Instructions to use upstage/llama-30b-instruct-2048 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use upstage/llama-30b-instruct-2048 with Transformers:

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

pipe = pipeline("text-generation", model="upstage/llama-30b-instruct-2048")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("upstage/llama-30b-instruct-2048")
model = AutoModelForCausalLM.from_pretrained("upstage/llama-30b-instruct-2048")

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use upstage/llama-30b-instruct-2048 with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "upstage/llama-30b-instruct-2048"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "upstage/llama-30b-instruct-2048",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/upstage/llama-30b-instruct-2048

SGLang

How to use upstage/llama-30b-instruct-2048 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/llama-30b-instruct-2048" \
    --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": "upstage/llama-30b-instruct-2048",
		"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 "upstage/llama-30b-instruct-2048" \
        --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": "upstage/llama-30b-instruct-2048",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use upstage/llama-30b-instruct-2048 with Docker Model Runner:
```
docker model run hf.co/upstage/llama-30b-instruct-2048
```

Need any help?

by hunkim - opened Jul 19, 2023

Discussion

hunkim

upstage org Jul 19, 2023

Please let us know. We are happy to help you.

wonhosong changed discussion title from Need any hlep? to Need any help? Jul 19, 2023

mechanicmuthu

Jul 19, 2023

Are you planning to make any 13B and 7B versions of this? Is there any link to the paper in which you describe how you have achieved state of the open art performance. I'm sure you must be starting the same process on llama-2 also. Also, the licensing restriction on this is flowing from llama license or you have imposed something on your own. Many things to ask in a single post. But you would understand my excitement. Thanks.

linkai-dl

Jul 20, 2023

if possible, could you release the code for fine tuning? and the prompt template?

hunkim

upstage org Jul 22, 2023

We will be working on 13B and 7B.

We are still collecting more data and changing the fine-tuning part. Once settled, we will consider opening more, including code for fine-tuning and the prompt template.

Thank you!

hunkim changed discussion status to closed Jul 22, 2023

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