Text Generation
Transformers
Safetensors
gpt_bigcode
code
Eval Results (legacy)
text-generation-inference
4-bit precision
gptq
Instructions to use TheBloke/starcoder-GPTQ with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use TheBloke/starcoder-GPTQ with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="TheBloke/starcoder-GPTQ")# Load model directly from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("TheBloke/starcoder-GPTQ") model = AutoModelForCausalLM.from_pretrained("TheBloke/starcoder-GPTQ") - Notebooks
- Google Colab
- Kaggle
- Local Apps
- vLLM
How to use TheBloke/starcoder-GPTQ with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "TheBloke/starcoder-GPTQ" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "TheBloke/starcoder-GPTQ", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker
docker model run hf.co/TheBloke/starcoder-GPTQ
- SGLang
How to use TheBloke/starcoder-GPTQ 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 "TheBloke/starcoder-GPTQ" \ --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": "TheBloke/starcoder-GPTQ", "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 "TheBloke/starcoder-GPTQ" \ --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": "TheBloke/starcoder-GPTQ", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }' - Docker Model Runner
How to use TheBloke/starcoder-GPTQ with Docker Model Runner:
docker model run hf.co/TheBloke/starcoder-GPTQ
GPU memory usage/requirement?
#1
by Bilibili - opened
Thanks for this work!
Since the original StarCoder requires 60+ GB GPU RAM for inference, I wonder what about the GPTQ version, and could the model run inference on V100-32G?
Bilibili changed discussion title from GPU memory usage peak? to GPU memory usage requirement?
Bilibili changed discussion title from GPU memory usage requirement? to GPU memory usage/requirement?
I'm totally new to GPTQ and am not exactly sure how to calculate the exacts, but it seems happy with 20-30 gigs from my CPU's ram, and I have only 12 gigs used in my GPU.
Yes 32GB is more than enough VRAM for nearly any model in GPTQ. This one needs around 12GB yeah