Text Generation
Transformers
Safetensors
English
llama
rp
erp
chat
storywriting
text-generation-inference
Instructions to use MarsupialAI/KitchenSink_103b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use MarsupialAI/KitchenSink_103b with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="MarsupialAI/KitchenSink_103b")# Load model directly from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("MarsupialAI/KitchenSink_103b") model = AutoModelForCausalLM.from_pretrained("MarsupialAI/KitchenSink_103b", device_map="auto") - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use MarsupialAI/KitchenSink_103b with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "MarsupialAI/KitchenSink_103b" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "MarsupialAI/KitchenSink_103b", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker
docker model run hf.co/MarsupialAI/KitchenSink_103b
- SGLang
How to use MarsupialAI/KitchenSink_103b with SGLang:
Install from pip and serve model
# Install SGLang from pip: pip install sglang # Start the SGLang server: python3 -m sglang.launch_server \ --model-path "MarsupialAI/KitchenSink_103b" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "MarsupialAI/KitchenSink_103b", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker images
docker run --gpus all \ --shm-size 32g \ -p 30000:30000 \ -v ~/.cache/huggingface:/root/.cache/huggingface \ --env "HF_TOKEN=<secret>" \ --ipc=host \ lmsysorg/sglang:latest \ python3 -m sglang.launch_server \ --model-path "MarsupialAI/KitchenSink_103b" \ --host 0.0.0.0 \ --port 30000 # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:30000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "MarsupialAI/KitchenSink_103b", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }' - Docker Model Runner
How to use MarsupialAI/KitchenSink_103b with Docker Model Runner:
docker model run hf.co/MarsupialAI/KitchenSink_103b
| license: llama2 | |
| language: | |
| - en | |
| tags: | |
| - rp | |
| - erp | |
| - chat | |
| - storywriting | |
| # Kitchen Sink 103b | |
|  | |
| This model is a rotating-stack merge of three 70b models in a 103b (120 layer) configuration inspired by Venus 103b. The result of | |
| this "frankenmerge" is a large model that contains a little bit of everything - including the kitchen sink. RP, chat, storywriting, | |
| and instruct are all well supported. It may or may not code well - I lack the expertise to test it in that capacity, but considering | |
| the source models, I suspect it is unlikely. | |
| Component models for the rotating stack are | |
| - royallab/Aetheria-L2-70B | |
| - lizpreciatior/lzlv_70b_fp16_hf | |
| - Sao10K/WinterGoddess-1.4x-70B-L2 | |
| Components of those models are purported to include: Nous-Hermes-Llama2-70b, Xwin-LM-7B-V0.1, Mythospice-70b, Euryale-1.3-L2-70B, | |
| tulu-2-dpo-70b, GOAT-70B-Storytelling, Platypus2-70B-instruct, Lila-70B, SunsetBoulevard, and some private LoRAs. | |
| As all components are based on Llama2 70b, native context length is 4k tokens. Coherency out to 8k is extremely good with rope scaling, | |
| but starts to decline beyond that. | |
| This model is uncensored and perfectly capable of generating objectionable material. However, it is not an explicitely-NSFW model, | |
| and it has never "gone rogue" and tried to insert NSFW content into SFW prompts in my experience. As with any LLM, no factual claims | |
| made by the model should be taken at face value. You know that boilerplate safety disclaimer that most professional models have? | |
| Assume this has it too. This model is for entertainment purposes only. | |
| FP16 and Q4_K_S GGUFs: https://huggingface.co/MarsupialAI/KitchenSink_103b_GGUF_Q4KS_FP16 | |
| Q2 and Q3 GGUFs: https://huggingface.co/bcse/KitchenSink-103b-GGUF | |
| iMatrix GGUFs: https://huggingface.co/MarsupialAI/KitchenSink_103b_iMatrix_GGUF | |
| EXL2 quants: https://huggingface.co/AzureBlack/KitchenSink_103b-2.5bpw-6h-exl2 https://huggingface.co/AzureBlack/KitchenSink_103b-3.5bpw-6h-exl2 https://huggingface.co/AzureBlack/KitchenSink_103b-4.5bpw-6h-exl2 | |
| If you create additional quants, please let me know and I will link them here as well. | |
| # Sample output | |
| ``` | |
| {{[INPUT]}} | |
| Write a detailed and humorous story about a cute and fluffy bunny that goes to a Gwar concert. | |
| {{[OUTPUT]}} | |
| Once upon a time, in a magical land called Fluffyville, there lived a cute and fluffy little bunny named Fuzzlesworth. Fuzzlesworth | |
| was your average, everyday bunny; he hopped around, nibbled on carrots, and generally lived a quiet, peaceful life. However, | |
| Fuzzlesworth had a secret: he was a closet metalhead. When no one else was around, he would crank up his tiny bunny-sized speakers | |
| and rock out to the loudest, most aggressive metal music he could find. His favorite band of all time was Gwar, the infamous | |
| intergalactic warriors who sprayed fake blood and played insane heavy metal. | |
| One day, Fuzzlesworth stumbled upon a flyer nailed to a tree trunk in the meadow that announced that Gwar would be performing at a | |
| nearby human concert venue. His fluffy little bunny heart skipped a beat! He simply had to go! But, he knew his friends and family | |
| would never understand his love for such an aggressive and intense music, so he decided to keep his plan a secret. | |
| The big night finally arrived, and Fuzzlesworth, dressed in his best leather bunny-sized studded vest and spiked collar, hopped | |
| his way to the concert venue. He stood outside the door, shaking with excitement and nervousness. The line of people waiting to | |
| enter was intimidating: there were tall, muscular men and women covered in spikes, chains, and tattoos, and Fuzzlesworth felt like | |
| a small, fluffy speck in a sea of metal. | |
| But Fuzzlesworth was determined to see his favorite band, so he summoned up his bunny-sized courage and hopped to the front of the | |
| line. The bouncer, a hulking behemoth of a man, looked down at Fuzzlesworth and raised a thick, hairy eyebrow. | |
| "Umm...I'm here for the Gwar concert," Fuzzlesworth squeaked, holding up his ticket. | |
| The bouncer stared at him incredulously for a moment, then let out a hearty laugh. "Aw, look at the little bunny in his metal gear! | |
| Alright, go on in, but stay out of trouble, you hear?" | |
| Fuzzlesworth's tail was wagging so hard, it could've powered a small wind turbine. He bounced his way inside the venue, taking in | |
| the sights and sounds of the raucous crowd. The air was thick with the scent of sweat, cheap beer, and dry-ice smoke. He found a | |
| small nook near the stage where he could see everything and waited for the show to start. | |
| Finally, the lights dimmed, and the band took the stage. Fuzzlesworth's eyes went wide as the members of Gwar, adorned in their | |
| grotesque, otherworldly costumes, began to thrash and wail on their instruments. He couldn't believe his fluffy little ears! He | |
| was actually at a Gwar concert! | |
| As the set went on, Fuzzlesworth got more and more into the music, headbanging and thrashing around in his own adorable bunny way. | |
| The other concertgoers couldn't help but smile and chuckle at the sight of the tiny bunny moshing it up in the front row. | |
| During one particularly intense song, Fuzzlesworth found himself lifted up by the mosh pit and deposited on stage. He looked out | |
| at the roaring crowd, his little bunny heart pounding. In a moment of metal madness, he jumped up on the speaker stack and began | |
| to shred an air-guitar solo, much to the amusement of the band and audience alike. | |
| At the end of the set, the band's lead singer, Oderus Urungus, spotted Fuzzlesworth and motioned him over. Fuzzlesworth, overcome | |
| with joy, hopped over and was lifted up onto the frontman's massive, spiked shoulder. | |
| "Ladies and gentlemen," Oderus bellowed into the microphone, "I'd like to introduce our special guest tonight! Give it up for | |
| Fuzzlesworth the Metal Bunny!" | |
| The crowd went wild, cheering and clapping as Fuzzlesworth soaked up the attention. He had never felt so alive! | |
| After the show, Fuzzlesworth got to meet the band, posing for photos and getting autographs. As he hopped back home that night, | |
| his head still buzzing with metal and excitement, he knew that his life would never be the same. He was no longer just a small, | |
| fluffy bunny from Fluffyville; he was Fuzzlesworth the Metal Bunny, and he wasn't afraid to rock out! | |
| ``` | |
| # Prompt format | |
| Seems to have the strongest affinity for Alpaca prompts, but Vicuna works as well. Considering the variety of components, most | |
| formats will probbaly work to some extent. | |
| # WTF is a rotating-stack merge? | |
| Inspired by Undi's experiments with stacked merges, Jeb Carter found that output quality and model initiative could be significantly | |
| improved by reversing the model order in the stack, and then doing a linear merge between the original and reversed stacks. That is | |
| what I did here. I created three passthrough stacked merges using the three source models (rotating the model order in each stack), | |
| and then doing a linear merge of all three stacks. The exact merge configs can be found in the recipe.txt file. | |