Update app.py

app.py

@@ -4,29 +4,25 @@ from peft import PeftModel
 import torch
 
 # -----------------------------
-# Model config
+# CPU-friendly model
 # -----------------------------
-MODEL_NAME = "unsloth/qwen2.5-math-1.5b-bnb-4bit"
+MODEL_NAME = "tiiuae/falcon-7b-instruct"  # smaller CPU-friendly model
 
 print("Loading tokenizer...")
 tokenizer = AutoTokenizer.from_pretrained(MODEL_NAME, use_fast=False)
 
-print("Loading model in 4-bit...")
+print("Loading model...")
+base_model = AutoModelForCausalLM.from_pretrained(
+    MODEL_NAME,
+    device_map=None,  # CPU only
+    torch_dtype=torch.float32
+)
+
+# Load LoRA if exists (optional)
 try:
-    base_model = AutoModelForCausalLM.from_pretrained(
-        MODEL_NAME,
-        device_map="auto",
-        dtype=torch.float16,  # newer transformers prefers dtype
-        low_cpu_mem_usage=True
-    )
-    # Load LoRA adapter if exists
-    try:
-        model = PeftModel.from_pretrained(base_model, MODEL_NAME, device_map="auto")
-    except:
-        model = base_model
-except Exception as e:
-    print("Error loading model:", e)
-    raise e
+    model = PeftModel.from_pretrained(base_model, MODEL_NAME, device_map=None)
+except:
+    model = base_model
 
 model.eval()
 
@@ -34,7 +30,7 @@ model.eval()
 # Response function
 # -----------------------------
 def respond(message, history, system_message, max_tokens, temperature, top_p):
-    # Limit max tokens for safety
+    # Limit max tokens for CPU safety
     if max_tokens > 128:
         max_tokens = 128
 
@@ -44,7 +40,7 @@ def respond(message, history, system_message, max_tokens, temperature, top_p):
         prompt += f"User: {h['content']}\n"
     prompt += f"User: {message}\nBot:"
 
-    inputs = tokenizer(prompt, return_tensors="pt").to(model.device)
+    inputs = tokenizer(prompt, return_tensors="pt")
 
     gen_config = GenerationConfig(
         max_new_tokens=max_tokens,
@@ -66,7 +62,7 @@ chatbot = gr.ChatInterface(
     type="messages",
     additional_inputs=[
         gr.Textbox(value="You are a friendly Chatbot.", label="System message"),
-        gr.Slider(minimum=1, maximum=2048, value=64, step=1, label="Max new tokens"),
+        gr.Slider(minimum=1, maximum=128, value=64, step=1, label="Max new tokens"),
         gr.Slider(minimum=0.1, maximum=2.0, value=0.7, step=0.1, label="Temperature"),
         gr.Slider(minimum=0.1, maximum=1.0, value=0.95, step=0.05, label="Top-p (nucleus sampling)"),
     ],
@@ -77,3 +73,4 @@ with gr.Blocks() as demo:
 
 if __name__ == "__main__":
     demo.launch(server_name="0.0.0.0", server_port=7860)
+