Instructions to use upstage/solar-pro-preview-instruct with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use upstage/solar-pro-preview-instruct with Transformers:

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

pipe = pipeline("text-generation", model="upstage/solar-pro-preview-instruct", trust_remote_code=True)
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# Load model directly
from transformers import AutoModelForCausalLM
model = AutoModelForCausalLM.from_pretrained("upstage/solar-pro-preview-instruct", trust_remote_code=True, dtype="auto")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use upstage/solar-pro-preview-instruct with vLLM:

Install from pip and serve model

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

Use Docker

docker model run hf.co/upstage/solar-pro-preview-instruct

SGLang

How to use upstage/solar-pro-preview-instruct 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 "upstage/solar-pro-preview-instruct" \
    --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": "upstage/solar-pro-preview-instruct",
		"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 "upstage/solar-pro-preview-instruct" \
        --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": "upstage/solar-pro-preview-instruct",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use upstage/solar-pro-preview-instruct with Docker Model Runner:
```
docker model run hf.co/upstage/solar-pro-preview-instruct
```

support vllm

#10

by CarrotAI - opened Sep 14, 2024

Discussion

CarrotAI

Sep 14, 2024

•

edited Sep 14, 2024

hello upstage
Thank you for sharing the model.

To make it easier for me to use it, I asked vllm for PR by referring to the given code. Can you explain the difference with Llama in more detail? And will the structure of the Pro model be changed in the future?

https://github.com/vllm-project/vllm/pull/8386

JustJaro

Sep 14, 2024

Yes please!

sungrae-park

upstage org Sep 16, 2024

@CarrotAI
Thank you for your interest!
The architectural difference from Llama is the presence of the BSKCN (Block level SKip CoNnection). The rope scaling mentioned in the PR is not different from Llama. We will officially review your PR and offer assistance in case any challenges or difficulties arise.

CarrotAI

Sep 18, 2024

Thank you. It's been merged.

CarrotAI changed discussion status to closed Sep 19, 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