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

Libraries

How to use Nexusflow/Starling-LM-7B-beta with Transformers:

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

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

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Nexusflow/Starling-LM-7B-beta")
model = AutoModelForCausalLM.from_pretrained("Nexusflow/Starling-LM-7B-beta")
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 Nexusflow/Starling-LM-7B-beta with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/Nexusflow/Starling-LM-7B-beta

SGLang

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

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

verbosity

by userBo - opened Mar 21, 2024

Discussion

userBo

Mar 21, 2024

This is a good model but the verbosity issue is still there like the alpha version. Why is that?

sparsh35

Mar 21, 2024

I think it is because it is trained on chat gpt data which is similarly verbose that's why it is verbose. It might be more good for more reasoning tasks as verbosity helps in getting to finally correct answer.

userBo

Mar 21, 2024

It's not just elaborate answers, but a simple "hello" would make it go on forever

banghua

Nexusflow org Mar 21, 2024

Thank you for your feedback. The model generation is still on the longer side but the verbosity is slightly improved compared with the alpha version. The unnecessary generation should be mitigated with a second round RLHF, which we plan to investigate in the near future. For now we release the beta version since it's better than alpha version in some of the reasoning tasks we tested.

banghua changed discussion status to closed Mar 21, 2024

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