Instructions to use MarinaraSpaghetti/NemoMix-Unleashed-12B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use MarinaraSpaghetti/NemoMix-Unleashed-12B with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="MarinaraSpaghetti/NemoMix-Unleashed-12B")# Load model directly from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("MarinaraSpaghetti/NemoMix-Unleashed-12B") model = AutoModelForCausalLM.from_pretrained("MarinaraSpaghetti/NemoMix-Unleashed-12B") - Inference
- Notebooks
- Google Colab
- Kaggle
- Local Apps
- vLLM
How to use MarinaraSpaghetti/NemoMix-Unleashed-12B with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "MarinaraSpaghetti/NemoMix-Unleashed-12B" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "MarinaraSpaghetti/NemoMix-Unleashed-12B", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker
docker model run hf.co/MarinaraSpaghetti/NemoMix-Unleashed-12B
- SGLang
How to use MarinaraSpaghetti/NemoMix-Unleashed-12B 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 "MarinaraSpaghetti/NemoMix-Unleashed-12B" \ --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": "MarinaraSpaghetti/NemoMix-Unleashed-12B", "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 "MarinaraSpaghetti/NemoMix-Unleashed-12B" \ --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": "MarinaraSpaghetti/NemoMix-Unleashed-12B", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }' - Docker Model Runner
How to use MarinaraSpaghetti/NemoMix-Unleashed-12B with Docker Model Runner:
docker model run hf.co/MarinaraSpaghetti/NemoMix-Unleashed-12B
Samplers, templates etc. - help, please :-D
So what do we actually do in the end to make it work?
Samplers: Temperature 1.0-1.25, Min P 0.01-0.1, DRY 0.8/1.75/2/0 DRY. All the rest - nullified. That's clear.
Context Template: ChatML? Chat ML-Names? Mistral?
Instruct: Mistral or Mistral + we need to do something in Silly Tavern in addition to selecting Mistral? Do we download your settings and use those instead because the SillyTavern ones will be wrong the way they go in a vanilla form? I don't understand how exactly it works - with your note on template from Mistral employee. I'm lost here, sorry for a bother :-P
Try this https://huggingface.co/Virt-io/SillyTavern-Presets (Mistral presets, both Context and Instruct)
Try this https://huggingface.co/Virt-io/SillyTavern-Presets (Mistral presets, both Context and Instruct)
Thank you very much. They work relatively well on Q8 but give me a complete rubbish at Q6. Any idea to why it happens? o.O
Still, great work in general.
Turn off system as user in instruct mode sequence
Turn off system as user in instruct mode sequence
You mean the system prompt? Why? If not, something else?
Also, you can add a little to rep pen, something like 1,05. For me, it helps model to not fall into repetition with certan cards.
Hey @sphiratrioth666 , I see that someone else already offered help, and I hope it works better! You can also use my recommended settings for the Story String and Instruct: https://huggingface.co/MarinaraSpaghetti/SillyTavern-Settings/tree/main/Customized/Mistral%20Improved%202%20Electric%20Boogaloo
Hey @sphiratrioth666 , I see that someone else already offered help, and I hope it works better! You can also use my recommended settings for the Story String and Instruct: https://huggingface.co/MarinaraSpaghetti/SillyTavern-Settings/tree/main/Customized/Mistral%20Improved%202%20Electric%20Boogaloo
Thx! I've tried both yours and Virt-io ones. Both work well :-) Thank you very much for response, you didn't have to go as far as those screenshots, much appreciated!



