unmasked_tokens\n            sorted_idxs_BSN = jnp.argsort(final_token_probs_BSN, axis=-1, descending=True)\n            mask_update_fn = jax.vmap(lambda msk, ids: msk.at[ids].set(idx_mask_N))\n            new_mask_BSN = mask_update_fn(mask_BSN, sorted_idxs_BSN)\n\n            new_carry = (rng, token_idxs_BSN, new_mask_BSN, action_tokens_EL)\n            return new_carry, None\n\n        def generation_step_fn(\n            carry: tuple[jax.Array, jax.Array], step_t: jax.Array\n        ) -> tuple[tuple[jax.Array, jax.Array], None]:\n            rng, current_token_idxs_BSN = carry\n            rng, step_rng = jax.random.split(rng)\n\n            # Mask current and future frames (i.e., t >= step_t)\n            mask_S = jnp.arange(seq_len) >= step_t\n            mask_BSN = jnp.broadcast_to(mask_S[None, :, None], (B, seq_len, N)).astype(\n                bool\n            )\n            masked_token_idxs_BSN = current_token_idxs_BSN * ~mask_BSN\n\n            # --- Initialize and run MaskGIT loop ---\n            init_carry_maskgit = (\n                step_rng,\n                masked_token_idxs_BSN,\n                mask_BSN,\n                action_tokens_EL,\n            )\n            final_carry_maskgit, _ = jax.lax.scan(\n                maskgit_step_fn, init_carry_maskgit, jnp.arange(steps)\n            )\n            updated_token_idxs = final_carry_maskgit[1]\n            new_carry = (rng, updated_token_idxs)\n            return new_carry, None\n\n        # --- Run the autoregressive generation using jax.lax.scan ---\n        initial_carry = (batch[""rng""], token_idxs_BSN)\n        timesteps_to_scan = jnp.arange(T, seq_len)\n        final_carry, _ = jax.lax.scan(\n            generation_step_fn, initial_carry, timesteps_to_scan\n        )\n        final_token_idxs = final_carry[1]\n\n        # --- Decode all tokens at once at the end ---\n        H, W = batch[""videos""].shape[2:4]\n        final_frames = self.tokenizer.decode(\n            final_token_idxs,\n            video_hw=(H, W),\n        )\n        return final_frames\n\n    def vq_encode(self, batch: Dict[str, jax.Array], training: bool) -> jax.Array:\n        # --- Preprocess videos ---\n        video_BTHWC = batch[""videos""]\n        lam_output = self.lam.vq_encode(video_BTHWC, training=training)\n        lam_indices_E = lam_output[""indices""]\n        return lam_indices_E\n\n\n# FIXME (f.srambical): add conversion script for old checkpoints\ndef restore_genie_components(\n    optimizer: nnx.Optimizer,\n    sharding: jax.sharding.NamedSharding,\n    rng: jax.Array,\n    args,\n) -> nnx.Optimizer:\n    """"""Restore pre-trained Genie components""""""\n    rngs = nnx.Rngs(rng)\n\n    # dummy values since we only use tx to initialize the dummy train states\n    dummy_tx = optax.adamw(\n        learning_rate=optax.constant_schedule(args.max_lr),\n        b1=0.9,\n        b2=0.9,\n        weight_decay=1e-4,\n    )\n    handler_registry = ocp.handlers.DefaultCheckpointHandlerRegistry()\n    handler_registry.add(\n        ""model_state"", ocp.args.PyTreeRestore, ocp.handlers.PyTreeCheckpointHandler\n    )\n\n    checkpoint_options = ocp.CheckpointManagerOptions(\n        step_format_fixed_length=6,\n    )\n    tokenizer_checkpoint_manager = ocp.CheckpointManager(\n        directory=args.tokenizer_checkpoint,\n        options=checkpoint_options,\n        handler_registry=handler_registry,\n    )\n    dummy_tokenizer = TokenizerVQVAE(\n        in_dim=args.image_channels,\n        model_dim=args.tokenizer_dim,\n        ffn_dim=args.tokenizer_ffn_dim,\n        latent_dim=args.latent_patch_dim,\n        num_latents=args.num_patch_latents,\n        patch_size=args.patch_size,\n        num_blocks=args.tokenizer_num_blocks,\n        num_heads=args.tokenizer_num_heads,\n        dropout=args.dropout,\n        codebook_dropout=args.dropout,\n        param_dtype=args.param_dtype,\n        dtype=args.dtype,\n        use_flash_attention=args.use_flash_attention,\n        rngs=rngs,\n    )\n    dummy_tokenizer_optimizer = nnx.Optimizer(dummy_tokenizer, dummy_tx)\n    dummy_tokenizer_optimizer_state = nnx.state(dummy_tokenizer_optimizer)\n    abstract_sharded_tokenizer_optimizer_state = _create_abstract_sharded_pytree(\n        dummy_tokenizer_optimizer_state, sharding\n    )\n    restored_tokenizer = tokenizer_checkpoint_manager.restore(\n        step=tokenizer_checkpoint_manager.latest_step(),\n        args=ocp.args.Composite(\n            model_state=ocp.args.PyTreeRestore(  # type: ignore\n                abstract_sharded_tokenizer_optimizer_state  # type: ignore\n            ),\n        ),\n    )[""model_state""]\n    nnx.update(dummy_tokenizer_optimizer.model, restored_tokenizer.model)\n    optimizer.model.tokenizer = dummy_tokenizer_optimizer.model\n    tokenizer_checkpoint_manager.close()\n\n    if args.lam_checkpoint:\n        lam_checkpoint_manager = ocp.CheckpointManager(\n            directory=args.lam_checkpoint,\n            options=checkpoint_options,\n            handler_registry=handler_registry,\n        )\n        dummy_lam = LatentActionModel(\n            in_dim=args.image_channels,\n            model_dim=args.lam_dim,\n            ffn_dim=args.lam_ffn_dim,\n            latent_dim=args.latent_patch_dim,\n            num_latents=args.num_latent_actions,\n            patch_size=args.lam_patch_size,\n            num_blocks=args.lam_num_blocks,\n            num_heads=args.lam_num_heads,\n            dropout=args.dropout,\n            codebook_dropout=args.dropout,\n            param_dtype=args.param_dtype,\n            dtype=args.dtype,\n            use_flash_attention=args.use_flash_attention,\n            rngs=rngs,\n        )\n        dummy_lam_optimizer = nnx.Optimizer(dummy_lam, dummy_tx)\n        dummy_lam_optimizer_state = nnx.state(dummy_lam_optimizer)\n        abstract_sharded_lam_optimizer_state = _create_abstract_sharded_pytree(\n            dummy_lam_optimizer_state, sharding\n        )\n        restored_lam_optimizer = lam_checkpoint_manager.restore(\n            step=lam_checkpoint_manager.latest_step(),\n            args=ocp.args.Composite(\n                model_state=ocp.args.PyTreeRestore(  # type: ignore\n                    abstract_sharded_lam_optimizer_state  # type: ignore\n                ),\n            ),\n        )[""model_state""]\n        nnx.update(dummy_lam_optimizer.model, restored_lam_optimizer.model)\n        optimizer.model.lam = dummy_lam_optimizer.model\n        # Remove the LAM decoder to save memory and avoid unnecessary computation.\n        del optimizer.model.lam.decoder\n        lam_checkpoint_manager.close()\n\n    return optimizer\n\n\ndef _create_abstract_sharded_pytree(\n    pytree_template: nnx.GraphState, sharding_spec: jax.sharding.NamedSharding\n) -> jax.Array:\n    """"""Replaces arrays in a pytree with ShapeDtypeStructs having the given sharding.""""""\n\n    def map_fn(leaf_template):\n        if hasattr(leaf_template, ""shape"") and hasattr(leaf_template, ""dtype""):\n            return jax.ShapeDtypeStruct(\n                leaf_template.shape, leaf_template.dtype, sharding=sharding_spec\n            )\n        return leaf_template\n\n    return jax.tree_util.tree_map(map_fn, pytree_template)\n",python,tab
+33,168545,"genie.py",6235,0,"",python,selection_command

927a8af5474e5654810c00ce2e09fd2de87d3e5722f33fa1090d867db114e403/crowd-code-7c8743a1-f55a-4e45-b7aa-0b3df3c9f3c91752835699286-2025_07_18-12.49.02.294/source.csv

@@ -0,0 +1,394 @@
+Sequence,Time,File,RangeOffset,RangeLength,Text,Language,Type
+1,4,"models/dynamics.py",0,0,"from typing import Dict, Any\n\nimport jax\nimport jax.numpy as jnp\nimport flax.linen as nn\nimport einops\n\nfrom utils.nn import STTransformer\n\n\nclass DynamicsMaskGIT(nn.Module):\n    """"""MaskGIT dynamics model""""""\n\n    model_dim: int\n    num_latents: int\n    num_blocks: int\n    num_heads: int\n    dropout: float\n    mask_limit: float\n    param_dtype: jnp.dtype\n    dtype: jnp.dtype\n\n    def setup(self):\n        self.dynamics = STTransformer(\n            self.model_dim,\n            self.num_latents,\n            self.num_blocks,\n            self.num_heads,\n            self.dropout,\n            self.param_dtype,\n            self.dtype,\n            spacial_bert=True,\n        )\n        self.patch_embed = nn.Embed(self.num_latents, self.model_dim)\n        self.mask_token = self.param(\n            ""mask_token"",\n            nn.initializers.lecun_uniform(),\n            (1, 1, 1, self.model_dim),\n        )\n        self.action_up = nn.Dense(\n            self.model_dim,\n            param_dtype=self.param_dtype,\n            dtype=self.dtype,\n        )\n\n    def __call__(self, batch: Dict[str, Any], training: bool = True) -> Dict[str, Any]:\n        # --- Mask videos ---\n        vid_embed = self.patch_embed(batch[""video_tokens""])\n        if training:\n            rng1, rng2 = jax.random.split(batch[""mask_rng""])\n            mask_prob = jax.random.uniform(rng1, minval=self.mask_limit)\n            mask = jax.random.bernoulli(rng2, mask_prob, vid_embed.shape[:-1])\n            mask = mask.at[:, 0].set(False)\n            vid_embed = jnp.where(jnp.expand_dims(mask, -1), self.mask_token, vid_embed)\n        else:\n            mask = None\n\n        # --- Predict transition ---\n        act_embed = self.action_up(batch[""latent_actions""])\n        vid_embed += jnp.pad(act_embed, ((0, 0), (1, 0), (0, 0), (0, 0)))\n        logits = self.dynamics(vid_embed)\n        return dict(token_logits=logits, mask=mask)\n\n\nclass DynamicsAutoregressive(nn.Module):\n    """"""Autoregressive (causal) dynamics model""""""\n\n    model_dim: int\n    num_latents: int\n    num_blocks: int\n    num_heads: int\n    dropout: float\n    param_dtype: jnp.dtype\n    dtype: jnp.dtype\n\n    def setup(self):\n        self.dynamics = STTransformer(\n            self.model_dim,\n            self.num_latents,\n            self.num_blocks,\n            self.num_heads,\n            self.dropout,\n            self.param_dtype,\n            self.dtype,\n            spacial_bert=False,\n        )\n        self.patch_embed = nn.Embed(self.num_latents, self.model_dim)\n        self.action_up = nn.Dense(\n            self.model_dim,\n            param_dtype=self.param_dtype,\n            dtype=self.dtype,\n        )\n\n    def __call__(self, batch: Dict[str, Any], training: bool = True) -> Dict[str, Any]:\n        vid_embed = self.patch_embed(batch[""video_tokens""])\n        act_embed = self.action_up(batch[""latent_actions""])\n        vid_embed += jnp.pad(act_embed, ((0, 0), (1, 0), (0, 0), (0, 0)))\n        vid_embed_padded = jnp.pad(vid_embed, ((0, 0), (0, 0), (1, 0), (0, 0)))\n        logits = self.dynamics(vid_embed_padded)[:,:,1:]\n        mask = jnp.ones(vid_embed.shape[:-1])\n        return dict(token_logits=logits, mask=mask)",python,tab
+2,402,"extension-output-pdoom-org.crowd-code-#1-crowd-code",0,0,"12:49:02 PM [info] Activating crowd-code\n12:49:02 PM [info] Recording started\n12:49:02 PM [info] Initializing git provider using file system watchers...\n12:49:02 PM [info] Git repository found\n12:49:02 PM [info] Git provider initialized successfully\n12:49:02 PM [info] Initial git state: [object Object]\n",Log,tab
+3,3198,"TERMINAL",0,0,"/bin/python3 /hkfs/home/project/hk-project-p0023960/tum_cte0515/.cursor-server/extensions/ms-python.python-2024.12.3-linux-x64/python_files/printEnvVariablesToFile.py /hkfs/home/project/hk-project-p0023960/tum_cte0515/.cursor-server/extensions/ms-python.python-2024.12.3-linux-x64/python_files/deactivate/bash/envVars.txt",,terminal_command
+4,3233,"TERMINAL",0,0,"]633;E;2025-07-18 12:49:05 /bin/python3 /hkfs/home/project/hk-project-p0023960/tum_cte0515/.cursor-server/extensions/ms-python.python-2024.12.3-linux-x64/python_files/printEnvVariablesToFile.py /hkfs/home/project/hk-project-p0023960/tum_cte0515/.cursor-server/extensions/ms-python.python-2024.12.3-linux-x64/python_files/deactivate/bash/envVars.txt;3a42ff33-fe06-4226-9417-7f10804f0a18]633;C",,terminal_output
+5,3297,"TERMINAL",0,0,"]0;tum_cte0515@hkn1991:/hkfs/home/project/hk-project-p0023960/tum_cte0515/.cursor-server/extensions/ms-python.python-2024.12.3-linux-x64/python_files/deactivate/bash]633;D;0",,terminal_output
+6,4955,"models/lam.py",0,0,"from typing import Dict, Any\n\nimport jax.numpy as jnp\nimport flax.linen as nn\n\nfrom utils.preprocess import patchify, unpatchify\nfrom utils.nn import STTransformer, VectorQuantizer\n\n\nclass LatentActionModel(nn.Module):\n    """"""Latent Action ST-ViVit VQ-VAE""""""\n\n    in_dim: int\n    model_dim: int\n    latent_dim: int\n    num_latents: int\n    patch_size: int\n    num_blocks: int\n    num_heads: int\n    dropout: float\n    codebook_dropout: float\n    param_dtype: jnp.dtype\n    dtype: jnp.dtype\n\n    def setup(self):\n        self.patch_token_dim = self.in_dim * self.patch_size**2\n        self.encoder = STTransformer(\n            self.model_dim,\n            self.latent_dim,\n            self.num_blocks,\n            self.num_heads,\n            self.dropout,\n            self.param_dtype,\n            self.dtype,\n        )\n        self.action_in = self.param(\n            ""action_in"",\n            nn.initializers.lecun_uniform(),\n            (1, 1, 1, self.patch_token_dim),\n        )\n        self.vq = VectorQuantizer(\n            self.latent_dim,\n            self.num_latents,\n            self.codebook_dropout,\n        )\n        self.patch_up = nn.Dense(\n            self.model_dim,\n            param_dtype=self.param_dtype,\n            dtype=self.dtype,\n        )\n        self.action_up = nn.Dense(\n            self.model_dim,\n            param_dtype=self.param_dtype,\n            dtype=self.dtype,\n        )\n        self.decoder = STTransformer(\n            self.model_dim,\n            self.patch_token_dim,\n            self.num_blocks,\n            self.num_heads,\n            self.dropout,\n            self.param_dtype,\n            self.dtype,\n        )\n\n    def __call__(self, batch: Dict[str, Any], training: bool = True) -> Dict[str, Any]:\n        # --- Encode + VQ ---\n        H, W = batch[""videos""].shape[2:4]\n        outputs = self.vq_encode(batch[""videos""], training)\n        video_action_patches = self.action_up(outputs[""z_q""]) + self.patch_up(\n            outputs[""patches""][:, :-1]\n        )\n        del outputs[""patches""]\n\n        # --- Decode ---\n        video_recon = self.decoder(video_action_patches)\n        video_recon = video_recon.astype(jnp.float32)\n        video_recon = nn.sigmoid(video_recon)\n        video_recon = video_recon.astype(self.dtype)\n        outputs[""recon""] = unpatchify(video_recon, self.patch_size, H, W)\n        return outputs\n\n    def vq_encode(self, videos: Any, training: bool = True) -> Dict[str, Any]:\n        # --- Preprocess videos ---\n        B, T = videos.shape[:2]\n        patches = patchify(videos, self.patch_size)\n        action_pad = jnp.broadcast_to(self.action_in, (B, T, 1, self.patch_token_dim))\n        padded_patches = jnp.concatenate((action_pad, patches), axis=2)\n\n        # --- Encode ---\n        z = self.encoder(padded_patches)  # (B, T, N, E)\n        # Get latent action for all future frames\n        z = z[:, 1:, 0]  # (B, T-1, E)\n\n        # --- Vector quantize ---\n        z = z.reshape(B * (T - 1), self.latent_dim)\n        z_q, z, emb, indices = self.vq(z, training)\n        z_q = z_q.reshape(B, T - 1, 1, self.latent_dim)\n        return dict(patches=patches, z_q=z_q, z=z, emb=emb, indices=indices)\n",python,tab
+7,11679,"models/lam.py",2162,0,"",python,selection_mouse
+8,11804,"models/lam.py",2159,3,"jnp",python,selection_mouse
+9,14630,"models/lam.py",2060,0,"",python,selection_mouse
+10,14633,"models/lam.py",2059,0,"",python,selection_command
+11,15171,"models/lam.py",2107,0,"",python,selection_mouse
+12,15318,"models/lam.py",2096,20,"video_action_patches",python,selection_mouse
+13,71040,"models/lam.py",2782,0,"",python,selection_mouse
+14,71169,"models/lam.py",2778,14,"padded_patches",python,selection_mouse
+15,72501,"models/lam.py",2696,0,"",python,selection_mouse
+16,72610,"models/lam.py",2695,2,"((",python,selection_mouse
+17,79356,"models/lam.py",2538,0,"",python,selection_mouse
+18,79489,"models/lam.py",2534,8,"patchify",python,selection_mouse
+19,80076,"models/lam.py",2529,0,"",python,selection_mouse
+20,80227,"models/lam.py",2524,7,"patches",python,selection_mouse
+21,81420,"models/lam.py",2529,0,"",python,selection_mouse
+22,82156,"models/lam.py",2524,7,"patches",python,selection_mouse
+23,88062,"models/lam.py",2669,0,"",python,selection_mouse
+24,88219,"models/lam.py",2663,14,"padded_patches",python,selection_mouse
+25,95005,"models/lam.py",2831,0,"",python,selection_mouse
+26,95165,"models/lam.py",2831,6,"action",python,selection_mouse
+27,95769,"models/lam.py",2669,0,"",python,selection_mouse
+28,95909,"models/lam.py",2663,14,"padded_patches",python,selection_mouse
+29,96023,"models/lam.py",2655,72,"        padded_patches = jnp.concatenate((action_pad, patches), axis=2)\n",python,selection_mouse
+30,96634,"models/lam.py",2669,0,"",python,selection_mouse
+31,96635,"models/lam.py",2663,14,"padded_patches",python,selection_mouse
+32,156184,"models/lam.py",2933,0,"",python,selection_mouse
+33,156229,"models/lam.py",2932,0,"",python,selection_command
+34,156811,"models/lam.py",3093,0,"",python,selection_mouse
+35,156852,"models/lam.py",3092,0,"",python,selection_command
+36,157452,"models/lam.py",3170,0,"",python,selection_mouse
+37,157453,"models/lam.py",3169,0,"",python,selection_command
+38,158312,"models/lam.py",2933,0,"",python,selection_mouse
+39,158313,"models/lam.py",2932,0,"",python,selection_command
+40,223766,"models/lam.py",2500,0,"",python,selection_mouse
+41,223922,"models/lam.py",2499,6,"videos",python,selection_mouse
+42,224169,"models/lam.py",2499,12,"videos.shape",python,selection_mouse
+43,224209,"models/lam.py",2499,13,"videos.shape[",python,selection_mouse
+44,224256,"models/lam.py",2499,14,"videos.shape[:",python,selection_mouse
+45,224257,"models/lam.py",2499,15,"videos.shape[:2",python,selection_mouse
+46,224301,"models/lam.py",2499,16,"videos.shape[:2]",python,selection_mouse
+47,224636,"models/lam.py",2515,0,"",python,selection_mouse
+48,224640,"models/lam.py",2514,0,"",python,selection_command
+49,225043,"models/lam.py",2515,0,"",python,selection_mouse
+50,225045,"models/lam.py",2514,0,"",python,selection_command
+51,225205,"models/lam.py",2514,1,"]",python,selection_mouse
+52,225207,"models/lam.py",2515,0,"",python,selection_command
+53,225270,"models/lam.py",2514,1,"]",python,selection_mouse
+54,225271,"models/lam.py",2510,5,"e[:2]",python,selection_mouse
+55,225271,"models/lam.py",2509,6,"pe[:2]",python,selection_mouse
+56,225272,"models/lam.py",2508,7,"ape[:2]",python,selection_mouse
+57,225300,"models/lam.py",2506,9,"shape[:2]",python,selection_mouse
+58,225348,"models/lam.py",2505,10,".shape[:2]",python,selection_mouse
+59,225392,"models/lam.py",2504,11,"s.shape[:2]",python,selection_mouse
+60,225392,"models/lam.py",2503,12,"os.shape[:2]",python,selection_mouse
+61,225393,"models/lam.py",2502,13,"eos.shape[:2]",python,selection_mouse
+62,225395,"models/lam.py",2501,14,"deos.shape[:2]",python,selection_mouse
+63,225485,"models/lam.py",2500,15,"ideos.shape[:2]",python,selection_mouse
+64,225984,"models/lam.py",2500,0,"",python,selection_mouse
+65,225985,"models/lam.py",2499,6,"videos",python,selection_mouse
+66,226228,"models/lam.py",2499,7,"videos.",python,selection_mouse
+67,226249,"models/lam.py",2499,12,"videos.shape",python,selection_mouse
+68,226295,"models/lam.py",2499,13,"videos.shape[",python,selection_mouse
+69,226338,"models/lam.py",2499,14,"videos.shape[:",python,selection_mouse
+70,226339,"models/lam.py",2499,15,"videos.shape[:2",python,selection_mouse
+71,226348,"models/lam.py",2499,16,"videos.shape[:2]",python,selection_mouse
+72,226644,"models/lam.py",2515,0,"",python,selection_mouse
+73,226648,"models/lam.py",2514,0,"",python,selection_command
+74,226797,"models/lam.py",2515,0,"",python,selection_mouse
+75,226799,"models/lam.py",2514,0,"",python,selection_command
+76,226966,"models/lam.py",2514,1,"]",python,selection_mouse
+77,227008,"models/lam.py",2515,0,"",python,selection_command
+78,227009,"models/lam.py",2514,1,"]",python,selection_mouse
+79,227013,"models/lam.py",2512,3,":2]",python,selection_mouse
+80,227054,"models/lam.py",2506,9,"shape[:2]",python,selection_mouse
+81,227100,"models/lam.py",2505,10,".shape[:2]",python,selection_mouse
+82,227125,"models/lam.py",2499,16,"videos.shape[:2]",python,selection_mouse
+83,227722,"models/lam.py",2500,0,"",python,selection_mouse
+84,228365,"models/lam.py",2383,0,"",python,selection_mouse
+85,228547,"models/lam.py",2377,9,"vq_encode",python,selection_mouse
+86,230444,"models/lam.py",2396,0,"",python,selection_mouse
+87,230609,"models/lam.py",2393,6,"videos",python,selection_mouse
+88,231398,"models/lam.py",2384,0,"",python,selection_mouse
+89,231528,"models/lam.py",2377,9,"vq_encode",python,selection_mouse
+90,232096,"models/lam.py",2396,0,"",python,selection_mouse
+91,232248,"models/lam.py",2393,6,"videos",python,selection_mouse
+92,232911,"models/lam.py",2384,0,"",python,selection_mouse
+93,233063,"models/lam.py",2377,9,"vq_encode",python,selection_mouse
+94,233900,"models/lam.py",2398,0,"",python,selection_mouse
+95,234064,"models/lam.py",2393,6,"videos",python,selection_mouse
+96,234808,"models/lam.py",2384,0,"",python,selection_mouse
+97,234931,"models/lam.py",2377,9,"vq_encode",python,selection_mouse
+98,235535,"models/lam.py",2471,0,"",python,selection_mouse
+99,236162,"models/lam.py",2396,0,"",python,selection_mouse
+100,236277,"models/lam.py",2393,6,"videos",python,selection_mouse
+101,237271,"models/lam.py",2382,0,"",python,selection_mouse
+102,237422,"models/lam.py",2377,9,"vq_encode",python,selection_mouse
+103,239596,"models/lam.py",2396,0,"",python,selection_mouse
+104,239714,"models/lam.py",2393,6,"videos",python,selection_mouse
+105,243202,"models/lam.py",2460,0,"",python,selection_mouse
+106,243586,"models/lam.py",2382,0,"",python,selection_mouse
+107,243748,"models/lam.py",2377,9,"vq_encode",python,selection_mouse
+108,247136,"models/dynamics.py",0,0,"",python,tab
+109,251589,"models/dynamics.py",2886,0,"",python,selection_mouse
+110,251711,"models/dynamics.py",2883,3,"pad",python,selection_mouse
+111,252204,"models/dynamics.py",2829,0,"",python,selection_mouse
+112,252345,"models/dynamics.py",2823,9,"action_up",python,selection_mouse
+113,258257,"models/dynamics.py",3031,0,"",python,selection_mouse
+114,258490,"models/dynamics.py",3031,1,"l",python,selection_mouse
+115,258506,"models/dynamics.py",3031,3,"lf.",python,selection_mouse
+116,258553,"models/dynamics.py",3031,4,"lf.d",python,selection_mouse
+117,258581,"models/dynamics.py",3031,5,"lf.dy",python,selection_mouse
+118,258582,"models/dynamics.py",3031,6,"lf.dyn",python,selection_mouse
+119,260088,"models/dynamics.py",3040,0,"",python,selection_mouse
+120,263134,"models/dynamics.py",2987,0,"",python,selection_mouse
+121,264608,"models/dynamics.py",2830,0,"",python,selection_mouse
+122,264753,"models/dynamics.py",2823,9,"action_up",python,selection_mouse
+123,275701,"genie.py",0,0,"from typing import Dict, Any\n\nimport optax\nimport jax\nimport jax.numpy as jnp\nimport flax.linen as nn\nfrom flax.training.train_state import TrainState\nimport orbax.checkpoint as ocp\n\nfrom models.dynamics import DynamicsMaskGIT, DynamicsAutoregressive\nfrom models.lam import LatentActionModel\nfrom models.tokenizer import TokenizerVQVAE\n\nimport os\nimport grain\n\n\nclass Genie(nn.Module):\n    """"""Genie model""""""\n\n    # --- Tokenizer ---\n    in_dim: int\n    tokenizer_dim: int\n    latent_patch_dim: int\n    num_patch_latents: int\n    patch_size: int\n    tokenizer_num_blocks: int\n    tokenizer_num_heads: int\n    # --- LAM ---\n    lam_dim: int\n    latent_action_dim: int\n    num_latent_actions: int\n    lam_patch_size: int\n    lam_num_blocks: int\n    lam_num_heads: int\n    lam_co_train: bool\n    # --- Dynamics ---\n    dyna_dim: int\n    dyna_num_blocks: int\n    dyna_num_heads: int\n    use_maskgit: bool\n    param_dtype: jnp.dtype\n    dtype: jnp.dtype\n    dropout: float = 0.0\n    mask_limit: float = 0.0\n\n    def setup(self):\n        self.tokenizer = TokenizerVQVAE(\n            in_dim=self.in_dim,\n            model_dim=self.tokenizer_dim,\n            latent_dim=self.latent_patch_dim,\n            num_latents=self.num_patch_latents,\n            patch_size=self.patch_size,\n            num_blocks=self.tokenizer_num_blocks,\n            num_heads=self.tokenizer_num_heads,\n            dropout=0.0,\n            codebook_dropout=0.0,\n            param_dtype=self.param_dtype,\n            dtype=self.dtype,\n        )\n        self.lam = LatentActionModel(\n            in_dim=self.in_dim,\n            model_dim=self.lam_dim,\n            latent_dim=self.latent_patch_dim,\n            num_latents=self.num_latent_actions,\n            patch_size=self.lam_patch_size,\n            num_blocks=self.lam_num_blocks,\n            num_heads=self.lam_num_heads,\n            dropout=0.0,\n            codebook_dropout=0.0,\n            param_dtype=self.param_dtype,\n            dtype=self.dtype,\n        )\n\n        if self.use_maskgit:\n            self.dynamics = DynamicsMaskGIT(\n                model_dim=self.dyna_dim,\n                num_latents=self.num_patch_latents,\n                num_blocks=self.dyna_num_blocks,\n                num_heads=self.dyna_num_heads,\n                dropout=self.dropout,\n                mask_limit=self.mask_limit,\n                param_dtype=self.param_dtype,\n                dtype=self.dtype,\n            ) \n        else:\n            self.dynamics = DynamicsAutoregressive(\n                model_dim=self.dyna_dim,\n                num_latents=self.num_patch_latents,\n                num_blocks=self.dyna_num_blocks,\n                num_heads=self.dyna_num_heads,\n                dropout=self.dropout,\n                param_dtype=self.param_dtype,\n                dtype=self.dtype,\n            )\n\n    def __call__(self, batch: Dict[str, Any], training: bool = True) -> Dict[str, Any]:\n        tokenizer_outputs = self.tokenizer.vq_encode(batch[""videos""], training=False)\n        lam_outputs = self.lam.vq_encode(batch[""videos""], training=False)\n        latent_actions = jax.lax.cond(\n            self.lam_co_train,\n            lambda: lam_outputs[""z_q""],\n            lambda: jax.lax.stop_gradient(lam_outputs[""z_q""])\n        )\n        outputs = dict(\n            video_tokens=jax.lax.stop_gradient(tokenizer_outputs[""indices""]),\n            latent_actions=latent_actions,\n        )\n        outputs[""mask_rng""] = batch[""mask_rng""]\n        dyna_outputs = self.dynamics(outputs, training)\n        outputs.update(dyna_outputs)\n        mle_indices = jnp.argmax(outputs[""token_logits""], axis=-1)\n        outputs[""recon""] = self.tokenizer.decode(\n            mle_indices, batch[""videos""].shape[2:4]\n        )\n        outputs[""lam_indices""] = lam_outputs[""indices""]\n        return outputs\n\n\n    def sample_causal(\n        self,\n        batch: Dict[str, Any],\n        seq_len: int,\n        temperature: float = 1,\n        sample_argmax: bool = False,\n    ):\n        """"""\n        Autoregressively samples up to `seq_len` future frames using the causal transformer backend.\n\n        - Input frames are tokenized once.\n        - Future frames are generated one at a time, each conditioned on all previous frames.\n        - All frames are detokenized in a single pass at the end.\n\n        Args:\n            batch: Dict with at least ""videos"" (B, T, H, W, C)\n            seq_len: total number of frames to generate (including context)\n            temperature: sampling temperature\n            sample_argmax: if True, use argmax instead of sampling\n\n        Returns:\n            Generated video frames (B, seq_len, H, W, C)\n        """"""\n        # --- Encode context frames ---\n        tokenizer_out = self.tokenizer.vq_encode(batch[""videos""], training=False)\n        token_idxs = tokenizer_out[""indices""]  # (B, T, N)\n        B, T, N = token_idxs.shape\n\n        # jax.debug.print(""token_idxs shape: {}"", token_idxs.shape)\n        # --- Prepare initial token sequence ---\n        # Pad with zeros for future frames\n        pad_shape = (B, seq_len - T, N)\n        token_idxs_full = jnp.concatenate(\n            [token_idxs, jnp.zeros(pad_shape, dtype=token_idxs.dtype)], axis=1\n        )  # (B, seq_len, N)\n\n        # --- Prepare latent actions ---\n        action_tokens = self.lam.vq.get_codes(batch[""latent_actions""]) # (B, S-1, )\n        # --- Autoregressive generation loop ---\n        rng = batch[""rng""]\n        for t in range(T, seq_len):\n            for n in range(30):\n                dyna_inputs = {\n                    ""video_tokens"": token_idxs_full,\n                    ""latent_actions"": action_tokens\n                    }\n                # jax.debug.print(""token_idxs_full 0: {}"", token_idxs_full[0,:,0])\n                dyna_outputs = self.dynamics(dyna_inputs, training=False)\n                # # dyna_outputs[""token_logits""]: (B, t, N, vocab_size)\n                # # We want the logits for the last time step (frame t-1 predicting t)\n                jax.debug.breakpoint()\n                next_token_logits = dyna_outputs[""token_logits""][:, t, n, :].astype(jnp.float32)  # (B, 1, vocab_size)\n\n                # Sample or argmax for each patch\n                if sample_argmax:\n                    next_token = jnp.argmax(next_token_logits, axis=-1)  # (B, 1)\n                else:\n                    rng, step_rng = jax.random.split(rng)\n                    next_token = jax.random.categorical(\n                        step_rng, next_token_logits / temperature, axis=-1\n                    )  # (B, 1)\n\n                # Insert the generated tokens into the sequence\n                token_idxs_full = token_idxs_full.at[:, t, n].set(next_token)\n\n        # --- Decode all tokens at once at the end ---\n        final_frames = self.tokenizer.decode(\n            token_idxs_full, video_hw=batch[""videos""].shape[2:4]\n        )\n        return final_frames\n\n\n    @nn.compact\n    def sample_maskgit(\n        self,\n        batch: Dict[str, Any],\n        seq_len: int,\n        steps: int = 25,\n        temperature: float = 1,\n        sample_argmax: bool = False,\n    ) -> Any:\n        """"""\n        Autoregressively samples up to `seq_len` future frames, following Figure 8 of the paper.\n\n        - Input frames are tokenized once.\n        - Future frames are generated autoregressively in token space.\n        - All frames are detokenized in a single pass.\n\n        Note:\n        - For interactive or step-wise sampling, detokenization should occur after each action.\n        - To maintain consistent tensor shapes across timesteps, all current and future frames are decoded at every step.\n        - Temporal causal structure is preserved by \n            a) reapplying the mask before each decoding step.\n            b) a temporal causal mask is applied within each ST-transformer block.\n\n        Dimension keys:\n            B: batch size  \n            T: number of input (conditioning) frames  \n            N: patches per frame  \n            S: sequence length  \n            A: action space  \n            D: model latent dimension\n        """"""\n        # --- Encode videos and actions ---\n        tokenizer_out = self.tokenizer.vq_encode(batch[""videos""], training=False)\n        token_idxs = tokenizer_out[""indices""] # (B, T, N)\n        B, T, N = token_idxs.shape\n        pad_shape = (B, seq_len - T, N)\n        pad = jnp.zeros(pad_shape, dtype=token_idxs.dtype)\n        token_idxs = jnp.concatenate([token_idxs, pad], axis=1) # (B, S, N)\n        action_tokens = self.lam.vq.get_codes(batch[""latent_actions""]) \n\n        MaskGITLoop = nn.scan(\n            MaskGITStep,\n            variable_broadcast=""params"",\n            split_rngs={""params"": False},\n            in_axes=0,\n            out_axes=0,\n            length=steps,\n        )\n    \n        loop_fn = MaskGITLoop(\n            dynamics=self.dynamics,\n            tokenizer=self.tokenizer,\n            temperature=temperature,\n            sample_argmax=sample_argmax,\n            steps=steps,\n        )\n\n        def generation_step_fn(carry, step_t):\n            rng, current_token_idxs = carry\n            rng, step_rng = jax.random.split(rng)\n\n            # Mask current and future frames (i.e., t >= step_t)\n            mask = jnp.arange(seq_len) >= step_t # (S,)\n            mask = jnp.broadcast_to(mask[None, :, None], (B, seq_len, N)) # (B, S, N)\n            mask = mask.astype(bool)\n            masked_token_idxs = current_token_idxs * ~mask\n\n            # --- Initialize and run MaskGIT loop ---\n            init_carry_maskgit = (\n                step_rng,\n                masked_token_idxs,\n                mask,\n                action_tokens,\n            )\n            final_carry_maskgit, _ = loop_fn(init_carry_maskgit, jnp.arange(steps))\n            updated_token_idxs = final_carry_maskgit[1]\n            new_carry = (rng, updated_token_idxs)\n            return new_carry, None\n\n        # --- Run the autoregressive generation using scan ---\n        initial_carry = (batch[""rng""], token_idxs)\n        timesteps_to_scan = jnp.arange(T, seq_len)\n        final_carry, _ = jax.lax.scan(\n            generation_step_fn,\n            initial_carry,\n            timesteps_to_scan\n        )\n        final_token_idxs = final_carry[1]\n\n        # --- Decode all tokens at once at the end ---\n        final_frames = self.tokenizer.decode(\n            final_token_idxs,\n            video_hw=batch[""videos""].shape[2:4],\n        )\n        return final_frames\n\n    def vq_encode(self, batch, training) -> Dict[str, Any]:\n        # --- Preprocess videos ---\n        lam_output = self.lam.vq_encode(batch[""videos""], training=training)\n        return lam_output[""indices""]\n\n\nclass MaskGITStep(nn.Module):\n    dynamics: nn.Module\n    tokenizer: nn.Module\n    temperature: float\n    sample_argmax: bool\n    steps: int\n\n    @nn.compact\n    def __call__(self, carry, x):\n        rng, token_idxs, mask, action_tokens = carry\n        step = x\n        N = token_idxs.shape[2]\n\n        # --- Construct + encode video ---\n        vid_embed = self.dynamics.patch_embed(token_idxs) # (B, S, N, D)\n        mask_token = self.dynamics.mask_token  # (1, 1, 1, D,)\n        mask_expanded = mask[..., None] # (B, S, N, 1) \n        vid_embed = jnp.where(mask_expanded, mask_token, vid_embed)\n\n        # --- Predict transition ---\n        act_embed = self.dynamics.action_up(action_tokens)\n        vid_embed += jnp.pad(act_embed, ((0, 0), (1, 0), (0, 0), (0, 0)))\n        unmasked_ratio = jnp.cos(jnp.pi * (step + 1) / (self.steps * 2))\n        step_temp = self.temperature * (1.0 - unmasked_ratio)\n        final_logits = self.dynamics.dynamics(vid_embed) / step_temp\n\n        # --- Sample new tokens for final frame ---\n        if self.sample_argmax:\n            sampled_token_idxs = jnp.argmax(final_logits, axis=-1)\n        else:\n            rng, _rng = jax.random.split(rng)\n            sampled_token_idxs = jax.random.categorical(_rng, final_logits)\n        gather_fn = jax.vmap(jax.vmap(jax.vmap(lambda x, y: x[y])))\n        final_token_probs = gather_fn(jax.nn.softmax(final_logits), sampled_token_idxs)\n        final_token_probs += ~mask\n        # Update masked tokens only\n        token_idxs = jnp.where(mask, sampled_token_idxs, token_idxs)\n\n        # --- Update mask ---\n        num_unmasked_tokens = jnp.round(N * (1.0 - unmasked_ratio)).astype(int)\n        idx_mask = jnp.arange(final_token_probs.shape[-1]) > num_unmasked_tokens\n        sorted_idxs = jnp.argsort(final_token_probs, axis=-1, descending=True)\n        mask_update_fn = jax.vmap(lambda msk, ids: msk.at[ids].set(idx_mask))\n        new_mask = mask_update_fn(mask, sorted_idxs)\n\n        new_carry = (rng, token_idxs, new_mask, action_tokens)\n        return new_carry, None\n\ndef restore_genie_components(\n    train_state: TrainState,\n    sharding: jax.sharding.NamedSharding,\n    grain_iterator: grain.DataLoaderIterator,\n    inputs: Dict[str, jax.Array],\n    rng: jax.Array,\n    args,\n):\n    """"""Restore pre-trained Genie components""""""\n    rng, _rng = jax.random.split(rng)\n\n    # dummy values since we only use tx to initialize the dummy train states\n    dummy_tx = optax.adamw(\n        learning_rate=optax.constant_schedule(args.max_lr),\n        b1=0.9,\n        b2=0.9,\n        weight_decay=1e-4,\n    )\n    handler_registry = ocp.handlers.DefaultCheckpointHandlerRegistry()\n    handler_registry.add('model_state', ocp.args.StandardRestore, ocp.handlers.StandardCheckpointHandler)\n    handler_registry.add('dataloader_state', grain.checkpoint.CheckpointRestore, grain.checkpoint.CheckpointHandler)\n    \n\n    checkpoint_options = ocp.CheckpointManagerOptions(\n        step_format_fixed_length=6,\n    )\n    tokenizer_checkpoint_manager = ocp.CheckpointManager(\n        directory=args.tokenizer_checkpoint,\n        options=checkpoint_options,\n        handler_registry=handler_registry,\n    )\n    dummy_tokenizer = TokenizerVQVAE(\n        in_dim=args.image_channels,\n        model_dim=args.tokenizer_dim,\n        latent_dim=args.latent_patch_dim,\n        num_latents=args.num_patch_latents,\n        patch_size=args.patch_size,\n        num_blocks=args.tokenizer_num_blocks,\n        num_heads=args.tokenizer_num_heads,\n        dropout=args.dropout,\n        codebook_dropout=args.dropout,\n        param_dtype=args.param_dtype,\n        dtype=args.dtype,\n    )\n    tokenizer_init_params = dummy_tokenizer.init(_rng, inputs)\n    dummy_tokenizer_train_state = TrainState.create(\n        apply_fn=dummy_tokenizer.apply, params=tokenizer_init_params, tx=dummy_tx\n    )\n    abstract_sharded_tokenizer_state = _create_abstract_sharded_pytree(\n        dummy_tokenizer_train_state, sharding\n    )\n    restored_tokenizer = tokenizer_checkpoint_manager.restore(\n        step=tokenizer_checkpoint_manager.latest_step(),\n        args=ocp.args.Composite(\n            model_state=ocp.args.StandardRestore(abstract_sharded_tokenizer_state),\n            dataloader_state=grain.checkpoint.CheckpointRestore(grain_iterator),\n        ),\n    )[""model_state""]\n    restored_tokenizer_params = restored_tokenizer.params[""params""]\n    train_state.params[""params""][""tokenizer""].update(restored_tokenizer_params)\n    tokenizer_checkpoint_manager.close()\n\n    if args.lam_checkpoint:\n        lam_checkpoint_manager = ocp.CheckpointManager(\n            directory=args.lam_checkpoint,\n            options=checkpoint_options,\n            handler_registry=handler_registry,\n        )\n        dummy_lam = LatentActionModel(\n            in_dim=args.image_channels,\n            model_dim=args.lam_dim,\n            latent_dim=args.latent_patch_dim,\n            num_latents=args.num_latent_actions,\n            patch_size=args.lam_patch_size,\n            num_blocks=args.lam_num_blocks,\n            num_heads=args.lam_num_heads,\n            dropout=args.dropout,\n            codebook_dropout=args.dropout,\n            param_dtype=args.param_dtype,\n            dtype=args.dtype,\n        )\n        lam_init_params = dummy_lam.init(_rng, inputs)\n        dummy_lam_train_state = TrainState.create(\n            apply_fn=dummy_lam.apply, params=lam_init_params, tx=dummy_tx\n        )\n        abstract_sharded_lam_state = _create_abstract_sharded_pytree(\n            dummy_lam_train_state, sharding\n        )\n        restored_lam = lam_checkpoint_manager.restore(\n            step=lam_checkpoint_manager.latest_step(),\n            args=ocp.args.Composite(\n                model_state=ocp.args.StandardRestore(abstract_sharded_lam_state),\n                dataloader_state=grain.checkpoint.CheckpointRestore(grain_iterator),\n            ),\n        )[""model_state""]\n        restored_lam_params = restored_lam.params[""params""]\n        # Genie does not initialize all LAM modules, thus we omit those extra modules during restoration\n        # (f.srambical) FIXME: Currently, this is a small HBM memory crunch since the LAM's decoder is loaded into HBM and immediately dicarded.\n        # A workaround would be to restore to host memory first, and only move the weights to HBM after pruning the decoder\n        restored_lam_params = {\n            k: v\n            for k, v in restored_lam_params.items()\n            if k in train_state.params[""params""][""lam""]\n        }\n        train_state.params[""params""][""lam""].update(restored_lam_params)\n        lam_checkpoint_manager.close()\n\n    return train_state\n\ndef _create_abstract_sharded_pytree(pytree_template, sharding_spec):\n    """"""Replaces arrays in a pytree with ShapeDtypeStructs having the given sharding.""""""\n\n    def map_fn(leaf_template):\n        if hasattr(leaf_template, ""shape"") and hasattr(leaf_template, ""dtype""):\n            return jax.ShapeDtypeStruct(\n                leaf_template.shape, leaf_template.dtype, sharding=sharding_spec\n            )\n        return leaf_template\n\n    return jax.tree_util.tree_map(map_fn, pytree_template)",python,tab
+124,307289,"genie.py",3010,0,"",python,selection_mouse
+125,307857,"genie.py",2931,0,"",python,selection_mouse
+126,308409,"genie.py",3020,0,"",python,selection_mouse
+127,308566,"genie.py",3017,9,"vq_encode",python,selection_mouse
+128,409573,"genie.py",3604,0,"",python,selection_mouse
+129,409576,"genie.py",3603,0,"",python,selection_command
+130,410046,"genie.py",3772,0,"",python,selection_mouse
+131,410054,"genie.py",3771,0,"",python,selection_command
+132,410559,"genie.py",3797,0,"",python,selection_mouse
+133,411067,"genie.py",3772,0,"",python,selection_mouse
+134,411080,"genie.py",3771,0,"",python,selection_command
+135,412609,"TERMINAL",0,0,"bash",,terminal_focus
+136,464986,"models/dynamics.py",0,0,"",python,tab
+137,469860,"models/lam.py",0,0,"",python,tab
+138,473097,"models/lam.py",2697,0,"",python,selection_mouse
+139,473666,"models/lam.py",2613,0,"",python,selection_mouse
+140,475909,"models/lam.py",2654,0,"\n        ",python,content
+141,476310,"models/lam.py",2663,0,"#",python,content
+142,476311,"models/lam.py",2664,0,"",python,selection_keyboard
+143,476601,"models/lam.py",2664,0," ",python,content
+144,476602,"models/lam.py",2665,0,"",python,selection_keyboard
+145,477268,"models/lam.py",2665,0,"D",python,content
+146,477269,"models/lam.py",2666,0,"",python,selection_keyboard
+147,477371,"models/lam.py",2666,0,"O",python,content
+148,477373,"models/lam.py",2667,0,"",python,selection_keyboard
+149,477751,"models/lam.py",2667,0," ",python,content
+150,477752,"models/lam.py",2668,0,"",python,selection_keyboard
+151,478055,"models/lam.py",2667,1,"",python,content
+152,478165,"models/lam.py",2666,1,"",python,content
+153,478312,"models/lam.py",2665,1,"",python,content
+154,479439,"models/lam.py",2665,0,"F",python,content
+155,479440,"models/lam.py",2666,0,"",python,selection_keyboard
+156,479576,"models/lam.py",2666,0,"I",python,content
+157,479577,"models/lam.py",2667,0,"",python,selection_keyboard
+158,479721,"models/lam.py",2667,0,"X",python,content
+159,479722,"models/lam.py",2668,0,"",python,selection_keyboard
+160,480012,"models/lam.py",2668,0,"M",python,content
+161,480013,"models/lam.py",2669,0,"",python,selection_keyboard
+162,480115,"models/lam.py",2669,0,"E",python,content
+163,480116,"models/lam.py",2670,0,"",python,selection_keyboard
+164,480284,"models/lam.py",2670,0," ",python,content
+165,480285,"models/lam.py",2671,0,"",python,selection_keyboard
+166,480618,"models/lam.py",2671,0,"M",python,content
+167,480619,"models/lam.py",2672,0,"",python,selection_keyboard
+168,480801,"models/lam.py",2672,0,"I",python,content
+169,480802,"models/lam.py",2673,0,"",python,selection_keyboard
+170,481171,"models/lam.py",2672,1,"",python,content
+171,481546,"models/lam.py",2672,0,"i",python,content
+172,481547,"models/lam.py",2673,0,"",python,selection_keyboard
+173,481820,"models/lam.py",2672,1,"",python,content
+174,482134,"models/lam.py",2671,1,"",python,content
+175,482288,"models/lam.py",2671,0,"m",python,content
+176,482289,"models/lam.py",2672,0,"",python,selection_keyboard
+177,482455,"models/lam.py",2672,0,"i",python,content
+178,482456,"models/lam.py",2673,0,"",python,selection_keyboard
+179,482555,"models/lam.py",2673,0,"h",python,content
+180,482556,"models/lam.py",2674,0,"",python,selection_keyboard
+181,482649,"models/lam.py",2674,0,"i",python,content
+182,482650,"models/lam.py",2675,0,"",python,selection_keyboard
+183,482729,"models/lam.py",2675,0,"r",python,content
+184,482730,"models/lam.py",2676,0,"",python,selection_keyboard
+185,482867,"models/lam.py",2676,0," ",python,content
+186,482868,"models/lam.py",2677,0,"",python,selection_keyboard
+187,483095,"models/lam.py",2677,0,"d",python,content
+188,483096,"models/lam.py",2678,0,"",python,selection_keyboard
+189,483251,"models/lam.py",2678,0,"o",python,content
+190,483252,"models/lam.py",2679,0,"",python,selection_keyboard
+191,483382,"models/lam.py",2679,0," ",python,content
+192,483383,"models/lam.py",2680,0,"",python,selection_keyboard
+193,483533,"models/lam.py",2680,0,"t",python,content
+194,483534,"models/lam.py",2681,0,"",python,selection_keyboard
+195,483598,"models/lam.py",2681,0,"h",python,content
+196,483599,"models/lam.py",2682,0,"",python,selection_keyboard
+197,483724,"models/lam.py",2682,0,"i",python,content
+198,483725,"models/lam.py",2683,0,"",python,selection_keyboard
+199,483772,"models/lam.py",2683,0,"s",python,content
+200,483773,"models/lam.py",2684,0,"",python,selection_keyboard
+201,483880,"models/lam.py",2684,0," ",python,content
+202,483881,"models/lam.py",2685,0,"",python,selection_keyboard
+203,484433,"models/lam.py",2685,0,"t",python,content
+204,484434,"models/lam.py",2686,0,"",python,selection_keyboard
+205,484520,"models/lam.py",2686,0,"h",python,content
+206,484521,"models/lam.py",2687,0,"",python,selection_keyboard
+207,484628,"models/lam.py",2687,0,"e",python,content
+208,484629,"models/lam.py",2688,0,"",python,selection_keyboard
+209,484715,"models/lam.py",2688,0," ",python,content
+210,484716,"models/lam.py",2689,0,"",python,selection_keyboard
+211,484805,"models/lam.py",2689,0,"o",python,content
+212,484806,"models/lam.py",2690,0,"",python,selection_keyboard
+213,484879,"models/lam.py",2690,0,"t",python,content
+214,484880,"models/lam.py",2691,0,"",python,selection_keyboard
+215,485012,"models/lam.py",2691,0,"h",python,content
+216,485012,"models/lam.py",2692,0,"",python,selection_keyboard
+217,485111,"models/lam.py",2692,0,"e",python,content
+218,485112,"models/lam.py",2693,0,"",python,selection_keyboard
+219,485171,"models/lam.py",2693,0,"r",python,content
+220,485173,"models/lam.py",2694,0,"",python,selection_keyboard
+221,485283,"models/lam.py",2694,0," ",python,content
+222,485284,"models/lam.py",2695,0,"",python,selection_keyboard
+223,485667,"models/lam.py",2695,0,"w",python,content
+224,485668,"models/lam.py",2696,0,"",python,selection_keyboard
+225,485881,"models/lam.py",2696,0,"a",python,content
+226,485882,"models/lam.py",2697,0,"",python,selection_keyboard
+227,486046,"models/lam.py",2697,0,"y",python,content
+228,486047,"models/lam.py",2698,0,"",python,selection_keyboard
+229,486139,"models/lam.py",2698,0," ",python,content
+230,486140,"models/lam.py",2699,0,"",python,selection_keyboard
+231,486310,"models/lam.py",2699,0,"a",python,content
+232,486311,"models/lam.py",2700,0,"",python,selection_keyboard
+233,486497,"models/lam.py",2700,0,"r",python,content
+234,486498,"models/lam.py",2701,0,"",python,selection_keyboard
+235,486789,"models/lam.py",2701,0,"o",python,content
+236,486789,"models/lam.py",2702,0,"",python,selection_keyboard
+237,486986,"models/lam.py",2702,0,"u",python,content
+238,486987,"models/lam.py",2703,0,"",python,selection_keyboard
+239,487032,"models/lam.py",2703,0,"n",python,content
+240,487034,"models/lam.py",2704,0,"",python,selection_keyboard
+241,487122,"models/lam.py",2704,0,"d",python,content
+242,487123,"models/lam.py",2705,0,"",python,selection_keyboard
+243,487824,"models/lam.py",2704,0,"",python,selection_command
+244,555157,"models/lam.py",3222,0,"",python,selection_mouse
+245,562923,"models/lam.py",2949,0,"",python,selection_mouse
+246,562925,"models/lam.py",2948,0,"",python,selection_command
+247,563531,"models/lam.py",2950,0,"",python,selection_mouse
+248,564154,"models/lam.py",2934,16,"  # (B, T-1, E)\n",python,selection_mouse
+249,564172,"models/lam.py",2933,17,"]  # (B, T-1, E)\n",python,selection_mouse
+250,564185,"models/lam.py",2932,18,"0]  # (B, T-1, E)\n",python,selection_mouse
+251,564200,"models/lam.py",2931,19," 0]  # (B, T-1, E)\n",python,selection_mouse
+252,564218,"models/lam.py",2930,20,", 0]  # (B, T-1, E)\n",python,selection_mouse
+253,564246,"models/lam.py",2879,71,"nt action for all future frames\n        z = z[:, 1:, 0]  # (B, T-1, E)\n",python,selection_mouse
+254,564264,"models/lam.py",2878,72,"ent action for all future frames\n        z = z[:, 1:, 0]  # (B, T-1, E)\n",python,selection_mouse
+255,564309,"models/lam.py",2877,73,"tent action for all future frames\n        z = z[:, 1:, 0]  # (B, T-1, E)\n",python,selection_mouse
+256,564354,"models/lam.py",2876,74,"atent action for all future frames\n        z = z[:, 1:, 0]  # (B, T-1, E)\n",python,selection_mouse
+257,564355,"models/lam.py",2875,75,"latent action for all future frames\n        z = z[:, 1:, 0]  # (B, T-1, E)\n",python,selection_mouse
+258,564355,"models/lam.py",2874,76," latent action for all future frames\n        z = z[:, 1:, 0]  # (B, T-1, E)\n",python,selection_mouse
+259,564356,"models/lam.py",2873,77,"t latent action for all future frames\n        z = z[:, 1:, 0]  # (B, T-1, E)\n",python,selection_mouse
+260,564398,"models/lam.py",2872,78,"et latent action for all future frames\n        z = z[:, 1:, 0]  # (B, T-1, E)\n",python,selection_mouse
+261,564399,"models/lam.py",2871,79,"Get latent action for all future frames\n        z = z[:, 1:, 0]  # (B, T-1, E)\n",python,selection_mouse
+262,564444,"models/lam.py",2870,80," Get latent action for all future frames\n        z = z[:, 1:, 0]  # (B, T-1, E)\n",python,selection_mouse
+263,564450,"models/lam.py",2920,30," = z[:, 1:, 0]  # (B, T-1, E)\n",python,selection_mouse
+264,564490,"models/lam.py",2919,31,"z = z[:, 1:, 0]  # (B, T-1, E)\n",python,selection_mouse
+265,564975,"models/lam.py",2919,0,"",python,selection_mouse
+266,567444,"models/lam.py",2736,0,"",python,selection_mouse
+267,568071,"models/lam.py",2739,0,"",python,selection_mouse
+268,568203,"models/lam.py",2735,11,"concatenate",python,selection_mouse
+269,568809,"models/lam.py",2753,0,"",python,selection_mouse
+270,569464,"models/lam.py",2677,0,"",python,selection_mouse
+271,569624,"models/lam.py",2677,2,"do",python,selection_mouse
+272,569807,"models/lam.py",2677,7,"do this",python,selection_mouse
+273,569847,"models/lam.py",2677,11,"do this the",python,selection_mouse
+274,569851,"models/lam.py",2677,17,"do this the other",python,selection_mouse
+275,569893,"models/lam.py",2677,18,"do this the other ",python,selection_mouse
+276,569894,"models/lam.py",2677,21,"do this the other way",python,selection_mouse
+277,569902,"models/lam.py",2677,28,"do this the other way around",python,selection_mouse
+278,570192,"models/lam.py",2705,0,"",python,selection_mouse
+279,570199,"models/lam.py",2704,0,"",python,selection_command
+280,570389,"models/lam.py",2699,6,"around",python,selection_mouse
+281,570435,"models/lam.py",2700,5,"round",python,selection_command
+282,570600,"models/lam.py",2700,58,"round\n        padded_patches = jnp.concatenate((action_pad",python,selection_mouse
+283,570641,"models/lam.py",2700,46,"round\n        padded_patches = jnp.concatenate",python,selection_mouse
+284,570690,"models/lam.py",2680,20,"this the other way a",python,selection_mouse
+285,570734,"models/lam.py",2679,21," this the other way a",python,selection_mouse
+286,570735,"models/lam.py",2677,23,"do this the other way a",python,selection_mouse
+287,570735,"models/lam.py",2676,24," do this the other way a",python,selection_mouse
+288,570777,"models/lam.py",2671,29,"mihir do this the other way a",python,selection_mouse
+289,570782,"models/lam.py",2700,28,"round\n        padded_patches",python,selection_mouse
+290,571048,"models/lam.py",2700,14,"round\n        ",python,selection_mouse
+291,571160,"models/lam.py",2663,37,"# FIXME mihir do this the other way a",python,selection_mouse
+292,571227,"models/lam.py",2664,36," FIXME mihir do this the other way a",python,selection_mouse
+293,571574,"models/lam.py",2664,0,"",python,selection_mouse
+294,571574,"models/lam.py",2664,1," ",python,selection_mouse
+295,571784,"models/lam.py",2664,6," FIXME",python,selection_mouse
+296,571784,"models/lam.py",2664,12," FIXME mihir",python,selection_mouse
+297,571816,"models/lam.py",2664,15," FIXME mihir do",python,selection_mouse
+298,571834,"models/lam.py",2664,20," FIXME mihir do this",python,selection_mouse
+299,571865,"models/lam.py",2664,21," FIXME mihir do this ",python,selection_mouse
+300,571868,"models/lam.py",2664,24," FIXME mihir do this the",python,selection_mouse
+301,571911,"models/lam.py",2664,25," FIXME mihir do this the ",python,selection_mouse
+302,571912,"models/lam.py",2664,30," FIXME mihir do this the other",python,selection_mouse
+303,571918,"models/lam.py",2606,59,"self.action_in, (B, T, 1, self.patch_token_dim))\n        # ",python,selection_mouse
+304,572011,"models/lam.py",2610,55,".action_in, (B, T, 1, self.patch_token_dim))\n        # ",python,selection_mouse
+305,572012,"models/lam.py",2611,54,"action_in, (B, T, 1, self.patch_token_dim))\n        # ",python,selection_mouse
+306,572664,"models/lam.py",2619,0,"",python,selection_mouse
+307,573004,"models/lam.py",2705,0,"",python,selection_mouse
+308,573005,"models/lam.py",2704,0,"",python,selection_command
+309,573166,"models/lam.py",2704,1,"d",python,selection_mouse
+310,573167,"models/lam.py",2702,2,"un",python,selection_mouse
+311,573211,"models/lam.py",2705,0,"",python,selection_command
+312,573213,"models/lam.py",2698,7," around",python,selection_mouse
+313,573213,"models/lam.py",2689,16,"other way around",python,selection_mouse
+314,573223,"models/lam.py",2681,24,"his the other way around",python,selection_mouse
+315,573235,"models/lam.py",2677,28,"do this the other way around",python,selection_mouse
+316,573262,"models/lam.py",2675,30,"r do this the other way around",python,selection_mouse
+317,573268,"models/lam.py",2672,33,"ihir do this the other way around",python,selection_mouse
+318,573286,"models/lam.py",2671,34,"mihir do this the other way around",python,selection_mouse
+319,573307,"models/lam.py",2670,35," mihir do this the other way around",python,selection_mouse
+320,573350,"models/lam.py",2669,36,"E mihir do this the other way around",python,selection_mouse
+321,573437,"models/lam.py",2668,37,"ME mihir do this the other way around",python,selection_mouse
+322,573449,"models/lam.py",2667,38,"XME mihir do this the other way around",python,selection_mouse
+323,573483,"models/lam.py",2666,39,"IXME mihir do this the other way around",python,selection_mouse
+324,573529,"models/lam.py",2665,40,"FIXME mihir do this the other way around",python,selection_mouse
+325,573571,"models/lam.py",2664,41," FIXME mihir do this the other way around",python,selection_mouse
+326,573660,"models/lam.py",2663,42,"# FIXME mihir do this the other way around",python,selection_mouse
+327,574100,"models/lam.py",2663,0,"",python,selection_mouse
+328,576800,"/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jafar/train_dynamics.py",0,0,"from dataclasses import dataclass, field\nimport os\n\nimport einops\nfrom flax.training.train_state import TrainState\nfrom jax.sharding import Mesh, PartitionSpec, NamedSharding\nfrom jax.experimental.mesh_utils import create_device_mesh\nimport optax\nimport orbax.checkpoint as ocp\nimport numpy as np\nimport dm_pix as pix\nimport jax\nimport jax.numpy as jnp\nimport tyro\nimport wandb\nimport grain\n\nfrom genie import Genie, restore_genie_components\nfrom utils.dataloader import get_dataloader\nfrom utils.lr_utils import get_lr_schedule\nfrom utils.parameter_utils import count_parameters_by_component\n\n@dataclass\nclass Args:\n    # Experiment\n    num_steps: int = 200_000\n    seed: int = 0\n    seq_len: int = 16\n    image_channels: int = 3\n    image_height: int = 90\n    image_width: int = 160\n    data_dir: str = """"\n    save_ckpt: bool = False\n    restore_ckpt: bool = False\n    # Optimization\n    batch_size: int = 36\n    init_lr: float = 0.0\n    max_lr: float = 3e-5\n    decay_end: float = 0.0\n    wsd_decay_steps: int = 10000 # NOTE: wsd_decay_steps will only be used when using a wsd-schedule\n    warmup_steps: int = 5000\n    lr_schedule : str = ""wsd"" # supported options: wsd, cos\n    # Tokenizer\n    tokenizer_dim: int = 512\n    latent_patch_dim: int = 32\n    num_patch_latents: int = 1024\n    patch_size: int = 4\n    tokenizer_num_blocks: int = 8\n    tokenizer_num_heads: int = 8\n    tokenizer_checkpoint: str = """"\n    # LAM\n    lam_dim: int = 512\n    latent_action_dim: int = 32\n    num_latent_actions: int = 6\n    lam_patch_size: int = 16\n    lam_num_blocks: int = 8\n    lam_num_heads: int = 8\n    lam_checkpoint: str = """"\n    # Dynamics\n    dyna_dim: int = 512\n    dyna_num_blocks: int = 12\n    dyna_num_heads: int = 8\n    dropout: float = 0.0\n    mask_limit: float = 0.5\n    use_maskgit: bool = False\n    param_dtype: jnp.dtype = jnp.float32\n    dtype: jnp.dtype = jnp.bfloat16\n    # Logging\n    log: bool = False\n    entity: str = """"\n    project: str = """"\n    name: str = ""train_dynamics""\n    tags: list[str] = field(default_factory=lambda: [""dynamics""])\n    log_interval: int = 5\n    log_image_interval: int = 250\n    ckpt_dir: str = """"\n    log_checkpoint_interval: int = 25000\n    log_checkpoint_keep_period: int = 20000\n    log_gradients: bool = False\n    wandb_id: str = """"\n\n\nargs = tyro.cli(Args)\n\n\ndef dynamics_loss_fn(params, state, inputs):\n    """"""Compute masked dynamics loss""""""\n    inputs[""videos""] = inputs[""videos""].astype(args.dtype) / 255.0\n    outputs = state.apply_fn(\n        params,\n        inputs,\n        training=True,\n        rngs={""params"": inputs[""rng""], ""dropout"": inputs[""dropout_rng""]},\n    )\n    mask = outputs[""mask""]\n    outputs[""token_logits""] = outputs[""token_logits""].astype(jnp.float32)\n    logits = outputs[""token_logits""]\n    targets = outputs[""video_tokens""]\n\n    # if not args.use_maskgit:\n        # logits = outputs[""token_logits""][:, :, :-1]\n        # targets = outputs[""video_tokens""][:, :, 1:]\n        # mask = outputs[""mask""][:, :, 1:] \n    ce_loss = optax.softmax_cross_entropy_with_integer_labels(\n        logits, targets\n    )\n    ce_loss = (mask * ce_loss).sum() / mask.sum()\n    acc = logits.argmax(-1) == targets\n    acc = (mask * acc).sum() / mask.sum()\n    select_probs = jax.nn.softmax(logits)\n    gt = inputs[""videos""].clip(0, 1).reshape(-1, *inputs[""videos""].shape[2:])\n    recon = outputs[""recon""].clip(0, 1).reshape(-1, *outputs[""recon""].shape[2:])\n    psnr = pix.psnr(gt, recon).mean() # type: ignore\n    ssim = pix.ssim(gt, recon).mean() # type: ignore\n    _, index_counts_lam = jnp.unique_counts(\n        jnp.ravel(outputs[""lam_indices""]), size=args.num_latent_actions, fill_value=0\n    )\n    _, index_counts_tokenizer = jnp.unique_counts(\n        jnp.ravel(outputs[""video_tokens""]), size=args.num_patch_latents, fill_value=0\n    )\n    codebook_usage_lam = (index_counts_lam != 0).mean()\n    codebook_usage_tokenizer = (index_counts_tokenizer != 0).mean()\n    metrics = dict(\n        cross_entropy_loss=ce_loss,\n        masked_token_accuracy=acc,\n        select_logit=logits.max(-1).mean(),\n        select_p=select_probs.max(-1).mean(),\n        entropy=jax.scipy.special.entr(select_probs).sum(-1).mean(),\n        psnr=psnr,\n        ssim=ssim,\n        codebook_usage_lam=codebook_usage_lam,\n        codebook_usage_tokenizer=codebook_usage_tokenizer,\n    )\n    return ce_loss, (outputs[""recon""], metrics)\n\n\n@jax.jit\ndef train_step(state, inputs):\n    """"""Update state and compute metrics""""""\n    grad_fn = jax.value_and_grad(dynamics_loss_fn, has_aux=True, allow_int=True)\n    (loss, (recon, metrics)), grads = grad_fn(state.params, state, inputs)\n    state = state.apply_gradients(grads=grads)\n    if args.log_gradients:\n        metrics[""gradients_std/""] = jax.tree.map(\n            lambda x: x.std(), grads[""params""][""dynamics""]\n        )\n    return state, loss, recon, metrics\n\n\nif __name__ == ""__main__"":\n    jax.distributed.initialize()\n    num_devices = jax.device_count()\n    if num_devices == 0:\n        raise ValueError(""No JAX devices found."")\n    print(f""Running on {num_devices} devices."")\n\n    if args.batch_size % num_devices != 0:\n        raise ValueError(\n            f""Global batch size {args.batch_size} must be divisible by ""\n            f""number of devices {num_devices}.""\n        )\n\n    per_device_batch_size_for_init = args.batch_size // num_devices\n\n    rng = jax.random.PRNGKey(args.seed)\n\n    # --- Initialize model ---\n    genie = Genie(\n        # Tokenizer\n        in_dim=args.image_channels,\n        tokenizer_dim=args.tokenizer_dim,\n        latent_patch_dim=args.latent_patch_dim,\n        num_patch_latents=args.num_patch_latents,\n        patch_size=args.patch_size,\n        tokenizer_num_blocks=args.tokenizer_num_blocks,\n        tokenizer_num_heads=args.tokenizer_num_heads,\n        # LAM\n        lam_dim=args.lam_dim,\n        latent_action_dim=args.latent_action_dim,\n        num_latent_actions=args.num_latent_actions,\n        lam_patch_size=args.lam_patch_size,\n        lam_num_blocks=args.lam_num_blocks,\n        lam_num_heads=args.lam_num_heads,\n        lam_co_train=not args.lam_checkpoint,\n        # Dynamics\n        dyna_dim=args.dyna_dim,\n        dyna_num_blocks=args.dyna_num_blocks,\n        dyna_num_heads=args.dyna_num_heads,\n        dropout=args.dropout,\n        mask_limit=args.mask_limit,\n        use_maskgit=args.use_maskgit,\n        param_dtype=args.param_dtype,\n        dtype=args.dtype,\n    )\n    rng, _rng = jax.random.split(rng)\n    image_shape = (args.image_height, args.image_width, args.image_channels)\n    dummy_inputs = dict(\n        videos=jnp.zeros(\n            (per_device_batch_size_for_init, args.seq_len, *image_shape),\n            dtype=args.dtype,\n        ),\n        action=jnp.zeros(\n            (per_device_batch_size_for_init, args.seq_len), dtype=args.dtype\n        ),\n        mask_rng=_rng,\n    )\n    rng, _rng = jax.random.split(rng)\n    init_params = genie.init(_rng, dummy_inputs)\n\n    param_counts = count_parameters_by_component(init_params)\n\n    if args.log and jax.process_index() == 0:\n        wandb_init_kwargs = {\n            ""entity"": args.entity,\n            ""project"": args.project,\n            ""name"": args.name,\n            ""tags"": args.tags,\n            ""group"": ""debug"",\n            ""config"": args,\n        }\n\n        if args.wandb_id:\n            wandb_init_kwargs.update(\n                {\n                    ""id"": args.wandb_id,\n                    ""resume"": ""allow"",\n                }\n            )\n        wandb.init(**wandb_init_kwargs)\n\n        wandb.config.update({""model_param_count"": param_counts})\n\n    print(""Parameter counts:"")\n    print(param_counts)\n\n    # --- Initialize optimizer ---\n    lr_schedule = get_lr_schedule(args.lr_schedule, \n                                  args.init_lr, \n                                  args.max_lr, \n                                  args.decay_end, \n                                  args.num_steps, \n                                  args.warmup_steps, \n                                  args.wsd_decay_steps)\n    tx = optax.adamw(learning_rate=lr_schedule, b1=0.9, b2=0.9, weight_decay=1e-4, mu_dtype=args.dtype)\n    train_state = TrainState.create(apply_fn=genie.apply, params=init_params, tx=tx)\n\n    device_mesh_arr = create_device_mesh((num_devices,))\n    mesh = Mesh(devices=device_mesh_arr, axis_names=(""data"",))\n\n    replicated_sharding = NamedSharding(mesh, PartitionSpec())\n    videos_sharding = NamedSharding(mesh, PartitionSpec(""data"", None, None, None, None))\n    train_state = jax.device_put(train_state, replicated_sharding)\n\n    # --- Initialize checkpoint manager ---\n    step = 0\n    handler_registry = ocp.handlers.DefaultCheckpointHandlerRegistry()\n    handler_registry.add(\n        ""model_state"", ocp.args.StandardSave, ocp.handlers.StandardCheckpointHandler\n    )\n    handler_registry.add(\n        ""model_state"", ocp.args.StandardRestore, ocp.handlers.StandardCheckpointHandler\n    )\n    handler_registry.add(""dataloader_state"", grain.checkpoint.CheckpointSave, grain.checkpoint.CheckpointHandler)  # type: ignore\n    handler_registry.add(""dataloader_state"", grain.checkpoint.CheckpointRestore, grain.checkpoint.CheckpointHandler)  # type: ignore\n\n    checkpoint_options = ocp.CheckpointManagerOptions(\n        save_interval_steps=args.log_checkpoint_interval,\n        max_to_keep=3,\n        keep_period=args.log_checkpoint_keep_period,\n        step_format_fixed_length=6,\n        cleanup_tmp_directories=True,\n    )\n\n    checkpoint_manager = ocp.CheckpointManager(\n        args.ckpt_dir,\n        options=checkpoint_options,\n        handler_registry=handler_registry,\n    )\n\n    # --- Create DataLoaderIterator from dataloader ---\n    array_record_files = [\n        os.path.join(args.data_dir, x)\n        for x in os.listdir(args.data_dir)\n        if x.endswith("".array_record"")\n    ]\n    grain_dataloader = get_dataloader(\n        array_record_files,\n        args.seq_len,\n        # NOTE: We deliberately pass the global batch size\n        # The dataloader shards the dataset across all processes\n        args.batch_size,\n        *image_shape,\n        num_workers=8,\n        prefetch_buffer_size=1,\n        seed=args.seed,\n    )\n    initial_state = grain_dataloader._create_initial_state()\n    grain_iterator = grain.DataLoaderIterator(grain_dataloader, initial_state)\n\n    # --- Restore checkpoint ---\n    if args.restore_ckpt:\n        # Restore full dynamics model\n        abstract_train_state = jax.tree_util.tree_map(\n            ocp.utils.to_shape_dtype_struct, train_state\n        )\n        restored = checkpoint_manager.restore(\n            checkpoint_manager.latest_step(),\n            args=ocp.args.Composite(\n                model_state=ocp.args.StandardRestore(abstract_train_state),\n                dataloader_state=grain.checkpoint.CheckpointRestore(grain_iterator),\n            ),\n        )\n        train_state = restored[""model_state""]\n        grain_iterator = restored[""dataloader_state""]\n        step = checkpoint_manager.latest_step() or 0\n        print(f""Restored dataloader and model state from step {step}"")\n    else:\n        # Restore from pre-trained tokenizer (and LAM)\n        train_state = restore_genie_components(\n            train_state, replicated_sharding, grain_iterator, dummy_inputs, rng, args\n        )\n\n    # --- TRAIN LOOP ---\n    dataloader = (jax.make_array_from_process_local_data(videos_sharding, elem) for elem in grain_iterator)  # type: ignore\n    while step < args.num_steps:\n        # for videos in dataloader:\n        videos = np.load(""overfit_dir/corner_8repl.npy"")\n        videos = jax.make_array_from_process_local_data(videos_sharding, videos)\n        while True:\n            # --- Train step ---\n            rng, _rng, _rng_dropout, _rng_mask = jax.random.split(rng, 4)\n\n            inputs = dict(\n                videos=videos,\n                rng=_rng,\n                dropout_rng=_rng_dropout,\n                mask_rng=_rng_mask,\n            )\n            train_state, loss, recon, metrics = train_step(train_state, inputs)\n            metrics[""lr""] = lr_schedule(step)\n            print(f""Step {step}, loss: {loss}"")\n            step += 1\n\n            # --- Logging ---\n            if args.log:\n                if step % args.log_interval == 0 and jax.process_index() == 0:\n                    wandb.log(\n                        {\n                            ""loss"": loss,\n                            ""step"": step,\n                            **metrics,\n                        }\n                    )\n                if step % args.log_image_interval == 0:\n                    gt_seq = inputs[""videos""][0].astype(jnp.float32) #/ 255.0\n                    recon_seq = recon[0].clip(0, 1)\n                    comparison_seq = jnp.concatenate((gt_seq, recon_seq), axis=1)\n                    comparison_seq = einops.rearrange(\n                        comparison_seq * 255, ""t h w c -> h (t w) c""\n                    )\n                    if jax.process_index() == 0:\n                        log_images = dict(\n                            image=wandb.Image(np.asarray(gt_seq[args.seq_len - 1])),\n                            recon=wandb.Image(np.asarray(recon_seq[args.seq_len - 1])),\n                            true_vs_recon=wandb.Image(\n                                np.asarray(comparison_seq.astype(np.uint8))\n                            ),\n                        )\n                        wandb.log(log_images)\n            # --- Checkpointing ---\n            if args.save_ckpt and step % args.log_checkpoint_interval == 0:\n                checkpoint_manager.save(\n                    step,\n                    args=ocp.args.Composite(\n                        model_state=ocp.args.StandardSave(train_state),\n                        dataloader_state=grain.checkpoint.CheckpointSave(\n                            grain_iterator\n                        ),\n                    ),\n                )\n                print(f""Saved checkpoint at step {step}"")\n            if step >= args.num_steps:\n                break\n\n    checkpoint_manager.close()\n",python,tab
+329,579332,"models/lam.py",0,0,"",python,tab
+330,580473,"train_dynamics.py",0,0,"from dataclasses import dataclass, field\nimport os\n\nimport einops\nfrom flax.training.train_state import TrainState\nfrom jax.sharding import Mesh, PartitionSpec, NamedSharding\nfrom jax.experimental.mesh_utils import create_device_mesh\nimport optax\nimport orbax.checkpoint as ocp\nimport numpy as np\nimport dm_pix as pix\nimport jax\nimport jax.numpy as jnp\nimport tyro\nimport wandb\nimport grain\n\nfrom genie import Genie, restore_genie_components\nfrom utils.dataloader import get_dataloader\nfrom utils.lr_utils import get_lr_schedule\nfrom utils.parameter_utils import count_parameters_by_component\n\n@dataclass\nclass Args:\n    # Experiment\n    num_steps: int = 200_000\n    seed: int = 0\n    seq_len: int = 16\n    image_channels: int = 3\n    image_height: int = 90\n    image_width: int = 160\n    data_dir: str = """"\n    save_ckpt: bool = False\n    restore_ckpt: bool = False\n    # Optimization\n    batch_size: int = 36\n    init_lr: float = 0.0\n    max_lr: float = 3e-5\n    decay_end: float = 0.0\n    wsd_decay_steps: int = 10000 # NOTE: wsd_decay_steps will only be used when using a wsd-schedule\n    warmup_steps: int = 5000\n    lr_schedule : str = ""wsd"" # supported options: wsd, cos\n    # Tokenizer\n    tokenizer_dim: int = 512\n    latent_patch_dim: int = 32\n    num_patch_latents: int = 1024\n    patch_size: int = 4\n    tokenizer_num_blocks: int = 8\n    tokenizer_num_heads: int = 8\n    tokenizer_checkpoint: str = """"\n    # LAM\n    lam_dim: int = 512\n    latent_action_dim: int = 32\n    num_latent_actions: int = 6\n    lam_patch_size: int = 16\n    lam_num_blocks: int = 8\n    lam_num_heads: int = 8\n    lam_checkpoint: str = """"\n    # Dynamics\n    dyna_dim: int = 512\n    dyna_num_blocks: int = 12\n    dyna_num_heads: int = 8\n    dropout: float = 0.0\n    mask_limit: float = 0.5\n    use_maskgit: bool = False\n    param_dtype: jnp.dtype = jnp.float32\n    dtype: jnp.dtype = jnp.bfloat16\n    # Logging\n    log: bool = False\n    entity: str = """"\n    project: str = """"\n    name: str = ""train_dynamics""\n    tags: list[str] = field(default_factory=lambda: [""dynamics""])\n    log_interval: int = 5\n    log_image_interval: int = 250\n    ckpt_dir: str = """"\n    log_checkpoint_interval: int = 25000\n    log_checkpoint_keep_period: int = 20000\n    log_gradients: bool = False\n    wandb_id: str = """"\n\n\nargs = tyro.cli(Args)\n\n\ndef dynamics_loss_fn(params, state, inputs):\n    """"""Compute masked dynamics loss""""""\n    inputs[""videos""] = inputs[""videos""].astype(args.dtype) / 255.0\n    outputs = state.apply_fn(\n        params,\n        inputs,\n        training=True,\n        rngs={""params"": inputs[""rng""], ""dropout"": inputs[""dropout_rng""]},\n    )\n    mask = outputs[""mask""]\n    outputs[""token_logits""] = outputs[""token_logits""].astype(jnp.float32)\n    logits = outputs[""token_logits""]\n    targets = outputs[""video_tokens""]\n\n    # if not args.use_maskgit:\n        # logits = outputs[""token_logits""][:, :, :-1]\n        # targets = outputs[""video_tokens""][:, :, 1:]\n        # mask = outputs[""mask""][:, :, 1:] \n    ce_loss = optax.softmax_cross_entropy_with_integer_labels(\n        logits, targets\n    )\n    ce_loss = (mask * ce_loss).sum() / mask.sum()\n    acc = logits.argmax(-1) == targets\n    acc = (mask * acc).sum() / mask.sum()\n    select_probs = jax.nn.softmax(logits)\n    gt = inputs[""videos""].clip(0, 1).reshape(-1, *inputs[""videos""].shape[2:])\n    recon = outputs[""recon""].clip(0, 1).reshape(-1, *outputs[""recon""].shape[2:])\n    psnr = pix.psnr(gt, recon).mean() # type: ignore\n    ssim = pix.ssim(gt, recon).mean() # type: ignore\n    _, index_counts_lam = jnp.unique_counts(\n        jnp.ravel(outputs[""lam_indices""]), size=args.num_latent_actions, fill_value=0\n    )\n    _, index_counts_tokenizer = jnp.unique_counts(\n        jnp.ravel(outputs[""video_tokens""]), size=args.num_patch_latents, fill_value=0\n    )\n    codebook_usage_lam = (index_counts_lam != 0).mean()\n    codebook_usage_tokenizer = (index_counts_tokenizer != 0).mean()\n    metrics = dict(\n        cross_entropy_loss=ce_loss,\n        masked_token_accuracy=acc,\n        select_logit=logits.max(-1).mean(),\n        select_p=select_probs.max(-1).mean(),\n        entropy=jax.scipy.special.entr(select_probs).sum(-1).mean(),\n        psnr=psnr,\n        ssim=ssim,\n        codebook_usage_lam=codebook_usage_lam,\n        codebook_usage_tokenizer=codebook_usage_tokenizer,\n    )\n    return ce_loss, (outputs[""recon""], metrics)\n\n\n@jax.jit\ndef train_step(state, inputs):\n    """"""Update state and compute metrics""""""\n    grad_fn = jax.value_and_grad(dynamics_loss_fn, has_aux=True, allow_int=True)\n    (loss, (recon, metrics)), grads = grad_fn(state.params, state, inputs)\n    state = state.apply_gradients(grads=grads)\n    if args.log_gradients:\n        metrics[""gradients_std/""] = jax.tree.map(\n            lambda x: x.std(), grads[""params""][""dynamics""]\n        )\n    return state, loss, recon, metrics\n\n\nif __name__ == ""__main__"":\n    jax.distributed.initialize()\n    num_devices = jax.device_count()\n    if num_devices == 0:\n        raise ValueError(""No JAX devices found."")\n    print(f""Running on {num_devices} devices."")\n\n    if args.batch_size % num_devices != 0:\n        raise ValueError(\n            f""Global batch size {args.batch_size} must be divisible by ""\n            f""number of devices {num_devices}.""\n        )\n\n    per_device_batch_size_for_init = args.batch_size // num_devices\n\n    rng = jax.random.PRNGKey(args.seed)\n\n    # --- Initialize model ---\n    genie = Genie(\n        # Tokenizer\n        in_dim=args.image_channels,\n        tokenizer_dim=args.tokenizer_dim,\n        latent_patch_dim=args.latent_patch_dim,\n        num_patch_latents=args.num_patch_latents,\n        patch_size=args.patch_size,\n        tokenizer_num_blocks=args.tokenizer_num_blocks,\n        tokenizer_num_heads=args.tokenizer_num_heads,\n        # LAM\n        lam_dim=args.lam_dim,\n        latent_action_dim=args.latent_action_dim,\n        num_latent_actions=args.num_latent_actions,\n        lam_patch_size=args.lam_patch_size,\n        lam_num_blocks=args.lam_num_blocks,\n        lam_num_heads=args.lam_num_heads,\n        lam_co_train=not args.lam_checkpoint,\n        # Dynamics\n        dyna_dim=args.dyna_dim,\n        dyna_num_blocks=args.dyna_num_blocks,\n        dyna_num_heads=args.dyna_num_heads,\n        dropout=args.dropout,\n        mask_limit=args.mask_limit,\n        use_maskgit=args.use_maskgit,\n        param_dtype=args.param_dtype,\n        dtype=args.dtype,\n    )\n    rng, _rng = jax.random.split(rng)\n    image_shape = (args.image_height, args.image_width, args.image_channels)\n    dummy_inputs = dict(\n        videos=jnp.zeros(\n            (per_device_batch_size_for_init, args.seq_len, *image_shape),\n            dtype=args.dtype,\n        ),\n        action=jnp.zeros(\n            (per_device_batch_size_for_init, args.seq_len), dtype=args.dtype\n        ),\n        mask_rng=_rng,\n    )\n    rng, _rng = jax.random.split(rng)\n    init_params = genie.init(_rng, dummy_inputs)\n\n    param_counts = count_parameters_by_component(init_params)\n\n    if args.log and jax.process_index() == 0:\n        wandb_init_kwargs = {\n            ""entity"": args.entity,\n            ""project"": args.project,\n            ""name"": args.name,\n            ""tags"": args.tags,\n            ""group"": ""debug"",\n            ""config"": args,\n        }\n\n        if args.wandb_id:\n            wandb_init_kwargs.update(\n                {\n                    ""id"": args.wandb_id,\n                    ""resume"": ""allow"",\n                }\n            )\n        wandb.init(**wandb_init_kwargs)\n\n        wandb.config.update({""model_param_count"": param_counts})\n\n    print(""Parameter counts:"")\n    print(param_counts)\n\n    # --- Initialize optimizer ---\n    lr_schedule = get_lr_schedule(args.lr_schedule, \n                                  args.init_lr, \n                                  args.max_lr, \n                                  args.decay_end, \n                                  args.num_steps, \n                                  args.warmup_steps, \n                                  args.wsd_decay_steps)\n    tx = optax.adamw(learning_rate=lr_schedule, b1=0.9, b2=0.9, weight_decay=1e-4, mu_dtype=args.dtype)\n    train_state = TrainState.create(apply_fn=genie.apply, params=init_params, tx=tx)\n\n    device_mesh_arr = create_device_mesh((num_devices,))\n    mesh = Mesh(devices=device_mesh_arr, axis_names=(""data"",))\n\n    replicated_sharding = NamedSharding(mesh, PartitionSpec())\n    videos_sharding = NamedSharding(mesh, PartitionSpec(""data"", None, None, None, None))\n    train_state = jax.device_put(train_state, replicated_sharding)\n\n    # --- Initialize checkpoint manager ---\n    step = 0\n    handler_registry = ocp.handlers.DefaultCheckpointHandlerRegistry()\n    handler_registry.add(\n        ""model_state"", ocp.args.StandardSave, ocp.handlers.StandardCheckpointHandler\n    )\n    handler_registry.add(\n        ""model_state"", ocp.args.StandardRestore, ocp.handlers.StandardCheckpointHandler\n    )\n    handler_registry.add(""dataloader_state"", grain.checkpoint.CheckpointSave, grain.checkpoint.CheckpointHandler)  # type: ignore\n    handler_registry.add(""dataloader_state"", grain.checkpoint.CheckpointRestore, grain.checkpoint.CheckpointHandler)  # type: ignore\n\n    checkpoint_options = ocp.CheckpointManagerOptions(\n        save_interval_steps=args.log_checkpoint_interval,\n        max_to_keep=3,\n        keep_period=args.log_checkpoint_keep_period,\n        step_format_fixed_length=6,\n        cleanup_tmp_directories=True,\n    )\n\n    checkpoint_manager = ocp.CheckpointManager(\n        args.ckpt_dir,\n        options=checkpoint_options,\n        handler_registry=handler_registry,\n    )\n\n    # --- Create DataLoaderIterator from dataloader ---\n    array_record_files = [\n        os.path.join(args.data_dir, x)\n        for x in os.listdir(args.data_dir)\n        if x.endswith("".array_record"")\n    ]\n    grain_dataloader = get_dataloader(\n        array_record_files,\n        args.seq_len,\n        # NOTE: We deliberately pass the global batch size\n        # The dataloader shards the dataset across all processes\n        args.batch_size,\n        *image_shape,\n        num_workers=8,\n        prefetch_buffer_size=1,\n        seed=args.seed,\n    )\n    initial_state = grain_dataloader._create_initial_state()\n    grain_iterator = grain.DataLoaderIterator(grain_dataloader, initial_state)\n\n    # --- Restore checkpoint ---\n    if args.restore_ckpt:\n        # Restore full dynamics model\n        abstract_train_state = jax.tree_util.tree_map(\n            ocp.utils.to_shape_dtype_struct, train_state\n        )\n        restored = checkpoint_manager.restore(\n            checkpoint_manager.latest_step(),\n            args=ocp.args.Composite(\n                model_state=ocp.args.StandardRestore(abstract_train_state),\n                dataloader_state=grain.checkpoint.CheckpointRestore(grain_iterator),\n            ),\n        )\n        train_state = restored[""model_state""]\n        grain_iterator = restored[""dataloader_state""]\n        step = checkpoint_manager.latest_step() or 0\n        print(f""Restored dataloader and model state from step {step}"")\n    else:\n        # Restore from pre-trained tokenizer (and LAM)\n        train_state = restore_genie_components(\n            train_state, replicated_sharding, grain_iterator, dummy_inputs, rng, args\n        )\n\n    # --- TRAIN LOOP ---\n    dataloader = (jax.make_array_from_process_local_data(videos_sharding, elem) for elem in grain_iterator)  # type: ignore\n    while step < args.num_steps:\n        # for videos in dataloader:\n        videos = np.load(""overfit_dir/corner_8repl.npy"")\n        videos = jax.make_array_from_process_local_data(videos_sharding, videos)\n        while True:\n            # --- Train step ---\n            rng, _rng, _rng_dropout, _rng_mask = jax.random.split(rng, 4)\n\n            inputs = dict(\n                videos=videos,\n                rng=_rng,\n                dropout_rng=_rng_dropout,\n                mask_rng=_rng_mask,\n            )\n            train_state, loss, recon, metrics = train_step(train_state, inputs)\n            metrics[""lr""] = lr_schedule(step)\n            print(f""Step {step}, loss: {loss}"")\n            step += 1\n\n            # --- Logging ---\n            if args.log:\n                if step % args.log_interval == 0 and jax.process_index() == 0:\n                    wandb.log(\n                        {\n                            ""loss"": loss,\n                            ""step"": step,\n                            **metrics,\n                        }\n                    )\n                if step % args.log_image_interval == 0:\n                    gt_seq = inputs[""videos""][0].astype(jnp.float32) #/ 255.0\n                    recon_seq = recon[0].clip(0, 1)\n                    comparison_seq = jnp.concatenate((gt_seq, recon_seq), axis=1)\n                    comparison_seq = einops.rearrange(\n                        comparison_seq * 255, ""t h w c -> h (t w) c""\n                    )\n                    if jax.process_index() == 0:\n                        log_images = dict(\n                            image=wandb.Image(np.asarray(gt_seq[args.seq_len - 1])),\n                            recon=wandb.Image(np.asarray(recon_seq[args.seq_len - 1])),\n                            true_vs_recon=wandb.Image(\n                                np.asarray(comparison_seq.astype(np.uint8))\n                            ),\n                        )\n                        wandb.log(log_images)\n            # --- Checkpointing ---\n            if args.save_ckpt and step % args.log_checkpoint_interval == 0:\n                checkpoint_manager.save(\n                    step,\n                    args=ocp.args.Composite(\n                        model_state=ocp.args.StandardSave(train_state),\n                        dataloader_state=grain.checkpoint.CheckpointSave(\n                            grain_iterator\n                        ),\n                    ),\n                )\n                print(f""Saved checkpoint at step {step}"")\n            if step >= args.num_steps:\n                break\n\n    checkpoint_manager.close()\n",python,tab
+331,591956,"genie.py",0,0,"",python,tab
+332,688925,"genie.py",5504,0,"",python,selection_mouse
+333,688934,"genie.py",5503,0,"",python,selection_command
+334,689596,"genie.py",5609,0,"",python,selection_mouse
+335,689637,"genie.py",5608,0,"",python,selection_command
+336,690705,"genie.py",5714,0,"",python,selection_mouse
+337,690707,"genie.py",5713,0,"",python,selection_command
+338,691487,"genie.py",5631,0,"",python,selection_mouse
+339,691487,"genie.py",5630,0,"",python,selection_command
+340,756358,"genie.py",5469,0,"",python,selection_mouse
+341,757517,"genie.py",5469,1,"2",python,content
+342,758824,"genie.py",5546,0,"",python,selection_mouse
+343,759698,"genie.py",5631,0,"",python,selection_mouse
+344,759708,"genie.py",5630,0,"",python,selection_command
+345,759993,"genie.py",5545,0,"",python,selection_mouse
+346,760169,"genie.py",5541,15,"token_idxs_full",python,selection_mouse
+347,760755,"genie.py",5600,0,"",python,selection_mouse
+348,760916,"genie.py",5596,13,"action_tokens",python,selection_mouse
+349,761613,"genie.py",5549,0,"",python,selection_mouse
+350,761767,"genie.py",5541,15,"token_idxs_full",python,selection_mouse
+351,762479,"genie.py",5602,0,"",python,selection_mouse
+352,762615,"genie.py",5596,13,"action_tokens",python,selection_mouse
+353,767312,"genie.py",5602,0,"",python,selection_mouse
+354,768315,"genie.py",5596,13,"action_tokens",python,selection_mouse
+355,776474,"TERMINAL",0,0,"bash",,terminal_focus
+356,779711,"genie.py",5603,0,"",python,selection_mouse
+357,779838,"genie.py",5596,13,"action_tokens",python,selection_mouse
+358,780696,"genie.py",5603,0,"",python,selection_mouse
+359,782745,"genie.py",5596,13,"action_tokens",python,selection_mouse
+360,783651,"genie.py",5603,0,"",python,selection_mouse
+361,784226,"genie.py",5596,13,"action_tokens",python,selection_mouse
+362,784989,"genie.py",5603,0,"",python,selection_mouse
+363,785492,"genie.py",5596,13,"action_tokens",python,selection_mouse
+364,786034,"genie.py",5603,0,"",python,selection_mouse
+365,787382,"genie.py",5596,13,"action_tokens",python,selection_mouse
+366,788115,"genie.py",5603,0,"",python,selection_mouse
+367,788525,"genie.py",5596,13,"action_tokens",python,selection_mouse
+368,789072,"genie.py",5603,0,"",python,selection_mouse
+369,789265,"genie.py",5596,13,"action_tokens",python,selection_mouse
+370,789951,"genie.py",5603,0,"",python,selection_mouse
+371,790570,"genie.py",5596,13,"action_tokens",python,selection_mouse
+372,791316,"genie.py",5603,0,"",python,selection_mouse
+373,791700,"genie.py",5596,13,"action_tokens",python,selection_mouse
+374,792268,"genie.py",5603,0,"",python,selection_mouse
+375,792451,"genie.py",5596,13,"action_tokens",python,selection_mouse
+376,793167,"genie.py",5603,0,"",python,selection_mouse
+377,794510,"genie.py",5596,13,"action_tokens",python,selection_mouse
+378,795142,"genie.py",5603,0,"",python,selection_mouse
+379,795521,"genie.py",5596,13,"action_tokens",python,selection_mouse
+380,795900,"genie.py",5558,52,"                    ""latent_actions"": action_tokens\n",python,selection_mouse
+381,796297,"genie.py",5603,0,"",python,selection_mouse
+382,796298,"genie.py",5596,13,"action_tokens",python,selection_mouse
+383,796497,"genie.py",5558,52,"                    ""latent_actions"": action_tokens\n",python,selection_mouse
+384,796867,"genie.py",5603,0,"",python,selection_mouse
+385,796868,"genie.py",5596,13,"action_tokens",python,selection_mouse
+386,797052,"genie.py",5558,52,"                    ""latent_actions"": action_tokens\n",python,selection_mouse
+387,797416,"genie.py",5603,0,"",python,selection_mouse
+388,797417,"genie.py",5596,13,"action_tokens",python,selection_mouse
+389,797585,"genie.py",5558,52,"                    ""latent_actions"": action_tokens\n",python,selection_mouse
+390,797927,"genie.py",5603,0,"",python,selection_mouse
+391,797928,"genie.py",5596,13,"action_tokens",python,selection_mouse
+392,798088,"genie.py",5558,52,"                    ""latent_actions"": action_tokens\n",python,selection_mouse
+393,798692,"genie.py",5603,0,"",python,selection_mouse

927a8af5474e5654810c00ce2e09fd2de87d3e5722f33fa1090d867db114e403/crowd-code-7ec41ffb-ddad-4e1d-b171-0513171669281757061617849-2025_09_05-10.41.46.448/source.csv

@@ -0,0 +1,44 @@
+Sequence,Time,File,RangeOffset,RangeLength,Text,Language,Type
+1,4,"train_tokenizer.py",0,0,"from dataclasses import dataclass, field\n\nimport os\nimport time\n\nimport einops\nfrom flax.training import orbax_utils\nfrom flax.training.train_state import TrainState\nimport optax\nimport orbax\nfrom orbax.checkpoint import PyTreeCheckpointer\nimport numpy as np\nimport dm_pix as pix\nimport jax\nimport jax.numpy as jnp\nimport tyro\nimport wandb\n\nfrom models.tokenizer import TokenizerVQVAE\nfrom utils.dataloader import get_dataloader\n\nts = int(time.time())\n\n\n@dataclass\nclass Args:\n    # Experiment\n    num_steps: int = 300_000\n    seed: int = 0\n    seq_len: int = 16\n    image_channels: int = 3\n    image_resolution: int = 64\n    data_dir: str = ""data/coinrun_episodes""\n    checkpoint: str = """"\n    # Optimization\n    vq_beta: float = 0.25\n    batch_size: int = 48\n    min_lr: float = 3e-4\n    max_lr: float = 3e-4\n    warmup_steps: int = 10000\n    # Tokenizer\n    model_dim: int = 512\n    latent_dim: int = 32\n    num_latents: int = 1024\n    patch_size: int = 4\n    num_blocks: int = 8\n    num_heads: int = 8\n    dropout: float = 0.0\n    codebook_dropout: float = 0.01\n    # Logging\n    log: bool = False\n    entity: str = """"\n    project: str = """"\n    log_interval: int = 5\n    log_image_interval: int = 250\n    ckpt_dir: str = """"\n    log_checkpoint_interval: int = 10000\n    log_gradients: bool = False\n    name: str = """"\n    tags: list[str] = field(default_factory=lambda: [""tokenizer""])\n    wandb_id: str = """"\n\n\nargs = tyro.cli(Args)\n\n\ndef tokenizer_loss_fn(params, state, inputs):\n    # --- Compute loss ---\n    outputs = state.apply_fn(\n        params, inputs, training=True, rngs={""dropout"": inputs[""rng""]}\n    )\n    mse = jnp.square(inputs[""videos""] - outputs[""recon""]).mean()\n    q_loss = jnp.square(jax.lax.stop_gradient(outputs[""emb""]) - outputs[""z""]).mean()\n    commitment_loss = jnp.square(\n        outputs[""emb""] - jax.lax.stop_gradient(outputs[""z""])\n    ).mean()\n    loss = mse + q_loss + args.vq_beta * commitment_loss\n\n    # --- Compute validation metrics ---\n    gt = inputs[""videos""].clip(0, 1).reshape(-1, *inputs[""videos""].shape[2:])\n    recon = outputs[""recon""].clip(0, 1).reshape(-1, *outputs[""recon""].shape[2:])\n    psnr = pix.psnr(gt, recon).mean()\n    ssim = pix.ssim(gt, recon).mean()\n    _, index_counts = jnp.unique_counts(\n        jnp.ravel(outputs[""indices""]), size=args.num_latents, fill_value=0\n    )\n    codebook_usage = (index_counts != 0).mean()\n    metrics = dict(\n        loss=loss,\n        mse=mse,\n        q_loss=q_loss,\n        commitment_loss=commitment_loss,\n        psnr=psnr,\n        ssim=ssim,\n        codebook_usage=codebook_usage,\n    )\n    return loss, (outputs[""recon""], metrics)\n\n\n@jax.jit\ndef train_step(state, inputs):\n    grad_fn = jax.value_and_grad(tokenizer_loss_fn, has_aux=True, allow_int=True)\n    (loss, (recon, metrics)), grads = grad_fn(state.params, state, inputs)\n    state = state.apply_gradients(grads=grads)\n    if args.log_gradients:\n        metrics[""encoder_gradients_std/""] = jax.tree.map(\n            lambda x: x.std(), grads[""params""][""encoder""]\n        )\n        metrics[""vq_gradients_std/""] = jax.tree.map(\n            lambda x: x.std(), grads[""params""][""vq""]\n        )\n        metrics[""decoder_gradients_std/""] = jax.tree.map(\n            lambda x: x.std(), grads[""params""][""decoder""]\n        )\n    return state, loss, recon, metrics\n\n\nif __name__ == ""__main__"":\n    rng = jax.random.PRNGKey(args.seed)\n    if args.log:\n        wandb_init_kwargs = {\n            ""entity"": args.entity,\n            ""project"": args.project,\n            ""name"": args.name,\n            ""tags"": args.tags,\n            ""group"": ""debug"",\n            ""config"": args,\n        }\n        if args.wandb_id:\n            wandb_init_kwargs.update(\n                {\n                    ""id"": args.wandb_id,\n                    ""resume"": ""allow"",\n                }\n            )\n        wandb.init(**wandb_init_kwargs)\n\n    # --- Initialize model ---\n    tokenizer = TokenizerVQVAE(\n        in_dim=args.image_channels,\n        model_dim=args.model_dim,\n        latent_dim=args.latent_dim,\n        num_latents=args.num_latents,\n        patch_size=args.patch_size,\n        num_blocks=args.num_blocks,\n        num_heads=args.num_heads,\n        dropout=args.dropout,\n        codebook_dropout=args.codebook_dropout,\n    )\n    rng, _rng = jax.random.split(rng)\n    image_shape = (args.image_resolution, args.image_resolution, args.image_channels)\n    inputs = dict(\n        videos=jnp.zeros(\n            (args.batch_size, args.seq_len, *image_shape), dtype=jnp.float32\n        ),\n    )\n    init_params = tokenizer.init(_rng, inputs)\n\n    # --- Load checkpoint ---\n    step = 0\n    if args.checkpoint:\n        init_params[""params""].update(\n            PyTreeCheckpointer().restore(args.checkpoint)[""model""][""params""][""params""]\n        )\n        # Assume checkpoint is of the form tokenizer_<timestamp>_<step>\n        step += int(args.checkpoint.split(""_"")[-1])\n\n    # --- Initialize optimizer ---\n    lr_schedule = optax.warmup_cosine_decay_schedule(\n        args.min_lr, args.max_lr, args.warmup_steps, args.num_steps\n    )\n    tx = optax.adamw(learning_rate=lr_schedule, b1=0.9, b2=0.9, weight_decay=1e-4)\n    train_state = TrainState.create(apply_fn=tokenizer.apply, params=init_params, tx=tx)\n\n    # --- TRAIN LOOP ---\n    dataloader = get_dataloader(args.data_dir, args.seq_len, args.batch_size)\n    while step < args.num_steps:\n        for videos in dataloader:\n            # --- Train step ---\n            rng, _rng = jax.random.split(rng)\n            inputs = dict(videos=videos, rng=_rng)\n            train_state, loss, recon, metrics = train_step(train_state, inputs)\n            print(f""Step {step}, loss: {loss}"")\n            step += 1\n\n            # --- Logging ---\n            if args.log:\n                if step % args.log_interval == 0:\n                    wandb.log({""loss"": loss, ""step"": step, **metrics})\n                if step % args.log_image_interval == 0:\n                    gt_seq = inputs[""videos""][0]\n                    recon_seq = recon[0].clip(0, 1)\n                    comparison_seq = jnp.concatenate((gt_seq, recon_seq), axis=1)\n                    comparison_seq = einops.rearrange(\n                        comparison_seq * 255, ""t h w c -> h (t w) c""\n                    )\n                    log_images = dict(\n                        image=wandb.Image(np.asarray(gt_seq[0])),\n                        recon=wandb.Image(np.asarray(recon_seq[0])),\n                        true_vs_recon=wandb.Image(\n                            np.asarray(comparison_seq.astype(np.uint8))\n                        ),\n                    )\n                    wandb.log(log_images)\n                if step % args.log_checkpoint_interval == 0:\n                    ckpt = {""model"": train_state}\n                    orbax_checkpointer = orbax.checkpoint.PyTreeCheckpointer()\n                    save_args = orbax_utils.save_args_from_target(ckpt)\n                    orbax_checkpointer.save(\n                        os.path.join(\n                            os.getcwd(), args.ckpt_dir, f""tokenizer_{ts}_{step}""\n                        ),\n                        ckpt,\n                        save_args=save_args,\n                    )\n            if step >= args.num_steps:\n                break\n",python,tab
+2,1363,"extension-output-pdoom-org.crowd-code-#1-crowd-code",0,0,"10:41:46 AM [info] Activating crowd-code\n10:41:46 AM [info] Recording started\n10:41:46 AM [info] Initializing git provider using file system watchers...\n",Log,tab
+3,1867,"extension-output-pdoom-org.crowd-code-#1-crowd-code",153,0,"10:41:47 AM [info] Git repository found\n10:41:47 AM [info] Git provider initialized successfully\n10:41:47 AM [info] Initial git state: [object Object]\n",Log,content
+4,5691,"train_tokenizer.py",0,0,"",python,tab
+5,53960,"train_tokenizer.py",138,0,"",python,selection_mouse
+6,58325,"train_tokenizer.py",7112,0,"",python,selection_mouse
+7,58366,"train_tokenizer.py",7111,0,"",python,selection_command
+8,159487,"TERMINAL",0,0,"undefined[tum_cte0515@hkn1991 jafar]$ idling",,terminal_command
+9,159551,"TERMINAL",0,0,"]633;C[?1049h(B[?7hEvery 1.0s: sinfo_t_idlehkn1991.localdomain: Fri Sep  5 10:44:25 2025Partition dev_cpuonly:     11 nodes idle\rPartition cpuonly:     57 nodes idle\rPartition dev_accelerated:\t 1 nodes idle\rPartition accelerated:     23 nodes idle\rPartition dev_accelerated-h100    :\t 0 nodes idle\rPartition accelerated-h100:\t 0 nodes idle\rPartition large:\t 5 nodes idle\rPartition accelerated-h200:\t 4 nodes idle",,terminal_output
+10,160574,"TERMINAL",0,0,"6",,terminal_output
+11,161591,"TERMINAL",0,0,"7",,terminal_output
+12,162626,"TERMINAL",0,0,"8",,terminal_output
+13,163666,"TERMINAL",0,0,"9",,terminal_output
+14,164777,"TERMINAL",0,0,"30",,terminal_output
+15,165806,"TERMINAL",0,0,"1",,terminal_output
+16,166849,"TERMINAL",0,0,"3",,terminal_output
+17,167831,"TERMINAL",0,0,"4",,terminal_output
+18,169142,"TERMINAL",0,0,"5",,terminal_output
+19,170027,"TERMINAL",0,0,"6",,terminal_output
+20,171110,"TERMINAL",0,0,"7",,terminal_output
+21,172057,"TERMINAL",0,0,"[?1049l\r[?1l>]0;tum_cte0515@hkn1991:~/Projects/jafar",,terminal_output
+22,173319,"TERMINAL",0,0,"queue",,terminal_command
+23,173390,"TERMINAL",0,0,"]633;C[?1049h(B[?7hEvery 1.0s: squeue --mehkn1991.localdomain: Fri Sep  5 10:44:39 2025JOBID PARTITION     NAME     USER ST\tTIME  NODES NODELIST(REASON)3465675 accelerat train_la tum_cte0  R   12:43:56\t  8 hkn[0703,0706-0707,0711-0715]3465676 accelerat train_la tum_cte0  R   12:43:56\t  8 hkn[0521-0528]3465677 accelerat train_la tum_cte0  R   12:43:56\t  8 hkn[0504,0520,0720,0722-0724,0728,0731]3466286 accelerat train_to tum_cte0  R   13:28:44\t  1 hkn07363466287 accelerat train_la tum_cte0  R   13:28:44\t  1 hkn0736",,terminal_output
+24,174408,"TERMINAL",0,0,"4077755",,terminal_output
+25,175471,"TERMINAL",0,0,"188866",,terminal_output
+26,176485,"TERMINAL",0,0,"299977",,terminal_output
+27,177528,"TERMINAL",0,0,"34:004:004:0088",,terminal_output
+28,178649,"TERMINAL",0,0,"411199",,terminal_output
+29,179612,"TERMINAL",0,0,"52225050",,terminal_output
+30,180679,"TERMINAL",0,0,"633311",,terminal_output
+31,181719,"TERMINAL",0,0,"744422",,terminal_output
+32,182779,"TERMINAL",0,0,"866644",,terminal_output
+33,183858,"TERMINAL",0,0,"5077755",,terminal_output
+34,184996,"TERMINAL",0,0,"188866",,terminal_output
+35,185902,"TERMINAL",0,0,"299977",,terminal_output
+36,186927,"TERMINAL",0,0,"310101088",,terminal_output
+37,188006,"TERMINAL",0,0,"411199",,terminal_output
+38,189079,"TERMINAL",0,0,"52229:009:00",,terminal_output
+39,190163,"TERMINAL",0,0,"633311",,terminal_output
+40,190333,"TERMINAL",0,0,"[?1049l\r[?1l>]0;tum_cte0515@hkn1991:~/Projects/jafar",,terminal_output
+41,200056,"TERMINAL",0,0,"cursor ../jasmine",,terminal_command
+42,200101,"TERMINAL",0,0,"]633;C",,terminal_output
+43,200182,"TERMINAL",0,0,"]0;tum_cte0515@hkn1991:~/Projects/jafar",,terminal_output

927a8af5474e5654810c00ce2e09fd2de87d3e5722f33fa1090d867db114e403/crowd-code-92e279fe-1f26-4694-b71c-d8f950b76bd11757499356165-2025_09_10-12.16.58.191/source.csv

@@ -0,0 +1,463 @@
+Sequence,Time,File,RangeOffset,RangeLength,Text,Language,Type
+1,4,"slurm/jobs/mihir/horeka/maskgit_big_runs/train_dynamics_8_nodes.sbatch",0,0,"#!/usr/bin/env bash\n\n#SBATCH --nodes=8\n#SBATCH --ntasks-per-node=4\n#SBATCH --time=48:00:00\n#SBATCH --partition=accelerated\n#SBATCH --cpus-per-task=5\n#SBATCH --gres=gpu:4\n#SBATCH --output=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/logs/logs_mihir/maskgit/dynamics-cotraining/%x_%j.log\n#SBATCH --error=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/logs/logs_mihir/maskgit/dynamics-cotraining/%x_%j.log\n#SBATCH --job-name=train_dynamics_maskgit_8_node\n#SBATCH --requeue\n#SBATCH --signal=b:usr1@300 # 5 min before timeout\n\n# --- signal trap to requeue job before timeout ---\nrequeue_job() {\n    echo ""[$(date)] caught sigusr1 (timeout warning), requeueing slurm job $SLURM_JOB_ID...""\n    # optional: trigger checkpoint saving here\n    # e.g., touch $checkpoint_dir/requeue_trigger\n    scontrol requeue $SLURM_JOB_ID\n    exit 0\n}\n\ntrap requeue_job sigusr1\n\n# set checkpoint flag based on restart count\nrestart_count=$(scontrol show job $SLURM_JOB_ID | grep -o 'Restarts=[0-9]*' | cut -d'=' -f2)\n\nif [ $restart_count -eq 0 ]; then\n    restore_ckpt_flag=""--no-restore-ckpt""\nelse\n    restore_ckpt_flag=""--restore-ckpt""\nfi\n\n\n\n# Log the sbatch script\ncat $0\n\nmodule unload mpi/openmpi/5.0\nmodule unload devel/cuda/12.4\nsource .venv/bin/activate\n\narray_records_dir=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/data_new/open_ai_minecraft_arrayrecords_chunked\n\njob_name=$SLURM_JOB_NAME\nslurm_job_id=$SLURM_JOB_ID\n\nCHECKPOINT_DIR=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/checkpoints/mihir/holiday/maskgit/$job_name/$slurm_job_id\nmkdir -p $CHECKPOINT_DIR\n\ntokenizer_ckpt_dir=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/checkpoints/big-runs/tokenizer/train_tokenizer_1e-4/3412401\n\nenv | grep SLURM\n\nsrun python train_dynamics.py \\n    --save_ckpt \\n    $restore_ckpt_flag \\n    --wandb_id $SLURM_JOB_ID \\n    --ckpt_dir $CHECKPOINT_DIR \\n    --batch_size=160 \\n    --init_lr=0 \\n    --dyna_type=maskgit \\n    --num_latent_actions=100 \\n    --max_lr=8e-5 \\n    --log_image_interval=1000 \\n    --log \\n    --log_checkpoint_interval=1000 \\n    --name=dynamics-maskgit-8-node-$slurm_job_id \\n    --tags dynamics maskgit 8-node post-launch-main \\n    --entity instant-uv \\n    --project jafar \\n    --dyna_dim=1024 \\n    --dyna_num_blocks=16 \\n    --dyna_num_heads=16 \\n    --dyna_ffn_dim=4096 \\n    --tokenizer_checkpoint=$tokenizer_ckpt_dir \\n    --data_dir $array_records_dir &\n\nchild_pid=$!\n\nwait $child_pid\n",shellscript,tab
+2,3124,"extension-output-pdoom-org.crowd-code-#1-crowd-code",0,0,"12:16:58 PM [info] Activating crowd-code\n12:16:58 PM [info] Recording started\n12:16:58 PM [info] Initializing git provider using file system watchers...\n12:16:59 PM [info] Git repository found\n12:16:59 PM [info] Git provider initialized successfully\n12:16:59 PM [info] Initial git state: [object Object]\n",Log,tab
+3,5660,"slurm/jobs/mihir/horeka/maskgit_big_runs/train_dynamics_8_nodes.sbatch",0,0,"",shellscript,tab
+4,23989,"TERMINAL",0,0,"source .venv/bin/activate",,terminal_command
+5,24001,"TERMINAL",0,0,"]633;C]0;tum_cte0515@hkn1993:~/Projects/jasmine",,terminal_output
+6,29626,"TERMINAL",0,0,"queue",,terminal_command
+7,29650,"TERMINAL",0,0,"]633;C[?1049h(B[?7hEvery 1.0s: squeue --mehkn1993.localdomain: Wed Sep 10 12:17:27 2025JOBID PARTITION     NAME     USER ST\tTIME  NODES NODELIST(REASON)",,terminal_output
+8,30414,"TERMINAL",0,0,"[?1049l\r[?1l>]0;tum_cte0515@hkn1993:~/Projects/jasmine",,terminal_output
+9,31320,"TERMINAL",0,0,"idling",,terminal_command
+10,31418,"TERMINAL",0,0,"]633;C[?1049h(B[?7hEvery 1.0s: sinfo_t_idlehkn1993.localdomain: Wed Sep 10 12:17:29 2025Partition dev_cpuonly:     10 nodes idle\rPartition cpuonly:     28 nodes idle\rPartition dev_accelerated:\t 0 nodes idle\rPartition accelerated:     29 nodes idle\rPartition dev_accelerated-h100    :\t 0 nodes idle\rPartition accelerated-h100:\t 0 nodes idle\rPartition large:\t 4 nodes idle\rPartition accelerated-h200:\t 0 nodes idle",,terminal_output
+11,32432,"TERMINAL",0,0,"30\t",,terminal_output
+12,33638,"TERMINAL",0,0,"1\t",,terminal_output
+13,34475,"TERMINAL",0,0,"[?1049l\r[?1l>]0;tum_cte0515@hkn1993:~/Projects/jasmine",,terminal_output
+14,38862,"slurm/jobs/mihir/horeka/preprocessing/coinrun_chunked.sh",0,0,"#!/usr/bin/env bash\n\n#SBATCH --nodes=1\n#SBATCH --ntasks-per-node=1\n#SBATCH --time=10:00:00\n#SBATCH --partition=large\n#SBATCH --cpus-per-task=16\n#SBATCH --output=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/logs/logs_mihir/preprocess/coinrun/%x_%j.log\n#SBATCH --error=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/logs/logs_mihir/preprocess/coinrun/%x_%j.log\n#SBATCH --job-name=preprocess_coinrun_chunked\n\nsource .venv/bin/activate\n\npython generate_dataset.py \\n    --min_episode_length 1000 \\n    --output_dir /hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/data_coinrun/coinrun_episodes_test \\n    --num_episodes 50",shellscript,tab
+15,61173,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",0,0,"#!/usr/bin/env bash\n\n#SBATCH --nodes=1\n#SBATCH --ntasks-per-node=4\n#SBATCH --time=00:20:00\n#SBATCH --partition=accelerated\n#SBATCH --cpus-per-task=5\n#SBATCH --gres=gpu:1\n#SBATCH --output=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/logs/logs_mihir/coinrun/tokenizer/%x_%j.log\n#SBATCH --error=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/logs/logs_mihir/coinrun/tokenizer/%x_%j.log\n#SBATCH --job-name=train_tokenizer_1e-4\n#SBATCH --requeue\n#SBATCH --signal=b:usr1@300 # 5 min before timeout\n\n# Log the sbatch script\ncat $0\n\nmodule unload mpi/openmpi/5.0\nmodule unload devel/cuda/12.4\nsource .venv/bin/activate\n\narray_records_dir_train=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/data_coinrun/coinrun_episodes_10m_gt_actions_split/train\n# array_records_dir_train=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/data_coinrun/coinrun_episodes_10m_chunked\narray_records_dir_val=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/data_coinrun/coinrun_episodes_10m_gt_actions_split/val\n\n\njob_name=$SLURM_JOB_NAME\nslurm_job_id=$SLURM_JOB_ID\n\nCHECKPOINT_DIR=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/checkpoints/big-runs/tokenizer/$job_name/$slurm_job_id\nmkdir -p $CHECKPOINT_DIR\n\nenv | grep SLURM\n\nsrun python train_tokenizer.py \\n    --save_ckpt \\n    --image_height=64 \\n    --image_width=64 \\n    --ckpt_dir $CHECKPOINT_DIR \\n    --batch_size=120 \\n    --init_lr=0 \\n    --max_lr=1e-4 \\n    --log_image_interval=250 \\n    --log_checkpoint_interval=250 \\n    --log \\n    --name=coinrun-tokenizer-singlechunk-$slurm_job_id \\n    --tags tokenizer coinrun dev \\n    --entity instant-uv \\n    --project jafar \\n    --warmup_steps 0 \\n    --wsd_decay_steps 0 \\n    --num_steps 250 \\n    --data_dir $array_records_dir_train\n",shellscript,tab
+16,75516,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",442,0,"",shellscript,selection_mouse
+17,75546,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",441,0,"",shellscript,selection_command
+18,79673,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1551,0,"",shellscript,selection_mouse
+19,79898,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1551,1,"s",shellscript,selection_mouse
+20,79900,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1551,3,"sin",shellscript,selection_mouse
+21,79900,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1551,4,"sing",shellscript,selection_mouse
+22,79901,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1551,5,"singl",shellscript,selection_mouse
+23,80001,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1551,6,"single",shellscript,selection_mouse
+24,80002,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1551,7,"singlec",shellscript,selection_mouse
+25,80059,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1551,8,"singlech",shellscript,selection_mouse
+26,80120,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1551,9,"singlechu",shellscript,selection_mouse
+27,80144,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1551,10,"singlechun",shellscript,selection_mouse
+28,80300,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1551,11,"singlechunk",shellscript,selection_mouse
+29,81146,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1551,11,"",shellscript,content
+30,82064,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1551,0,"d",shellscript,content
+31,82065,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1552,0,"",shellscript,selection_keyboard
+32,82250,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1552,0,"a",shellscript,content
+33,82251,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1553,0,"",shellscript,selection_keyboard
+34,82313,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1553,0,"t",shellscript,content
+35,82314,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1554,0,"",shellscript,selection_keyboard
+36,82397,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1554,0,"a",shellscript,content
+37,82399,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1555,0,"",shellscript,selection_keyboard
+38,82493,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1555,0,"s",shellscript,content
+39,82494,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1556,0,"",shellscript,selection_keyboard
+40,82705,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1556,0,"e",shellscript,content
+41,82706,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1557,0,"",shellscript,selection_keyboard
+42,82774,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1557,0,"t",shellscript,content
+43,82775,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1558,0,"",shellscript,selection_keyboard
+44,83305,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1558,0,"-",shellscript,content
+45,83306,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1559,0,"",shellscript,selection_keyboard
+46,83659,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1559,0,"t",shellscript,content
+47,83660,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1560,0,"",shellscript,selection_keyboard
+48,83770,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1560,0,"e",shellscript,content
+49,83771,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1561,0,"",shellscript,selection_keyboard
+50,83910,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1561,0,"s",shellscript,content
+51,83911,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1562,0,"",shellscript,selection_keyboard
+52,83961,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1562,0,"t",shellscript,content
+53,83962,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1563,0,"",shellscript,selection_keyboard
+54,84571,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1562,0,"",shellscript,selection_command
+55,92932,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1760,0,"",shellscript,selection_mouse
+56,93056,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1750,23,"array_records_dir_train",shellscript,selection_mouse
+57,96922,"TERMINAL",0,0,"cd $ws_dir",,terminal_command
+58,96934,"TERMINAL",0,0,"]633;C]0;tum_cte0515@hkn1993:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared",,terminal_output
+59,100051,"TERMINAL",0,0,"cd data_coinrun/",,terminal_command
+60,100728,"TERMINAL",0,0,"ls",,terminal_command
+61,100779,"TERMINAL",0,0,"]633;C",,terminal_output
+62,100866,"TERMINAL",0,0,"array_records     coinrun_episodes_10m          coinrun_episodes_10m_chunked_2   coinrun_episodes_10m_gt_actions_split  dev\r\ncoinrun_episodes  coinrun_episodes_10m_chunked  coinrun_episodes_10m_gt_actions  coinrun_episodes_test\r\n]0;tum_cte0515@hkn1993:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/data_coinrun",,terminal_output
+63,105341,"TERMINAL",0,0,"cd coinrun_episodes_test",,terminal_command
+64,106167,"TERMINAL",0,0,"pwd",,terminal_command
+65,111333,"TERMINAL",0,0,"ls",,terminal_command
+66,111395,"TERMINAL",0,0,"]633;Ccoinrun_episodes_0000.array_record  coinrun_episodes_0001.array_record  coinrun_episodes_0002.array_record  metadata.json\r\n]0;tum_cte0515@hkn1993:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/data_coinrun/coinrun_episodes_test",,terminal_output
+67,116969,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",656,0,"",shellscript,selection_mouse
+68,118998,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",656,0,"/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/data_coinrun/coinrun_episodes_test",shellscript,content
+69,120133,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",746,0,"\n",shellscript,content
+70,122782,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",747,0,"a",shellscript,content
+71,122782,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",748,0,"",shellscript,selection_keyboard
+72,123001,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",748,0,"r",shellscript,content
+73,123002,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",749,0,"",shellscript,selection_keyboard
+74,123195,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",749,0,"r",shellscript,content
+75,123196,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",750,0,"",shellscript,selection_keyboard
+76,123373,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",750,0,"a",shellscript,content
+77,123374,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",751,0,"",shellscript,selection_keyboard
+78,123549,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",751,0,"y",shellscript,content
+79,123550,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",752,0,"",shellscript,selection_keyboard
+80,123722,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",752,0,"_",shellscript,content
+81,123723,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",753,0,"",shellscript,selection_keyboard
+82,124747,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",747,6,"array_records_dir_train",shellscript,content
+83,125466,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",770,0,"=",shellscript,content
+84,125466,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",771,0,"",shellscript,selection_keyboard
+85,125945,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",770,0,"",shellscript,selection_command
+86,126439,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",747,0,"",shellscript,selection_command
+87,127192,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",747,0,"#",shellscript,content
+88,127192,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",748,0,"",shellscript,selection_keyboard
+89,127255,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",748,0," ",shellscript,content
+90,127256,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",749,0,"",shellscript,selection_keyboard
+91,127525,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",748,0,"",shellscript,selection_command
+92,137035,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1638,0,"",shellscript,selection_command
+93,139810,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1559,0,"",shellscript,selection_mouse
+94,139818,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1558,0,"",shellscript,selection_command
+95,174447,"TERMINAL",0,0,"cat metadata.json",,terminal_command
+96,187629,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1400,0,"",shellscript,selection_mouse
+97,187634,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1399,0,"",shellscript,selection_command
+98,192227,"TERMINAL",0,0,"dev",,terminal_command
+99,202007,"TERMINAL",0,0,"git branch",,terminal_command
+100,202122,"TERMINAL",0,0,"]633;C[?1h=\r  action-mapper\r\n  add-wandb-name-and-tags\r\n  before-nnx\r\n  causal-mem-reduce\r\n  causal-spatiotemporal-kv-cache\r\n  causal-st-transformer\r\n  causal-transformer-dynamics-model\r\n  causal-transformer-nnx-no-kv-cache\r\n  coinrun-gt-actions\r\n  convert-to-jax-array-in-iter\r\n  correct-batched-sampling\r\n  dev\r\n:",,terminal_output
+101,202673,"TERMINAL",0,0,"\r  dont-let-tf-see-gpu\r\n:",,terminal_output
+102,203441,"TERMINAL",0,0,"\r  feat/darkness-filter\r\n:\r  feat/explicit-image-dims\r\n:\r  fix-action-padding-lam-future-information-access\r\n:\r  fix-sampling\r\n:\r  fix-transformer-forwardpass\r\n:\r  fix/spatiotemporal-pe-once-in-STTransformer\r\n:\r  grad-norm-log-and-clip\r\n:\r  grain-dataloader\r\n:\r* input_pipeline/add-npy2array_record\r\n:\r  logging-variants\r\n:\r  lr-schedules\r\n:",,terminal_output
+103,203710,"TERMINAL",0,0,"\r  main\r\n:\r  maskgit-different-maskprob-per-sample\r\n:",,terminal_output
+104,203828,"TERMINAL",0,0,"\r  maskgit-sampling-iterative-unmasking-fix\r\n:\r  metrics-logging-for-dynamics-model\r\n:\r  monkey-patch\r\n:\r  new-arch-sampling\r\n:\r  preprocess_video\r\n:",,terminal_output
+105,203835,"TERMINAL",0,0,"\r  refactor-tmp\r\n:\r  revised-dataloader\r\n:\r  runner\r\n:",,terminal_output
+106,203954,"TERMINAL",0,0,"\r  runner-grain\r\n:\r  sample-ali-branch\r\n:\r  sample-from-different-topologies\r\n:",,terminal_output
+107,204315,"TERMINAL",0,0,"\rM  main\r\n\r:",,terminal_output
+108,204550,"TERMINAL",0,0,"\rM  lr-schedules\r\n\r:",,terminal_output
+109,205135,"TERMINAL",0,0,"\rM  logging-variants\r\n\r:\rM* input_pipeline/add-npy2array_record\r\n\r:",,terminal_output
+110,205246,"TERMINAL",0,0,"\rM  grain-dataloader\r\n\r:\rM  grad-norm-log-and-clip\r\n\r:\rM  fix/spatiotemporal-pe-once-in-STTransformer\r\n\r:",,terminal_output
+111,205453,"TERMINAL",0,0,"\rM  fix-transformer-forwardpass\r\n\r:",,terminal_output
+112,206225,"TERMINAL",0,0,"\r[?1l>]0;tum_cte0515@hkn1993:~/Projects/jasmine",,terminal_output
+113,210801,"TERMINAL",0,0,"sync-runner",,terminal_command
+114,210870,"TERMINAL",0,0,"]633;Csending incremental file list\r\n",,terminal_output
+115,215172,"TERMINAL",0,0,"r",,terminal_output
+116,215388,"TERMINAL",0,0,"u",,terminal_output
+117,215581,"TERMINAL",0,0,"n",,terminal_output
+118,215996,"TERMINAL",0,0,"n",,terminal_output
+119,216184,"TERMINAL",0,0,"./\r\nREADME.md\r\natari.log\r\ngenerate_dataset.py\r\ngenie.py\r\nrequirements.txt\r\nsample.py\r\ntrain_dynamics.py\r\ntrain_lam.py\r\ntrain_tokenizer.py\r\n",,terminal_output
+120,216446,"TERMINAL",0,0,"e",,terminal_output
+121,216499,"TERMINAL",0,0,"r",,terminal_output
+122,217367,"TERMINAL",0,0,"input_pipeline/download/\r\ninput_pipeline/download/download_array_records.sh\r\ninput_pipeline/download/openai/\r\ninput_pipeline/download/openai/download_actions_files.py\r\ninput_pipeline/preprocess/\r\ninput_pipeline/preprocess/pngs_to_array_records.py\r\ninput_pipeline/preprocess/video_to_array_records.py\r\nmodels/\r\nmodels/dynamics.py\r\nmodels/lam.py\r\nmodels/tokenizer.py\r\nslurm/jobs/mihir/horeka/\r\nslurm/jobs/mihir/horeka/coinrun/train_dyn_single_gpu.sh\r\nslurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh\r\nslurm/jobs/mihir/horeka/lam/coinrun/tmp\r\nslurm/jobs/mihir/horeka/lamap/\r\nslurm/jobs/mihir/horeka/lamap/coinrun_lam_base.sbatch\r\nslurm/jobs/mihir/horeka/lamap/coinrun_lam_base_dev.sh\r\nslurm/jobs/mihir/horeka/preprocessing/\r\nslurm/jobs/mihir/horeka/preprocessing/coinrun_chunked.sh\r\nslurm/jobs/mihir/horeka/preprocessing/preprocess_atari.sbatch\r\nslurm/jobs/mihir/horeka/preprocessing/preprocess_minecraft_multi_chunk.sbatch\r\nslurm/jobs/mihir/horeka/preprocessing/preprocess_minecraft_single_chunk.sbatch\r\ntests/\r\ntests/test_dataloader.py\r\nutils/\r\nutils/dataloader.py\r\nutils/lr_utils.py\r\nutils/nn.py\r\nutils/parameter_utils.py\r\nutils/preprocess.py\r\n",,terminal_output
+123,217724,"TERMINAL",0,0,"\r\nsent 317,315 bytes  received 839 bytes  42,420.53 bytes/sec\r\ntotal size is 128,665,823  speedup is 404.41\r\n]0;tum_cte0515@hkn1993:~/Projects/jasmine",,terminal_output
+124,218194,"TERMINAL",0,0,"runner",,terminal_command
+125,221290,"TERMINAL",0,0,"sbatch slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",,terminal_command
+126,221368,"TERMINAL",0,0,"]633;CSubmitted batch job 3482644\r\n]0;tum_cte0515@hkn1993:~/Projects/jasmine_jobs",,terminal_output
+127,224747,"TERMINAL",0,0,"dev",,terminal_command
+128,225736,"TERMINAL",0,0,"queue",,terminal_command
+129,225829,"TERMINAL",0,0,"]633;C[?1049h(B[?7hEvery 1.0s: squeue --mehkn1993.localdomain: Wed Sep 10 12:20:43 2025JOBID PARTITION     NAME     USER ST\tTIME  NODES NODELIST(REASON)3482644 accelerat train_to tum_cte0 PD\t0:00\t  1 (None)",,terminal_output
+130,226945,"TERMINAL",0,0,"4\t",,terminal_output
+131,227019,"TERMINAL",0,0,"[?1049l\r[?1l>]0;tum_cte0515@hkn1993:~/Projects/jasmine",,terminal_output
+132,228854,"TERMINAL",0,0,"idling",,terminal_command
+133,228898,"TERMINAL",0,0,"]633;C[?1049h(B[?7hEvery 1.0s: sinfo_t_idlehkn1993.localdomain: Wed Sep 10 12:20:46 2025Partition dev_cpuonly:     10 nodes idle\rPartition cpuonly:     21 nodes idle\rPartition dev_accelerated:\t 1 nodes idle\rPartition accelerated:     26 nodes idle\rPartition dev_accelerated-h100    :\t 0 nodes idle\rPartition accelerated-h100:\t 0 nodes idle\rPartition large:\t 4 nodes idle\rPartition accelerated-h200:\t 0 nodes idle",,terminal_output
+134,229911,"TERMINAL",0,0,"7\t",,terminal_output
+135,231057,"TERMINAL",0,0,"8\t",,terminal_output
+136,231792,"TERMINAL",0,0,"[?1049l\r[?1l>]0;tum_cte0515@hkn1993:~/Projects/jasmine",,terminal_output
+137,234616,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1169,0,"",shellscript,selection_mouse
+138,234626,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1168,0,"",shellscript,selection_command
+139,236891,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",111,0,"",shellscript,selection_mouse
+140,237873,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",111,0,"d",shellscript,content
+141,237875,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",112,0,"",shellscript,selection_keyboard
+142,238030,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",112,0,"e",shellscript,content
+143,238031,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",113,0,"",shellscript,selection_keyboard
+144,238265,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",113,0,"v",shellscript,content
+145,238267,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",114,0,"",shellscript,selection_keyboard
+146,238452,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",114,0,":",shellscript,content
+147,238453,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",115,0,"",shellscript,selection_keyboard
+148,239090,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",114,1,"",shellscript,content
+149,239374,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",114,0,"_",shellscript,content
+150,239375,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",115,0,"",shellscript,selection_keyboard
+151,239546,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",114,0,"",shellscript,selection_command
+152,247615,"TERMINAL",0,0,"sbatch slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",,terminal_command
+153,247667,"TERMINAL",0,0,"]633;C",,terminal_output
+154,247712,"TERMINAL",0,0,"Submitted batch job 3482647\r\n]0;tum_cte0515@hkn1993:~/Projects/jasmine",,terminal_output
+155,249746,"TERMINAL",0,0,"queue",,terminal_command
+156,249824,"TERMINAL",0,0,"]633;C[?1049h(B[?7hEvery 1.0s: squeue --mehkn1993.localdomain: Wed Sep 10 12:21:07 2025JOBID PARTITION     NAME     USER ST\tTIME  NODES NODELIST(REASON)3482644 accelerat train_to tum_cte0 PD\t0:00\t  1 (None)3482647 dev_accel train_to tum_cte0 PD\t0:00\t  1 (None)",,terminal_output
+157,250836,"TERMINAL",0,0,"8\t",,terminal_output
+158,251401,"TERMINAL",0,0,"9\t",,terminal_output
+159,252587,"TERMINAL",0,0,"10\t",,terminal_output
+160,253384,"TERMINAL",0,0,"1\t",,terminal_output
+161,254366,"TERMINAL",0,0,"2\t",,terminal_output
+162,255768,"TERMINAL",0,0,"3\t",,terminal_output
+163,256389,"TERMINAL",0,0,"4\t",,terminal_output
+164,257583,"TERMINAL",0,0,"5\t",,terminal_output
+165,258532,"TERMINAL",0,0,"6\t",,terminal_output
+166,259563,"TERMINAL",0,0,"7\t",,terminal_output
+167,260772,"TERMINAL",0,0,"8\t",,terminal_output
+168,261809,"TERMINAL",0,0,"9 Rhkn0817 Rhkn0401",,terminal_output
+169,262715,"TERMINAL",0,0,"2011",,terminal_output
+170,263698,"TERMINAL",0,0,"122",,terminal_output
+171,264824,"TERMINAL",0,0,"233",,terminal_output
+172,266036,"TERMINAL",0,0,"344",,terminal_output
+173,267051,"TERMINAL",0,0,"455",,terminal_output
+174,268023,"TERMINAL",0,0,"566",,terminal_output
+175,269152,"TERMINAL",0,0,"677",,terminal_output
+176,269932,"TERMINAL",0,0,"[?1049l\r[?1l>]0;tum_cte0515@hkn1993:~/Projects/jasmine",,terminal_output
+177,271743,"TERMINAL",0,0,"scancel 3482647",,terminal_command
+178,273551,"TERMINAL",0,0,"queue",,terminal_command
+179,273672,"TERMINAL",0,0,"]633;C[?1049h(B[?7hEvery 1.0s: squeue --mehkn1993.localdomain: Wed Sep 10 12:21:31 2025JOBID PARTITION     NAME     USER ST\tTIME  NODES NODELIST(REASON)3482644 accelerat train_to tum_cte0  R\t0:12\t  1 hkn08173482647 dev_accel train_to tum_cte0 CG\t0:10\t  1 hkn0401",,terminal_output
+180,274783,"TERMINAL",0,0,"23",,terminal_output
+181,275816,"TERMINAL",0,0,"34",,terminal_output
+182,276724,"TERMINAL",0,0,"45",,terminal_output
+183,277771,"TERMINAL",0,0,"56",,terminal_output
+184,278845,"TERMINAL",0,0,"67",,terminal_output
+185,279974,"TERMINAL",0,0,"78",,terminal_output
+186,280965,"TERMINAL",0,0,"89",,terminal_output
+187,281946,"TERMINAL",0,0,"921",,terminal_output
+188,282997,"TERMINAL",0,0,"412",,terminal_output
+189,284101,"TERMINAL",0,0,"23",,terminal_output
+190,285163,"TERMINAL",0,0,"34",,terminal_output
+191,286171,"TERMINAL",0,0,"45",,terminal_output
+192,287241,"TERMINAL",0,0,"56",,terminal_output
+193,288245,"TERMINAL",0,0,"67",,terminal_output
+194,289328,"TERMINAL",0,0,"78",,terminal_output
+195,290315,"TERMINAL",0,0,"89",,terminal_output
+196,291476,"TERMINAL",0,0,"930",,terminal_output
+197,292455,"TERMINAL",0,0,"501",,terminal_output
+198,293620,"TERMINAL",0,0,"12",,terminal_output
+199,294567,"TERMINAL",0,0,"23",,terminal_output
+200,295616,"TERMINAL",0,0,"34",,terminal_output
+201,296699,"TERMINAL",0,0,"45",,terminal_output
+202,297921,"TERMINAL",0,0,"56",,terminal_output
+203,298894,"TERMINAL",0,0,"\r67",,terminal_output
+204,300073,"TERMINAL",0,0,"78",,terminal_output
+205,300891,"TERMINAL",0,0,"89",,terminal_output
+206,301977,"TERMINAL",0,0,"940",,terminal_output
+207,303149,"TERMINAL",0,0,"2:001",,terminal_output
+208,304050,"TERMINAL",0,0,"13",,terminal_output
+209,305117,"TERMINAL",0,0,"34",,terminal_output
+210,306284,"TERMINAL",0,0,"45",,terminal_output
+211,307253,"TERMINAL",0,0,"56",,terminal_output
+212,308301,"TERMINAL",0,0,"67",,terminal_output
+213,309194,"TERMINAL",0,0,"78",,terminal_output
+214,310308,"TERMINAL",0,0,"89",,terminal_output
+215,311648,"TERMINAL",0,0,"950",,terminal_output
+216,312324,"TERMINAL",0,0,"101",,terminal_output
+217,313431,"TERMINAL",0,0,"12",,terminal_output
+218,314407,"TERMINAL",0,0,"23",,terminal_output
+219,315524,"TERMINAL",0,0,"34",,terminal_output
+220,316647,"TERMINAL",0,0,"45",,terminal_output
+221,317636,"TERMINAL",0,0,"56",,terminal_output
+222,318576,"TERMINAL",0,0,"67",,terminal_output
+223,319737,"TERMINAL",0,0,"78",,terminal_output
+224,320817,"TERMINAL",0,0,"89",,terminal_output
+225,322348,"TERMINAL",0,0,"91:00",,terminal_output
+226,322575,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1700,0,"",shellscript,selection_mouse
+227,322612,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",1699,0,"",shellscript,selection_command
+228,322935,"TERMINAL",0,0,"201",,terminal_output
+229,323973,"TERMINAL",0,0,"12",,terminal_output
+230,325218,"TERMINAL",0,0,"",,terminal_focus
+231,325509,"TERMINAL",0,0,"23",,terminal_output
+232,326120,"TERMINAL",0,0,"34",,terminal_output
+233,327369,"TERMINAL",0,0,"46",,terminal_output
+234,328332,"TERMINAL",0,0,"67",,terminal_output
+235,328998,"TERMINAL",0,0,"source /home/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/bin/activate",,terminal_command
+236,329132,"TERMINAL",0,0,"]633;C]0;tum_cte0515@hkn1993:~/Projects/jasmine",,terminal_output
+237,329190,"TERMINAL",0,0,"logs",,terminal_command
+238,329282,"TERMINAL",0,0,"]633;C]0;tum_cte0515@hkn1993:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/logs/logs_mihir",,terminal_output
+239,329373,"TERMINAL",0,0,"78",,terminal_output
+240,330304,"TERMINAL",0,0,"89",,terminal_output
+241,331470,"TERMINAL",0,0,"910",,terminal_output
+242,332525,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",0,0,"",shellscript,tab
+243,332880,"TERMINAL",0,0,"301",,terminal_output
+244,333323,"TERMINAL",0,0,"12",,terminal_output
+245,334277,"TERMINAL",0,0,"23",,terminal_output
+246,335925,"TERMINAL",0,0,"34",,terminal_output
+247,336549,"TERMINAL",0,0,"45",,terminal_output
+248,337407,"TERMINAL",0,0,"56",,terminal_output
+249,338539,"TERMINAL",0,0,"67",,terminal_output
+250,339614,"TERMINAL",0,0,"78",,terminal_output
+251,340593,"TERMINAL",0,0,"89",,terminal_output
+252,341627,"TERMINAL",0,0,"920",,terminal_output
+253,342604,"TERMINAL",0,0,"401",,terminal_output
+254,343758,"TERMINAL",0,0,"12",,terminal_output
+255,344553,"TERMINAL",0,0,"cd coinrun/tokenizer/",,terminal_command
+256,344755,"TERMINAL",0,0,"]633;C]0;tum_cte0515@hkn1993:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/logs/logs_mihir/coinrun/tokenizer",,terminal_output
+257,344836,"TERMINAL",0,0,"23",,terminal_output
+258,345816,"TERMINAL",0,0,"ls",,terminal_command
+259,345934,"TERMINAL",0,0,"]633;Ctrain_tokenizer_1e-4_3414046.log  train_tokenizer_1e-4_3482644.log  train_tokenizer_1e-4_3482647.log\r\n]0;tum_cte0515@hkn1993:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/logs/logs_mihir/coinrun/tokenizer",,terminal_output
+260,345982,"TERMINAL",0,0,"34",,terminal_output
+261,346913,"TERMINAL",0,0,"45",,terminal_output
+262,347913,"TERMINAL",0,0,"56",,terminal_output
+263,349088,"TERMINAL",0,0,"67",,terminal_output
+264,349962,"TERMINAL",0,0,"79",,terminal_output
+265,351152,"TERMINAL",0,0,"930",,terminal_output
+266,352109,"TERMINAL",0,0,"501",,terminal_output
+267,352869,"TERMINAL",0,0,"tail -f train_tokenizer_1e-4_3482644.log",,terminal_command
+268,352975,"TERMINAL",0,0,"]633;C    num_devices = jax.device_count()\r\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 907, in device_count\r\n    return int(get_backend(backend).device_count())\r\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 876, in get_backend\r\n    return _get_backend_uncached(platform)\r\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 855, in _get_backend_uncached\r\n    bs = backends()\r\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 758, in backends\r\n    raise RuntimeError(err_msg)\r\nRuntimeError: Unable to initialize backend 'cuda': INTERNAL: no supported devices found for platform CUDA (you may need to uninstall the failing plugin package, or set JAX_PLATFORMS=cpu to skip this backend.)\r\n",,terminal_output
+269,353132,"TERMINAL",0,0,"12",,terminal_output
+270,354488,"TERMINAL",0,0,"23",,terminal_output
+271,355378,"TERMINAL",0,0,"34",,terminal_output
+272,356434,"TERMINAL",0,0,"45",,terminal_output
+273,356838,"TERMINAL",0,0,"^C\r\n]0;tum_cte0515@hkn1993:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/logs/logs_mihir/coinrun/tokenizer",,terminal_output
+274,357606,"TERMINAL",0,0,"56",,terminal_output
+275,358469,"TERMINAL",0,0,"watch",,terminal_focus
+276,358548,"TERMINAL",0,0,"67",,terminal_output
+277,359403,"TERMINAL",0,0,"78",,terminal_output
+278,359637,"TERMINAL",0,0,"[?1049l\r[?1l>]0;tum_cte0515@hkn1993:~/Projects/jasmine",,terminal_output
+279,361610,"TERMINAL",0,0,"scancel 3482644",,terminal_command
+280,361666,"TERMINAL",0,0,"]633;C]0;tum_cte0515@hkn1993:~/Projects/jasmine",,terminal_output
+281,362935,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",0,0,"",shellscript,tab
+282,364016,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",634,0,"",shellscript,selection_mouse
+283,364034,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",633,0,"",shellscript,selection_command
+284,366006,"TERMINAL",0,0,"bash",,terminal_focus
+285,369355,"TERMINAL",0,0,"bash",,terminal_focus
+286,376821,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",0,0,"",shellscript,tab
+287,378682,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",635,0,"",shellscript,selection_command
+288,381176,"TERMINAL",0,0,"sbatch slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",,terminal_command
+289,383036,"TERMINAL",0,0,"bash",,terminal_focus
+290,384023,"TERMINAL",0,0,"bash",,terminal_focus
+291,385355,"TERMINAL",0,0,"queue",,terminal_command
+292,385425,"TERMINAL",0,0,"]633;C[?1049h(B[?7hEvery 1.0s: squeue --mehkn1993.localdomain: Wed Sep 10 12:23:23 2025JOBID PARTITION     NAME     USER ST\tTIME  NODES NODELIST(REASON)3482644 accelerat train_to tum_cte0 CG\t1:40\t  1 hkn08173482648 dev_accel train_to tum_cte0  R\t0:04\t  1 hkn0401",,terminal_output
+293,386467,"TERMINAL",0,0,"45",,terminal_output
+294,386791,"TERMINAL",0,0,"bash",,terminal_focus
+295,387303,"TERMINAL",0,0,"56",,terminal_output
+296,388315,"TERMINAL",0,0,"67",,terminal_output
+297,388656,"TERMINAL",0,0,"ls",,terminal_command
+298,388730,"TERMINAL",0,0,"]633;Ctrain_tokenizer_1e-4_3414046.log  train_tokenizer_1e-4_3482644.log  train_tokenizer_1e-4_3482647.log\r\n]0;tum_cte0515@hkn1993:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/logs/logs_mihir/coinrun/tokenizer",,terminal_output
+299,389462,"TERMINAL",0,0,"78",,terminal_output
+300,390448,"TERMINAL",0,0,"89",,terminal_output
+301,391051,"TERMINAL",0,0,"ls",,terminal_command
+302,391098,"TERMINAL",0,0,"]633;Ctrain_tokenizer_1e-4_3414046.log  train_tokenizer_1e-4_3482644.log  train_tokenizer_1e-4_3482647.log\r\n]0;tum_cte0515@hkn1993:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/logs/logs_mihir/coinrun/tokenizer",,terminal_output
+303,391576,"TERMINAL",0,0,"910",,terminal_output
+304,392609,"TERMINAL",0,0,"301",,terminal_output
+305,393716,"TERMINAL",0,0,"12",,terminal_output
+306,394835,"TERMINAL",0,0,"23",,terminal_output
+307,395882,"TERMINAL",0,0,"34",,terminal_output
+308,397039,"TERMINAL",0,0,"45",,terminal_output
+309,397883,"TERMINAL",0,0,"56",,terminal_output
+310,398873,"TERMINAL",0,0,"67",,terminal_output
+311,399115,"TERMINAL",0,0,"tail -f train_tokenizer_1e-4_3482648.log",,terminal_command
+312,399214,"TERMINAL",0,0,"]633;Ctail: cannot open 'train_tokenizer_1e-4_3482648.log' for reading: No such file or directory\r\ntail: no files remaining\r\n]0;tum_cte0515@hkn1993:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/logs/logs_mihir/coinrun/tokenizer",,terminal_output
+313,400134,"TERMINAL",0,0,"78",,terminal_output
+314,400982,"TERMINAL",0,0,"89",,terminal_output
+315,401513,"TERMINAL",0,0,"tail -f train_tokenizer_1e-4_3482648.log",,terminal_command
+316,401618,"TERMINAL",0,0,"]633;Ctail: cannot open 'train_tokenizer_1e-4_3482648.log' for reading: No such file or directory\r\ntail: no files remaining\r\n]0;tum_cte0515@hkn1993:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/logs/logs_mihir/coinrun/tokenizer",,terminal_output
+317,402363,"TERMINAL",0,0,"920",,terminal_output
+318,403391,"TERMINAL",0,0,"402",,terminal_output
+319,404432,"TERMINAL",0,0,"23",,terminal_output
+320,405435,"TERMINAL",0,0,"34",,terminal_output
+321,406724,"TERMINAL",0,0,"45",,terminal_output
+322,407758,"TERMINAL",0,0,"\r58 dev_accel R\t0:26401",,terminal_output
+323,408155,"TERMINAL",0,0,"lw",,terminal_command
+324,408223,"TERMINAL",0,0,"]633;Cbash: lw: command not found...\r\n",,terminal_output
+325,408732,"TERMINAL",0,0,"67",,terminal_output
+326,409500,"TERMINAL",0,0,"]0;tum_cte0515@hkn1993:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/logs/logs_mihir/coinrun/tokenizer",,terminal_output
+327,409714,"TERMINAL",0,0,"78",,terminal_output
+328,410600,"TERMINAL",0,0,"89",,terminal_output
+329,410919,"TERMINAL",0,0,"^C",,terminal_command
+330,411007,"TERMINAL",0,0,"]633;C]0;tum_cte0515@hkn1993:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/logs/logs_mihir/coinrun/tokenizer",,terminal_output
+331,411549,"TERMINAL",0,0,"ls",,terminal_command
+332,411676,"TERMINAL",0,0,"]633;Ctrain_tokenizer_1e-4_3414046.log  train_tokenizer_1e-4_3482647.log\r\ntrain_tokenizer_1e-4_3482644.log  train_tokenizer_1e-4_3482648.log\r\n]0;tum_cte0515@hkn1993:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/logs/logs_mihir/coinrun/tokenizer",,terminal_output
+333,411725,"TERMINAL",0,0,"930",,terminal_output
+334,412872,"TERMINAL",0,0,"501",,terminal_output
+335,413968,"TERMINAL",0,0,"12",,terminal_output
+336,414808,"TERMINAL",0,0,"23",,terminal_output
+337,415081,"TERMINAL",0,0,"tail -f train_tokenizer_1e-4_3482648.log",,terminal_command
+338,415163,"TERMINAL",0,0,"]633;CSLURM_NNODES=1\r\nSLURM_SUBMIT_HOST=hkn1993.localdomain\r\nSLURM_JOB_ID=3482648\r\nSLURM_NODEID=0\r\nSLURM_CONF=/etc/slurm/slurm.conf\r\nSLURM_JOB_NAME=train_tokenizer_1e-4\r\nSLURM_NTASKS_PER_NODE=4\r\nSLURM_JOB_GID=502226\r\nSLURM_JOB_NODELIST=hkn0401\r\nGpuFreq=control_disabled\r\n",,terminal_output
+339,415604,"TERMINAL",0,0,"34",,terminal_output
+340,416706,"TERMINAL",0,0,"45",,terminal_output
+341,417794,"TERMINAL",0,0,"56",,terminal_output
+342,418820,"TERMINAL",0,0,"67",,terminal_output
+343,419823,"TERMINAL",0,0,"78",,terminal_output
+344,420920,"TERMINAL",0,0,"89",,terminal_output
+345,422085,"TERMINAL",0,0,"940",,terminal_output
+346,423090,"TERMINAL",0,0,"4:001",,terminal_output
+347,424162,"TERMINAL",0,0,"12",,terminal_output
+348,425508,"TERMINAL",0,0,"23",,terminal_output
+349,426448,"TERMINAL",0,0,"34",,terminal_output
+350,427243,"TERMINAL",0,0,"46",,terminal_output
+351,428362,"TERMINAL",0,0,"67",,terminal_output
+352,428476,"TERMINAL",0,0,"2025-09-10 12:24:06.019374: W external/xla/xla/service/platform_util.cc:220] unable to create StreamExecutor for CUDA:1: CUDA error: Failed call to cuDeviceGet: CUDA_ERROR_INVALID_DEVICE: invalid device ordinal\r\nTraceback (most recent call last):\r\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 742, in backends\r\n    backend = _init_backend(platform)\r\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 828, in _init_backend\r\n    backend = registration.factory()\r\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 528, in factory\r\n    return xla_client.make_c_api_client(\r\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jaxlib/xla_client.py"", line 153, in make_c_api_client\r\n2025-09-10 12:24:06.023260: W external/xla/xla/service/platform_util.cc:220] unable to create StreamExecutor for CUDA:3: CUDA error: Failed call to cuDeviceGet: CUDA_ERROR_INVALID_DEVICE: invalid device ordinal\r\nTraceback (most recent call last):\r\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 742, in backends\r\n    backend = _init_backend(platform)\r\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 828, in _init_backend\r\n    return _xla.get_c_api_client(plugin_name, options, distributed_client)\r\njaxlib._jax.XlaRuntimeError: INTERNAL: no supported devices found for platform CUDA\r\n\r\nDuring handling of the above exception, another exception occurred:\r\n\r\nTraceback (most recent call last):\r\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/train_tokenizer.py"", line 145, in <module>\r\n    backend = registration.factory()\r\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 528, in factory\r\n    num_devices = jax.device_count()\r\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 907, in device_count\r\n    return xla_client.make_c_api_client(\r\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jaxlib/xla_client.py"", line 153, in make_c_api_client\r\n    return int(get_backend(backend).device_count())\r\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 876, in get_backend\r\n    return _xla.get_c_api_client(plugin_name, options, distributed_client)\r\njaxlib._jax.XlaRuntimeError: INTERNAL: no supported devices found for platform CUDA\r\n\r\nDuring handling of the above exception, another exception occurred:\r\n\r\nTraceback (most recent call last):\r\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/train_tokenizer.py"", line 145, in <module>\r\n    return _get_backend_uncached(platform)\r\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 855, in _get_backend_uncached\r\n    num_devices = jax.device_count()\r\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 907, in device_count\r\n    bs = backends()\r\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 758, in backends\r\n    return int(get_backend(backend).device_count())\r\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 876, in get_backend\r\n    raise RuntimeError(err_msg)\r\nRuntimeError: Unable to initialize backend 'cuda': INTERNAL: no supported devices found for platform CUDA (you may need to uninstall the failing plugin package, or set JAX_PLATFORMS=cpu to skip this backend.)\r\n    return _get_backend_uncached(platform)\r\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 855, in _get_backend_uncached\r\n    bs = backends()\r\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 758, in backends\r\n    raise RuntimeError(err_msg)\r\nRuntimeError: Unable to initialize backend 'cuda': INTERNAL: no supported devices found for platform CUDA (you may need to uninstall the failing plugin package, or set JAX_PLATFORMS=cpu to skip this backend.)\r\n2025-09-10 12:24:06.035570: W external/xla/xla/service/platform_util.cc:220] unable to create StreamExecutor for CUDA:2: CUDA error: Failed call to cuDeviceGet: CUDA_ERROR_INVALID_DEVICE: invalid device ordinal\r\nTraceback (most recent call last):\r\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 742, in backends\r\n    backend = _init_backend(platform)\r\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 828, in _init_backend\r\n    backend = registration.factory()\r\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 528, in factory\r\n    return xla_client.make_c_api_client(\r\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jaxlib/xla_client.py"", line 153, in make_c_api_client\r\n    return _xla.get_c_api_client(plugin_name, options, distributed_client)\r\njaxlib._jax.XlaRuntimeError: INTERNAL: no supported devices found for platform CUDA\r\n\r\nDuring handling of the above exception, another exception occurred:\r\n\r\nTraceback (most recent call last):\r\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/train_tokenizer.py"", line 145, in <module>\r\n    num_devices = jax.device_count()\r\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 907, in device_count\r\n    return int(get_backend(backend).device_count())\r\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 876, in get_backend\r\n    return _get_backend_uncached(platform)\r\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 855, in _get_backend_uncached\r\n    bs = backends()\r\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 758, in backends\r\n    raise RuntimeError(err_msg)\r\nRuntimeError: Unable to initialize backend 'cuda': INTERNAL: no supported devices found for platform CUDA (you may need to uninstall the failing plugin package, or set JAX_PLATFORMS=cpu to skip this backend.)\r\n",,terminal_output
+353,429404,"TERMINAL",0,0,"78",,terminal_output
+354,430463,"TERMINAL",0,0,"89",,terminal_output
+355,431556,"TERMINAL",0,0,"950",,terminal_output
+356,432737,"TERMINAL",0,0,"101",,terminal_output
+357,433906,"TERMINAL",0,0,"12",,terminal_output
+358,435071,"TERMINAL",0,0,"23",,terminal_output
+359,435948,"TERMINAL",0,0,"34",,terminal_output
+360,436767,"TERMINAL",0,0,"45",,terminal_output
+361,438034,"TERMINAL",0,0,"56",,terminal_output
+362,439498,"TERMINAL",0,0,"67",,terminal_output
+363,440592,"TERMINAL",0,0,"78",,terminal_output
+364,441652,"TERMINAL",0,0,"89",,terminal_output
+365,442246,"TERMINAL",0,0,"91:00",,terminal_output
+366,443079,"TERMINAL",0,0,"201",,terminal_output
+367,444770,"TERMINAL",0,0,"12",,terminal_output
+368,445738,"TERMINAL",0,0,"23",,terminal_output
+369,446864,"TERMINAL",0,0,"34",,terminal_output
+370,447823,"TERMINAL",0,0,"45",,terminal_output
+371,449009,"TERMINAL",0,0,"56",,terminal_output
+372,449577,"TERMINAL",0,0,"78",,terminal_output
+373,450598,"TERMINAL",0,0,"89",,terminal_output
+374,451909,"TERMINAL",0,0,"910",,terminal_output
+375,453162,"TERMINAL",0,0,"301",,terminal_output
+376,453568,"TERMINAL",0,0,"watch",,terminal_focus
+377,454215,"TERMINAL",0,0,"12",,terminal_output
+378,454538,"TERMINAL",0,0,"23",,terminal_output
+379,454763,"TERMINAL",0,0,"[?1049l\r[?1l>]0;tum_cte0515@hkn1993:~/Projects/jasmine",,terminal_output
+380,457423,"TERMINAL",0,0,"scancel 3482648",,terminal_command
+381,458058,"TERMINAL",0,0,"tail",,terminal_focus
+382,458199,"TERMINAL",0,0,"srun: Job step aborted: Waiting up to 32 seconds for job step to finish.\r\nslurmstepd: error: *** JOB 3482648 ON hkn0401 CANCELLED AT 2025-09-10T12:24:35 ***\r\nsrun: got SIGCONT\r\nsrun: forcing job termination\r\nslurmstepd: error: *** STEP 3482648.0 ON hkn0401 CANCELLED AT 2025-09-10T12:24:35 ***\r\n",,terminal_output
+383,458489,"TERMINAL",0,0,"^C\r\n]0;tum_cte0515@hkn1993:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/logs/logs_mihir/coinrun/tokenizer",,terminal_output
+384,461173,"TERMINAL",0,0,"ls",,terminal_command
+385,461483,"TERMINAL",0,0,"]633;Ctrain_tokenizer_1e-4_3414046.log  train_tokenizer_1e-4_3482647.log\r\ntrain_tokenizer_1e-4_3482644.log  train_tokenizer_1e-4_3482648.log\r\n]0;tum_cte0515@hkn1993:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/logs/logs_mihir/coinrun/tokenizer",,terminal_output
+386,463277,"/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/logs/logs_mihir/coinrun/tokenizer/train_tokenizer_1e-4_3482648.log",0,0,"#!/usr/bin/env bash\n\n#SBATCH --nodes=1\n#SBATCH --ntasks-per-node=4\n#SBATCH --time=00:20:00\n#SBATCH --partition=dev_accelerated\n#SBATCH --cpus-per-task=5\n#SBATCH --gres=gpu:1\n#SBATCH --output=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/logs/logs_mihir/coinrun/tokenizer/%x_%j.log\n#SBATCH --error=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/logs/logs_mihir/coinrun/tokenizer/%x_%j.log\n#SBATCH --job-name=train_tokenizer_1e-4\n#SBATCH --requeue\n#SBATCH --signal=b:usr1@300 # 5 min before timeout\n\n# Log the sbatch script\ncat $0\n\nmodule unload mpi/openmpi/5.0\nmodule unload devel/cuda/12.4\nsource .venv/bin/activate\n\narray_records_dir_train=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/data_coinrun/coinrun_episodes_test\n# array_records_dir_train=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/data_coinrun/coinrun_episodes_10m_gt_actions_split/train\n# array_records_dir_train=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/data_coinrun/coinrun_episodes_10m_chunked\narray_records_dir_val=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/data_coinrun/coinrun_episodes_10m_gt_actions_split/val\n\n\njob_name=$SLURM_JOB_NAME\nslurm_job_id=$SLURM_JOB_ID\n\nCHECKPOINT_DIR=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/checkpoints/big-runs/tokenizer/$job_name/$slurm_job_id\nmkdir -p $CHECKPOINT_DIR\n\nenv | grep SLURM\n\nsrun python train_tokenizer.py \\n    --save_ckpt \\n    --image_height=64 \\n    --image_width=64 \\n    --ckpt_dir $CHECKPOINT_DIR \\n    --batch_size=120 \\n    --init_lr=0 \\n    --max_lr=1e-4 \\n    --log_image_interval=250 \\n    --log_checkpoint_interval=250 \\n    --log \\n    --name=coinrun-tokenizer-dataset-test-$slurm_job_id \\n    --tags tokenizer coinrun dev \\n    --entity instant-uv \\n    --project jafar \\n    --warmup_steps 0 \\n    --wsd_decay_steps 0 \\n    --num_steps 250 \\n    --data_dir $array_records_dir_train\nSLURM_JOB_USER=tum_cte0515\nSLURM_TASKS_PER_NODE=4\nSLURM_JOB_UID=999226\nSLURM_TASK_PID=3106821\nSLURM_JOB_GPUS=0\nSLURM_LOCALID=0\nSLURM_SUBMIT_DIR=/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine\nSLURMD_NODENAME=hkn0401\nSLURM_JOB_START_TIME=1757499799\nSLURM_CLUSTER_NAME=hk\nSLURM_JOB_END_TIME=1757500999\nSLURM_CPUS_ON_NODE=24\nSLURM_JOB_CPUS_PER_NODE=24\nSLURM_GPUS_ON_NODE=1\nSLURM_GTIDS=0\nSLURM_JOB_PARTITION=dev_accelerated\nSLURM_TRES_PER_TASK=cpu=5\nSLURM_OOM_KILL_STEP=0\nSLURM_JOB_NUM_NODES=1\nSLURM_JOBID=3482648\nSLURM_JOB_QOS=normal\nSLURM_PROCID=0\nSLURM_CPUS_PER_TASK=5\nSLURM_NTASKS=4\nSLURM_TOPOLOGY_ADDR=hkibb.hkibbi1.hkibbi1e11.hkn0401\nSLURM_TOPOLOGY_ADDR_PATTERN=switch.switch.switch.node\nSLURM_SCRIPT_CONTEXT=prolog_task\nSLURM_NODELIST=hkn0401\nSLURM_JOB_ACCOUNT=hk-project-p0023960\nSLURM_PRIO_PROCESS=0\nSLURM_NPROCS=4\nSLURM_NNODES=1\nSLURM_SUBMIT_HOST=hkn1993.localdomain\nSLURM_JOB_ID=3482648\nSLURM_NODEID=0\nSLURM_CONF=/etc/slurm/slurm.conf\nSLURM_JOB_NAME=train_tokenizer_1e-4\nSLURM_NTASKS_PER_NODE=4\nSLURM_JOB_GID=502226\nSLURM_JOB_NODELIST=hkn0401\nGpuFreq=control_disabled\n2025-09-10 12:24:06.019374: W external/xla/xla/service/platform_util.cc:220] unable to create StreamExecutor for CUDA:1: CUDA error: Failed call to cuDeviceGet: CUDA_ERROR_INVALID_DEVICE: invalid device ordinal\nTraceback (most recent call last):\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 742, in backends\n    backend = _init_backend(platform)\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 828, in _init_backend\n    backend = registration.factory()\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 528, in factory\n    return xla_client.make_c_api_client(\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jaxlib/xla_client.py"", line 153, in make_c_api_client\n2025-09-10 12:24:06.023260: W external/xla/xla/service/platform_util.cc:220] unable to create StreamExecutor for CUDA:3: CUDA error: Failed call to cuDeviceGet: CUDA_ERROR_INVALID_DEVICE: invalid device ordinal\nTraceback (most recent call last):\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 742, in backends\n    backend = _init_backend(platform)\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 828, in _init_backend\n    return _xla.get_c_api_client(plugin_name, options, distributed_client)\njaxlib._jax.XlaRuntimeError: INTERNAL: no supported devices found for platform CUDA\n\nDuring handling of the above exception, another exception occurred:\n\nTraceback (most recent call last):\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/train_tokenizer.py"", line 145, in <module>\n    backend = registration.factory()\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 528, in factory\n    num_devices = jax.device_count()\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 907, in device_count\n    return xla_client.make_c_api_client(\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jaxlib/xla_client.py"", line 153, in make_c_api_client\n    return int(get_backend(backend).device_count())\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 876, in get_backend\n    return _xla.get_c_api_client(plugin_name, options, distributed_client)\njaxlib._jax.XlaRuntimeError: INTERNAL: no supported devices found for platform CUDA\n\nDuring handling of the above exception, another exception occurred:\n\nTraceback (most recent call last):\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/train_tokenizer.py"", line 145, in <module>\n    return _get_backend_uncached(platform)\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 855, in _get_backend_uncached\n    num_devices = jax.device_count()\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 907, in device_count\n    bs = backends()\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 758, in backends\n    return int(get_backend(backend).device_count())\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 876, in get_backend\n    raise RuntimeError(err_msg)\nRuntimeError: Unable to initialize backend 'cuda': INTERNAL: no supported devices found for platform CUDA (you may need to uninstall the failing plugin package, or set JAX_PLATFORMS=cpu to skip this backend.)\n    return _get_backend_uncached(platform)\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 855, in _get_backend_uncached\n    bs = backends()\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 758, in backends\n    raise RuntimeError(err_msg)\nRuntimeError: Unable to initialize backend 'cuda': INTERNAL: no supported devices found for platform CUDA (you may need to uninstall the failing plugin package, or set JAX_PLATFORMS=cpu to skip this backend.)\n2025-09-10 12:24:06.035570: W external/xla/xla/service/platform_util.cc:220] unable to create StreamExecutor for CUDA:2: CUDA error: Failed call to cuDeviceGet: CUDA_ERROR_INVALID_DEVICE: invalid device ordinal\nTraceback (most recent call last):\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 742, in backends\n    backend = _init_backend(platform)\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 828, in _init_backend\n    backend = registration.factory()\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 528, in factory\n    return xla_client.make_c_api_client(\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jaxlib/xla_client.py"", line 153, in make_c_api_client\n    return _xla.get_c_api_client(plugin_name, options, distributed_client)\njaxlib._jax.XlaRuntimeError: INTERNAL: no supported devices found for platform CUDA\n\nDuring handling of the above exception, another exception occurred:\n\nTraceback (most recent call last):\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/train_tokenizer.py"", line 145, in <module>\n    num_devices = jax.device_count()\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 907, in device_count\n    return int(get_backend(backend).device_count())\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 876, in get_backend\n    return _get_backend_uncached(platform)\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 855, in _get_backend_uncached\n    bs = backends()\n  File ""/hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/.venv/lib/python3.10/site-packages/jax/_src/xla_bridge.py"", line 758, in backends\n    raise RuntimeError(err_msg)\nRuntimeError: Unable to initialize backend 'cuda': INTERNAL: no supported devices found for platform CUDA (you may need to uninstall the failing plugin package, or set JAX_PLATFORMS=cpu to skip this backend.)\nsrun: Job step aborted: Waiting up to 32 seconds for job step to finish.\nslurmstepd: error: *** JOB 3482648 ON hkn0401 CANCELLED AT 2025-09-10T12:24:35 ***\nsrun: got SIGCONT\nsrun: forcing job termination\nslurmstepd: error: *** STEP 3482648.0 ON hkn0401 CANCELLED AT 2025-09-10T12:24:35 ***\n",log,tab
+387,485481,"TERMINAL",0,0,"bash",,terminal_focus
+388,486531,"TERMINAL",0,0,"bash",,terminal_focus
+389,499618,"TERMINAL",0,0,"salloc --time=01:00:00 --partition=dev_accelerated --nodes=1 --gres=gpu:1  --cpus-per-task=8",,terminal_command
+390,499686,"TERMINAL",0,0,"]633;Csalloc: Pending job allocation 3482653\r\nsalloc: job 3482653 queued and waiting for resources\r\n",,terminal_output
+391,500971,"TERMINAL",0,0,"salloc: job 3482653 has been allocated resources\r\nsalloc: Granted job allocation 3482653\r\n",,terminal_output
+392,501073,"TERMINAL",0,0,"salloc: Waiting for resource configuration\r\n",,terminal_output
+393,503080,"TERMINAL",0,0,"s",,terminal_output
+394,503220,"TERMINAL",0,0,"o",,terminal_output
+395,503324,"TERMINAL",0,0,"u",,terminal_output
+396,503379,"TERMINAL",0,0,"r",,terminal_output
+397,503609,"TERMINAL",0,0,"c",,terminal_output
+398,503739,"TERMINAL",0,0,"e",,terminal_output
+399,503850,"TERMINAL",0,0," ",,terminal_output
+400,503927,"TERMINAL",0,0,".",,terminal_output
+401,504031,"TERMINAL",0,0,"v",,terminal_output
+402,504533,"TERMINAL",0,0,"e",,terminal_output
+403,504676,"TERMINAL",0,0,"n",,terminal_output
+404,504866,"TERMINAL",0,0,"v",,terminal_output
+405,505245,"TERMINAL",0,0,"/",,terminal_output
+406,505439,"TERMINAL",0,0,"b",,terminal_output
+407,505799,"TERMINAL",0,0,"i",,terminal_output
+408,505912,"TERMINAL",0,0,"n",,terminal_output
+409,506619,"TERMINAL",0,0,"/",,terminal_output
+410,506849,"TERMINAL",0,0,"a",,terminal_output
+411,506986,"TERMINAL",0,0,"c",,terminal_output
+412,507149,"TERMINAL",0,0,"c",,terminal_output
+413,507284,"TERMINAL",0,0,"e",,terminal_output
+414,507360,"TERMINAL",0,0,"l",,terminal_output
+415,507783,"TERMINAL",0,0,"e",,terminal_output
+416,507961,"TERMINAL",0,0,"r",,terminal_output
+417,508262,"TERMINAL",0,0,"a",,terminal_output
+418,508527,"TERMINAL",0,0,"t",,terminal_output
+419,508662,"TERMINAL",0,0,"e",,terminal_output
+420,509062,"TERMINAL",0,0,"\r\n",,terminal_output
+421,513649,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",0,0,"",shellscript,tab
+422,518757,"TERMINAL",0,0,"sh",,terminal_output
+423,518865,"TERMINAL",0,0," ",,terminal_output
+424,519419,"TERMINAL",0,0,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",,terminal_output
+425,520348,"TERMINAL",0,0,"\r\n",,terminal_output
+426,522468,"slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",0,0,"",shellscript,tab
+427,528106,"TERMINAL",0,0,"salloc: Nodes hkn0403 are ready for job\r\n",,terminal_output
+428,528277,"TERMINAL",0,0,"source .venv/bin/accelerate\r\nsh slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh\r\n",,terminal_output
+429,529767,"TERMINAL",0,0,"]0;tum_cte0515@hkn0403:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/logs/logs_mihir/coinrun/tokenizer[?2004h[tum_cte0515@hkn0403 tokenizer]$ source .venv/bin/accelerate\r\n[?2004l\rbash: .venv/bin/accelerate: No such file or directory\r\n]0;tum_cte0515@hkn0403:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/logs/logs_mihir/coinrun/tokenizer[?2004h[tum_cte0515@hkn0403 tokenizer]$ sh slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh\r\n[?2004l\rsh: slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh: No such file or directory\r\n]0;tum_cte0515@hkn0403:/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/logs/logs_mihir/coinrun/tokenizer[?2004h[tum_cte0515@hkn0403 tokenizer]$ ",,terminal_output
+430,533110,"TERMINAL",0,0,"d",,terminal_output
+431,533581,"TERMINAL",0,0,"e",,terminal_output
+432,535138,"TERMINAL",0,0,"v",,terminal_output
+433,535365,"TERMINAL",0,0,"\r\n[?2004l\r]0;tum_cte0515@hkn0403:~/Projects/jasmine[?2004h[tum_cte0515@hkn0403 jasmine]$ ",,terminal_output
+434,535668,"TERMINAL",0,0,"dev",,terminal_output
+435,535948,"TERMINAL",0,0,"sh slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",,terminal_output
+436,536284,"TERMINAL",0,0,"\rource .venv/bin/accelerate",,terminal_output
+437,536890,"TERMINAL",0,0,"\r\n[?2004l\rbash: .venv/bin/accelerate: No such file or directory\r\n]0;tum_cte0515@hkn0403:~/Projects/jasmine[?2004h[tum_cte0515@hkn0403 jasmine]$ ",,terminal_output
+438,537365,"TERMINAL",0,0,"source .venv/bin/accelerate",,terminal_output
+439,537540,"TERMINAL",0,0,"dev",,terminal_output
+440,538048,"TERMINAL",0,0,"source .venv/bin/accelerate",,terminal_output
+441,538858,"TERMINAL",0,0,"",,terminal_output
+442,542000,"TERMINAL",0,0,"source .venv/bin/accelerate",,terminal_output
+443,542619,"TERMINAL",0,0,"",,terminal_output
+444,543719,"TERMINAL",0,0,"",,terminal_output
+445,543863,"TERMINAL",0,0,"",,terminal_output
+446,544014,"TERMINAL",0,0,"",,terminal_output
+447,544278,"TERMINAL",0,0,"",,terminal_output
+448,544423,"TERMINAL",0,0,"",,terminal_output
+449,544824,"TERMINAL",0,0,"",,terminal_output
+450,545254,"TERMINAL",0,0,"",,terminal_output
+451,545374,"TERMINAL",0,0,"tivate",,terminal_output
+452,545932,"TERMINAL",0,0,"\r\n[?2004l\r]0;tum_cte0515@hkn0403:~/Projects/jasmine[?2004h(jasmine) [tum_cte0515@hkn0403 jasmine]$ ",,terminal_output
+453,546405,"TERMINAL",0,0,"source .venv/bin/activate",,terminal_output
+454,546639,"TERMINAL",0,0,"celerate",,terminal_output
+455,547159,"TERMINAL",0,0,"dev",,terminal_output
+456,547780,"TERMINAL",0,0,"sh slurm/jobs/mihir/horeka/coinrun/train_tokenizer_single_gpu.sh",,terminal_output
+457,548519,"TERMINAL",0,0,"\r\n[?2004l\r#!/usr/bin/env bash\r\n\r\n#SBATCH --nodes=1\r\n#SBATCH --ntasks-per-node=4\r\n#SBATCH --time=00:20:00\r\n#SBATCH --partition=dev_accelerated\r\n#SBATCH --cpus-per-task=5\r\n#SBATCH --gres=gpu:1\r\n#SBATCH --output=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/logs/logs_mihir/coinrun/tokenizer/%x_%j.log\r\n#SBATCH --error=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/logs/logs_mihir/coinrun/tokenizer/%x_%j.log\r\n#SBATCH --job-name=train_tokenizer_1e-4\r\n#SBATCH --requeue\r\n#SBATCH --signal=b:usr1@300 # 5 min before timeout\r\n\r\n# Log the sbatch script\r\ncat $0\r\n\r\nmodule unload mpi/openmpi/5.0\r\nmodule unload devel/cuda/12.4\r\nsource .venv/bin/activate\r\n\r\narray_records_dir_train=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/data_coinrun/coinrun_episodes_test\r\n# array_records_dir_train=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/data_coinrun/coinrun_episodes_10m_gt_actions_split/train\r\n# array_records_dir_train=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/data_coinrun/coinrun_episodes_10m_chunked\r\narray_records_dir_val=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/data_coinrun/coinrun_episodes_10m_gt_actions_split/val\r\n\r\n\r\njob_name=$SLURM_JOB_NAME\r\nslurm_job_id=$SLURM_JOB_ID\r\n\r\nCHECKPOINT_DIR=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/checkpoints/big-runs/tokenizer/$job_name/$slurm_job_id\r\nmkdir -p $CHECKPOINT_DIR\r\n\r\nenv | grep SLURM\r\n\r\nsrun python train_tokenizer.py \\r\n    --save_ckpt \\r\n    --image_height=64 \\r\n    --image_width=64 \\r\n    --ckpt_dir $CHECKPOINT_DIR \\r\n    --batch_size=120 \\r\n    --init_lr=0 \\r\n    --max_lr=1e-4 \\r\n    --log_image_interval=250 \\r\n    --log_checkpoint_interval=250 \\r\n    --log \\r\n    --name=coinrun-tokenizer-dataset-test-$slurm_job_id \\r\n    --tags tokenizer coinrun dev \\r\n    --entity instant-uv \\r\n    --project jafar \\r\n    --warmup_steps 0 \\r\n    --wsd_decay_steps 0 \\r\n    --num_steps 250 \\r\n    --data_dir $array_records_dir_train\r\n",,terminal_output
+458,548758,"TERMINAL",0,0,"SLURM_STEP_NUM_TASKS=1\r\nSLURM_JOB_USER=tum_cte0515\r\nSLURM_TASKS_PER_NODE=1\r\nSLURM_JOB_UID=999226\r\nSLURM_TASK_PID=67647\r\nSLURM_JOB_GPUS=0\r\nSLURM_LOCALID=0\r\nSLURM_SUBMIT_DIR=/hkfs/work/workspace/scratch/tum_ind3695-jafa_ws_shared/logs/logs_mihir/coinrun/tokenizer\r\nSLURMD_NODENAME=hkn0403\r\nSLURM_JOB_START_TIME=1757499919\r\nSLURM_STEP_NODELIST=hkn0403\r\nSLURM_CLUSTER_NAME=hk\r\nSLURM_JOB_END_TIME=1757503519\r\nSLURM_PMI2_SRUN_PORT=32775\r\nSLURM_CPUS_ON_NODE=8\r\nSLURM_JOB_CPUS_PER_NODE=8\r\nSLURM_GPUS_ON_NODE=1\r\nSLURM_GTIDS=0\r\nSLURM_JOB_PARTITION=dev_accelerated\r\nSLURM_TRES_PER_TASK=cpu=8\r\nSLURM_OOM_KILL_STEP=0\r\nSLURM_JOB_NUM_NODES=1\r\nSLURM_STEPID=4294967290\r\nSLURM_JOBID=3482653\r\nSLURM_PTY_PORT=44507\r\nSLURM_JOB_QOS=normal\r\nSLURM_LAUNCH_NODE_IPADDR=10.0.7.201\r\nSLURM_PTY_WIN_ROW=32\r\nSLURM_PMI2_PROC_MAPPING=(vector,(0,1,1))\r\nSLURMD_DEBUG=2\r\nSLURM_PROCID=0\r\nSLURM_CPUS_PER_TASK=8\r\nSLURM_TOPOLOGY_ADDR=hkibb.hkibbi1.hkibbi1e11.hkn0403\r\nSLURM_TOPOLOGY_ADDR_PATTERN=switch.switch.switch.node\r\nSLURM_SRUN_COMM_HOST=10.0.7.201\r\nSLURM_SCRIPT_CONTEXT=prolog_task\r\nSLURM_PTY_WIN_COL=129\r\nSLURM_NODELIST=hkn0403\r\nSLURM_SRUN_COMM_PORT=43229\r\nSLURM_STEP_ID=4294967290\r\nSLURM_JOB_ACCOUNT=hk-project-p0023960\r\nSLURM_PRIO_PROCESS=0\r\nSLURM_NNODES=1\r\nSLURM_SUBMIT_HOST=hkn1993.localdomain\r\nSLURM_JOB_ID=3482653\r\nSLURM_NODEID=0\r\nSLURM_STEP_NUM_NODES=1\r\nSLURM_STEP_TASKS_PER_NODE=1\r\nSLURM_MPI_TYPE=pmi2\r\nSLURM_PMI2_STEP_NODES=hkn0403\r\nSLURM_CONF=/etc/slurm/slurm.conf\r\nSLURM_JOB_NAME=interactive\r\nSLURM_STEP_LAUNCHER_PORT=43229\r\nSLURM_JOB_GID=502226\r\nSLURM_JOB_NODELIST=hkn0403\r\n",,terminal_output
+459,548839,"TERMINAL",0,0,"GpuFreq=control_disabled\r\n",,terminal_output
+460,571084,"TERMINAL",0,0,"wandb: Currently logged in as: mihir-mahajan2002 (instant-uv) to https://api.wandb.ai. Use `wandb login --relogin` to force relogin\r\n",,terminal_output
+461,572176,"TERMINAL",0,0,"wandb: creating run\r\nwandb: Tracking run with wandb version 0.21.3\r\nwandb: Run data is saved locally in /hkfs/home/project/hk-project-p0023960/tum_cte0515/Projects/jasmine/wandb/run-20250910_122628-qot85m4u\r\nwandb: Run `wandb offline` to turn off syncing.\r\nwandb: Syncing run coinrun-tokenizer-dataset-test-3482653\r\nwandb: ⭐️ View project at https://wandb.ai/instant-uv/jafar\r\nwandb: 🚀 View run at https://wandb.ai/instant-uv/jafar/runs/qot85m4u\r\n",,terminal_output
+462,573635,"TERMINAL",0,0,"Running on 1 devices.\r\nCounting all components: ['decoder', 'encoder', 'vq']\r\nParameter counts:\r\n{'decoder': 16858736, 'encoder': 16858752, 'vq': 32768, 'total': 33750256}\r\nStarting training from step 0...\r\n",,terminal_output

927a8af5474e5654810c00ce2e09fd2de87d3e5722f33fa1090d867db114e403/crowd-code-9c2d9ac2-2076-4ff2-8381-5264acd089541759350296590-2025_10_01-22.25.31.836/source.csv

@@ -0,0 +1,8 @@
+Sequence,Time,File,RangeOffset,RangeLength,Text,Language,Type
+1,5,"slurm/jobs/franz/berlin/coinrun/submission_debug/coinrun_dynamics_base_patch_size_4_action_prepend.sh",0,0,"#!/usr/bin/env bash\n\n#SBATCH --nodes=1\n#SBATCH --ntasks-per-node=1\n#SBATCH --time=24:00:00\n#SBATCH --cpus-per-task=8\n#SBATCH --gres=gpu:1\n#SBATCH --output=/fast/project/HFMI_SynergyUnit/jafar_ws/logs/franz/coinrun/dynamics/%x_%j.log\n#SBATCH --error=/fast/project/HFMI_SynergyUnit/jafar_ws/logs/franz/coinrun/dynamics/%x_%j.log\n#SBATCH --job-name=dynamics_coinrun_mila_submission_patch_size_4_action_prepend_branch\n#SBATCH --requeue\n#SBATCH --signal=b:usr1@300 # 5 min before timeout\n\n# --- signal trap to requeue job before timeout ---\nrequeue_job() {\n    echo ""[$(date)] caught sigusr1 (timeout warning), requeueing slurm job $SLURM_JOB_ID...""\n    # optional: trigger checkpoint saving here\n    # e.g., touch $checkpoint_dir/requeue_trigger\n    scontrol requeue $SLURM_JOB_ID\n    exit 0\n}\n\ntrap requeue_job sigusr1\n\n# set checkpoint flag based on restart count\nrestart_count=$(scontrol show job $SLURM_JOB_ID | grep -o 'Restarts=[0-9]*' | cut -d'=' -f2)\n\nif [ $restart_count -eq 0 ]; then\n    restore_ckpt_flag=""--no-restore-ckpt""\nelse\n    restore_ckpt_flag=""--restore-ckpt""\nfi\n\n\n\n# Log the sbatch script\ncat $0\n\nsource .venv/bin/activate\n\njob_name=$SLURM_JOB_NAME\nslurm_job_id=$SLURM_JOB_ID\n\ntags=""coinrun dynamics 500m_dataset mila_submission debug patch_size_4 action_prepend_branch""\n\narray_records_dir=""/fast/project/HFMI_SynergyUnit/jafar_ws/data/coinrun/array_records_500m_seed_w_increment""\ntokenizer_ckpt_dir=""/fast/project/HFMI_SynergyUnit/jafar_ws/checkpoints/coinrun/tokenizer/tokenizer_coinrun_500m_dataset_29490/""\nCHECKPOINT_DIR=""/fast/project/HFMI_SynergyUnit/jafar_ws/checkpoints/coinrun/dynamics/${job_name}/${slurm_job_id}""\nmkdir -p $CHECKPOINT_DIR\n\nenv | grep SLURM\n\n# Check if the current branch is the main branch\ncurrent_branch=$(git rev-parse --abbrev-ref HEAD)\nif [ ""$current_branch"" != ""prepend-action-maskgit"" ]; then\n    echo ""This script must be run from the prepend-action-maskgit branch. Current branch is $current_branch. Exiting.""\n    exit 1\nfi\n\nsrun python jasmine/train_dynamics.py \\n    --patch_size=4 \\n    --save_ckpt \\n    $restore_ckpt_flag \\n    --wandb_id $SLURM_JOB_ID \\n    --ckpt_dir $CHECKPOINT_DIR \\n    --name=""${job_name}_${slurm_job_id}"" \\n    --tags ${tags} \\n    --entity instant-uv \\n    --project jafar \\n    --tokenizer_checkpoint=""${tokenizer_ckpt_dir}"" \\n    --val_data_dir=""${array_records_dir}/val"" \\n    --data_dir=""${array_records_dir}/train"" &\n\nchild_pid=$!\n\nwait $child_pid\n\n",shellscript,tab
+2,330,"extension-output-pdoom-org.crowd-code-#1-crowd-code",0,0,"10:25:31 PM [info] Activating crowd-code\n10:25:31 PM [info] Recording started\n10:25:31 PM [info] Initializing git provider using file system watchers...\n",Log,tab
+3,567,"extension-output-pdoom-org.crowd-code-#1-crowd-code",153,0,"10:25:32 PM [info] Git repository found\n10:25:32 PM [info] Git provider initialized successfully\n10:25:32 PM [info] Initial git state: [object Object]\n",Log,content
+4,373443,"slurm/jobs/franz/berlin/coinrun/submission_debug/coinrun_dynamics_base_patch_size_4_action_prepend.sh",0,0,"",shellscript,tab
+5,373446,"slurm/jobs/franz/berlin/coinrun/submission_debug/coinrun_dynamics_base_patch_size_4_action_prepend.sh",955,0,"",shellscript,selection_mouse
+6,1320333,"slurm/jobs/franz/berlin/coinrun/submission_debug/coinrun_dynamics_base_patch_size_4_action_prepend.sh",0,0,"Switched from branch 'prepend-action-maskgit' to 'main'",shellscript,git_branch_checkout
+7,1360341,"slurm/jobs/franz/berlin/coinrun/submission_debug/coinrun_dynamics_base_patch_size_4_action_prepend.sh",0,0,"Switched from branch 'main' to 'dynamics_coinrun_500m_dataset_29519'",shellscript,git_branch_checkout