Instructions to use kaitchup/Mistral-NeMo-Minitron-8B-Base-AutoRound-GPTQ-sym-4bit with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use kaitchup/Mistral-NeMo-Minitron-8B-Base-AutoRound-GPTQ-sym-4bit with Transformers:

# Use a pipeline as a high-level helper
from transformers import pipeline

pipe = pipeline("text-generation", model="kaitchup/Mistral-NeMo-Minitron-8B-Base-AutoRound-GPTQ-sym-4bit")

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("kaitchup/Mistral-NeMo-Minitron-8B-Base-AutoRound-GPTQ-sym-4bit")
model = AutoModelForCausalLM.from_pretrained("kaitchup/Mistral-NeMo-Minitron-8B-Base-AutoRound-GPTQ-sym-4bit", device_map="auto")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use kaitchup/Mistral-NeMo-Minitron-8B-Base-AutoRound-GPTQ-sym-4bit with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "kaitchup/Mistral-NeMo-Minitron-8B-Base-AutoRound-GPTQ-sym-4bit"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "kaitchup/Mistral-NeMo-Minitron-8B-Base-AutoRound-GPTQ-sym-4bit",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/kaitchup/Mistral-NeMo-Minitron-8B-Base-AutoRound-GPTQ-sym-4bit

SGLang

How to use kaitchup/Mistral-NeMo-Minitron-8B-Base-AutoRound-GPTQ-sym-4bit 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 "kaitchup/Mistral-NeMo-Minitron-8B-Base-AutoRound-GPTQ-sym-4bit" \
    --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": "kaitchup/Mistral-NeMo-Minitron-8B-Base-AutoRound-GPTQ-sym-4bit",
		"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 "kaitchup/Mistral-NeMo-Minitron-8B-Base-AutoRound-GPTQ-sym-4bit" \
        --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": "kaitchup/Mistral-NeMo-Minitron-8B-Base-AutoRound-GPTQ-sym-4bit",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use kaitchup/Mistral-NeMo-Minitron-8B-Base-AutoRound-GPTQ-sym-4bit with Docker Model Runner:
```
docker model run hf.co/kaitchup/Mistral-NeMo-Minitron-8B-Base-AutoRound-GPTQ-sym-4bit
```

Mistral-NeMo-Minitron-8B-Base-AutoRound-GPTQ-sym-4bit / README.md

bnjmnmarie

Update README.md

798fb2d verified about 2 years ago

preview code

Raw History Blame Contribute Delete

1.03 kB

	---
	language:
	- en
	library_name: transformers
	tags:
	- AutoRound
	license: apache-2.0
	---

	Warning: This model poorly performs. I ran the quantization three times but it never produced a good model. I recommend using the asymmetric quantization ([kaitchup/Mistral-NeMo-Minitron-8B-Base-AutoRound-GPTQ-asym-4bit](https://huggingface.co/kaitchup/Mistral-NeMo-Minitron-8B-Base-AutoRound-GPTQ-asym-4bit)) version instead.


	## Model Details

	This is [nvidia/Mistral-NeMo-Minitron-8B-Base](https://huggingface.co/nvidia/Mistral-NeMo-Minitron-8B-Base) quantized with AutoRound (symmetric quantization) to 4-bit. The model has been created, tested, and evaluated by The Kaitchup. It is compatible with the main inference frameworks, e.g., TGI and vLLM.

	Details on the quantization process and evaluation:
	[Mistral-NeMo: 4.1x Smaller with Quantized Minitron](https://kaitchup.substack.com/p/mistral-nemo-41x-smaller-with-quantized)


	- Developed by: [The Kaitchup](https://kaitchup.substack.com/)
	- License: Apache license 2.0