Instructions to use stabilityai/stablelm-tuned-alpha-7b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use stabilityai/stablelm-tuned-alpha-7b with Transformers:

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

pipe = pipeline("text-generation", model="stabilityai/stablelm-tuned-alpha-7b")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("stabilityai/stablelm-tuned-alpha-7b")
model = AutoModelForCausalLM.from_pretrained("stabilityai/stablelm-tuned-alpha-7b")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use stabilityai/stablelm-tuned-alpha-7b with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/stabilityai/stablelm-tuned-alpha-7b

SGLang

How to use stabilityai/stablelm-tuned-alpha-7b 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 "stabilityai/stablelm-tuned-alpha-7b" \
    --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": "stabilityai/stablelm-tuned-alpha-7b",
		"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 "stabilityai/stablelm-tuned-alpha-7b" \
        --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": "stabilityai/stablelm-tuned-alpha-7b",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use stabilityai/stablelm-tuned-alpha-7b with Docker Model Runner:
```
docker model run hf.co/stabilityai/stablelm-tuned-alpha-7b
```

GGML f16, q4_0, q4_1, q4_2, q4_3

by oeathus - opened Apr 26, 2023

Discussion

oeathus

Apr 26, 2023

I converted and quantized these with https://github.com/ggerganov/ggml/ on a MacBook pro M1 w/ 16GB RAM.

https://huggingface.co/oeathus/stablelm-tuned-alpha-7b-ggml-q4

andy-og

Apr 28, 2023

Can you give me a heads up on how to plug these in and perform some local inference on my mac? Here is what I have so far:

def hugging_local(text="Can you please let us know more details about your "):
    from transformers import AutoTokenizer, AutoModelForCausalLM

    tokenizer = AutoTokenizer.from_pretrained("stabilityai/stablelm-tuned-alpha-7b")

    model = AutoModelForCausalLM.from_pretrained("stabilityai/stablelm-tuned-alpha-7b")

    from langchain.llms import HuggingFacePipeline
    llm = HuggingFacePipeline(model=model, tokenizer=tokenizer)

    template = """Question: {question}

    Answer: """
    prompt = PromptTemplate(template=template, input_variables=["question"])
    llm_chain = LLMChain(prompt=prompt, llm=llm)

    question = "Who won the FIFA World Cup in the year 1994? "

    print(llm_chain.run(question))

    return

if __name__ == '__main__':

    # result = hugging_lang()
    # result = hugging_raw(text=test_text)
    result = hugging_local(text=test_text)

    print(result)

oeathus

Apr 28, 2023

I'm still wrapping my head around the GGML format. My understanding is that it is a custom serialized binary format that sorta zips the parameters and other essentials on top of the actual neural net. I don't think you can run these with the Hugging Face transformers library, but I'm not terribly confident about that.

andy-og

Apr 28, 2023

Okay, yeah, I am struggling. I also was trying to use the hosted inference and it just times out constantly.

ldilov

May 14, 2023

ldilov/stablelm-tuned-alpha-7b-4bit-128g-descact-sym-true-sequential

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