Instructions to use berkeley-nest/Starling-LM-7B-alpha with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use berkeley-nest/Starling-LM-7B-alpha with Transformers:

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

pipe = pipeline("text-generation", model="berkeley-nest/Starling-LM-7B-alpha")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("berkeley-nest/Starling-LM-7B-alpha")
model = AutoModelForCausalLM.from_pretrained("berkeley-nest/Starling-LM-7B-alpha")
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]:]))

Inference
Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use berkeley-nest/Starling-LM-7B-alpha with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/berkeley-nest/Starling-LM-7B-alpha

SGLang

How to use berkeley-nest/Starling-LM-7B-alpha 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 "berkeley-nest/Starling-LM-7B-alpha" \
    --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": "berkeley-nest/Starling-LM-7B-alpha",
		"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 "berkeley-nest/Starling-LM-7B-alpha" \
        --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": "berkeley-nest/Starling-LM-7B-alpha",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use berkeley-nest/Starling-LM-7B-alpha with Docker Model Runner:
```
docker model run hf.co/berkeley-nest/Starling-LM-7B-alpha
```

What could be instruction fine tuning prompt for this model?

#22

by andreaKIM - opened Dec 9, 2023

Discussion

andreaKIM

Dec 9, 2023

Hello, It is so impressive to hit 70B performance with only 10 times smaller model.
I want to try instruction fine tuning However I got no idea about prompt.
Where could instruction go in the prompt? (system message + instruction)
Thank you!

banghua

Berkeley-Nest org Dec 9, 2023

Thank you! The prompt for fine-tuning will be the same as the one used for chat, which is provided in model card and also below:

During both SFT and RLHF, the model does not follow system message. So I would suggest put all system message together with instruction inside "GPT4 Correct User:".

andreaKIM

Dec 9, 2023

Really quick answer for me :)
I would try with pleasure.
By the way, why the model could not follow system message? is there any reason?
Thank you again!

banghua

Berkeley-Nest org Dec 9, 2023

It's because we mostly trained on the prompt style that does not come with system message (usually the fine-tuning dataset won't come with prompts with standard system message). But if you just put system message inside "GPT4 Correct User:" and prompt it appropriately, it shall be able to follow it well.

andreaKIM

Dec 9, 2023

Thanks again! I have one more question.
Does "generation_config.json" file affect the training?
Because some other models in the hub contain other parameters such as temperature,do_sample,...
Does it affect on training? If it does, then can i adjust that parameters also?

banghua

Berkeley-Nest org Dec 22, 2023

Sorry for the late reply. I think it won't affect training. If you want to use RLHF then the generation parameters might affect the samples you collected. But for SFT it won't affect anything.

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