Instructions to use TheBloke/WizardCoder-15B-1.0-GPTQ with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use TheBloke/WizardCoder-15B-1.0-GPTQ with Transformers:

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

pipe = pipeline("text-generation", model="TheBloke/WizardCoder-15B-1.0-GPTQ")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("TheBloke/WizardCoder-15B-1.0-GPTQ")
model = AutoModelForCausalLM.from_pretrained("TheBloke/WizardCoder-15B-1.0-GPTQ")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use TheBloke/WizardCoder-15B-1.0-GPTQ with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "TheBloke/WizardCoder-15B-1.0-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/WizardCoder-15B-1.0-GPTQ",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/TheBloke/WizardCoder-15B-1.0-GPTQ

SGLang

How to use TheBloke/WizardCoder-15B-1.0-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/WizardCoder-15B-1.0-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/WizardCoder-15B-1.0-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/WizardCoder-15B-1.0-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/WizardCoder-15B-1.0-GPTQ",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use TheBloke/WizardCoder-15B-1.0-GPTQ with Docker Model Runner:
```
docker model run hf.co/TheBloke/WizardCoder-15B-1.0-GPTQ
```

8bit GPTQ model

by 0dd1er - opened Jun 17, 2023

Discussion

0dd1er

Jun 17, 2023

Hi,
thank you for your valuable work!!
Do you plan to release a 8bit GPTQ version of WizardCoder-15B?
I assume that would fit perferctly on a RTX4090 :-)

TheBloke

Owner Jun 20, 2023

That's an interesting thought. I'd not thought about doing that but yeah I certainly could.

Fairly soon I plan to go back and start doing more GPTQs with other parameters, so I will add this to the list to do then.

rombodawg

Jul 6, 2023

I would 100% love a 8-bit model too in gptq! GPTQ is my favorite type of model since oobagooba doesnt support ggml currently, but 4-bit quantization just seems to harm the quality of the model a little too much. So i second this

TheBloke

Owner Jul 7, 2023

OK, noted. I've not yet had a chance to re-do my GPTQs but I'm expecting to get some new HW soon which will enable me to do that, so hopefully fairly soon.

astronauti

Jul 28, 2023

Hey, just wanted to check up on the 8bit variant :)

DQ83

Sep 7, 2023

@TheBloke would be great if they will also work with Exlama. recently the config.json are missing the pad_token configurations which leads the model not to load

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