Instructions to use upstage/Llama-2-70b-instruct with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use upstage/Llama-2-70b-instruct with Transformers:

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

pipe = pipeline("text-generation", model="upstage/Llama-2-70b-instruct")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("upstage/Llama-2-70b-instruct")
model = AutoModelForCausalLM.from_pretrained("upstage/Llama-2-70b-instruct")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use upstage/Llama-2-70b-instruct with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "upstage/Llama-2-70b-instruct"
# 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-2-70b-instruct",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/upstage/Llama-2-70b-instruct

SGLang

How to use upstage/Llama-2-70b-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/Llama-2-70b-instruct" \
    --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-2-70b-instruct",
		"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-2-70b-instruct" \
        --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-2-70b-instruct",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use upstage/Llama-2-70b-instruct with Docker Model Runner:
```
docker model run hf.co/upstage/Llama-2-70b-instruct
```

Does this finetune let you use 4096 context?

by Panchovix - opened Jul 24, 2023

Discussion

Panchovix

Jul 24, 2023

Hi there, very impressive results!

Was wondering, checking the file config.json, the

max_position_embeddings variable is set to 2048, while for llama-2 (https://huggingface.co/meta-llama/Llama-2-70b-hf/blob/main/config.json), this value is set to 4096.

Would this model be able to do 4096 context, as llama-2-70b?

Limerobot

upstage org Jul 25, 2023

Hi,

Yes, it would be possible to set the max_seq_len to 4096.

The reason our max_position_embeddings in the config is set to 2048 is because we based our work on a previous version of the Llama2 model, as you can see in this link (https://huggingface.co/meta-llama/Llama-2-70b-hf/blob/de00c41a63fb46d805f85f92fe5418a9633bb97d/config.json).

Panchovix changed discussion status to closed Jul 25, 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