Instructions to use codellama/CodeLlama-7b-hf with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use codellama/CodeLlama-7b-hf with Transformers:

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

pipe = pipeline("text-generation", model="codellama/CodeLlama-7b-hf")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("codellama/CodeLlama-7b-hf")
model = AutoModelForCausalLM.from_pretrained("codellama/CodeLlama-7b-hf")

Inference
Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use codellama/CodeLlama-7b-hf with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "codellama/CodeLlama-7b-hf"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "codellama/CodeLlama-7b-hf",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/codellama/CodeLlama-7b-hf

SGLang

How to use codellama/CodeLlama-7b-hf 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 "codellama/CodeLlama-7b-hf" \
    --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": "codellama/CodeLlama-7b-hf",
		"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 "codellama/CodeLlama-7b-hf" \
        --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": "codellama/CodeLlama-7b-hf",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use codellama/CodeLlama-7b-hf with Docker Model Runner:
```
docker model run hf.co/codellama/CodeLlama-7b-hf
```

Error in deployment in sagemaker

by Akanshu - opened Aug 25, 2023

Discussion

Akanshu

Aug 25, 2023

•

edited Aug 25, 2023

RuntimeError(f"weight {tensor_name} does not exist")
RuntimeError: weight model.layers.0.self_attn.rotary_emb.inv_freq does not exist
Error: ShardCannotStart

Thankyou in advance and please help in this deployment.

akgq

Aug 28, 2023

Facing the same issue Error: ShardCannotStart while deploying CodeLlama via hugging face

Choms

Aug 28, 2023

Hey, you need the mainline version of the 🤗 transformers from git to run this (https://huggingface.co/codellama/CodeLlama-7b-hf#model-use), there's no container for it yet on sagemaker (I guess you both are using the 0.9.3 container), you'll have to run it outside of sagemaker or load it on a notebook instance directly (that's what I'm doing for now, until this is supported)

Akanshu

Aug 28, 2023

I just made the notebook instance and inside that i have created a jupyter notebook and ran this code…can you please elaborate how to deploy.

Choms

Aug 28, 2023

That's about as far as I've got, I'm following the documentation here: https://huggingface.co/docs/transformers/main/model_doc/code_llama
plus the pip install from git on the readme from this model, then just use the notebook to play with it, as said there's no easy way to deploy it as an actual interference endpoint (you could build your own container with the required versions though), good luck!

PS: you can use thebloke's gptq build and run it on multi GPU if you pip install auto-gptq optimum

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