Instructions to use bofenghuang/vigostral-7b-chat with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use bofenghuang/vigostral-7b-chat with Transformers:

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

pipe = pipeline("text-generation", model="bofenghuang/vigostral-7b-chat")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("bofenghuang/vigostral-7b-chat")
model = AutoModelForCausalLM.from_pretrained("bofenghuang/vigostral-7b-chat", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=40)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use bofenghuang/vigostral-7b-chat with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "bofenghuang/vigostral-7b-chat"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "bofenghuang/vigostral-7b-chat",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/bofenghuang/vigostral-7b-chat

SGLang

How to use bofenghuang/vigostral-7b-chat 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 "bofenghuang/vigostral-7b-chat" \
    --host 0.0.0.0 \
    --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "bofenghuang/vigostral-7b-chat",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

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 "bofenghuang/vigostral-7b-chat" \
        --host 0.0.0.0 \
        --port 30000
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:30000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "bofenghuang/vigostral-7b-chat",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use bofenghuang/vigostral-7b-chat with Docker Model Runner:
```
docker model run hf.co/bofenghuang/vigostral-7b-chat
```

Running an inference server using Docker + vLLM

by YorelNation - opened Oct 18, 2023

Discussion

YorelNation

Oct 18, 2023

Hi,

Would it be possible to somehow deploy vigostral the same way we can deploy mistral via their recommended method: https://docs.mistral.ai/quickstart
Can I simply run:

docker run --gpus all
-e HF_TOKEN=$HF_TOKEN -p 8000:8000
ghcr.io/mistralai/mistral-src/vllm:latest
--host 0.0.0.0
--model bofenghuang/vigostral-7b-chat

I don't have the hardware to try this yet that's why I'm asking :)

Thanks

bofenghuang

Owner Oct 18, 2023

Hi,

Thanks for your message. I will look into it :)

bofenghuang

Owner Oct 25, 2023

Hi @YorelNation ,

The Mistral AI version has not yet been updated to support the prompt format of the Vigostral model.

However, I have managed to create another Docker image that also leverages vLLM for inference. You can use it as follows:

# Launch inference engine
docker run --gpus '"device=0"' \
    -e HF_TOKEN=$HF_TOKEN -p 8000:8000 \
    ghcr.io/bofenghuang/vigogne/vllm:latest \
    --host 0.0.0.0 \
    --model bofenghuang/vigostral-7b-chat

# Launch inference engine on mutli-GPUs (4 here)
docker run --gpus all \
    -e HF_TOKEN=$HF_TOKEN -p 8000:8000 \
    ghcr.io/bofenghuang/vigogne/vllm:latest \
    --host 0.0.0.0 \
    --tensor-parallel-size 4 \
    --model bofenghuang/vigostral-7b-chat

# Launch inference engine using the quantized AWQ version
# Note only supports Ampere or newer GPUs
docker run --gpus '"device=0"' \
    -e HF_TOKEN=$HF_TOKEN -p 8000:8000 \
    ghcr.io/bofenghuang/vigogne/vllm:latest \
    --host 0.0.0.0 \
    --quantization awq \
    --model TheBloke/Vigostral-7B-Chat-AWQ

# Launch inference engine using the downloaded weights
docker run --gpus '"device=0"' \
    -p 8000:8000 \
    -v /path/to/model/:/mnt/model/ \
    ghcr.io/bofenghuang/vigogne/vllm:latest \
    --host 0.0.0.0 \
    --model="/mnt/model/"

YorelNation

Oct 25, 2023

Thanks ! Will try this

Upload images, audio, and videos by dragging in the text input, pasting, or clicking here.

Tap or paste here to upload images

· Sign up or log in to comment