Instructions to use allenai/open-instruct-gpt4-alpaca-7b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use allenai/open-instruct-gpt4-alpaca-7b with Transformers:

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

pipe = pipeline("text-generation", model="allenai/open-instruct-gpt4-alpaca-7b")

# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("allenai/open-instruct-gpt4-alpaca-7b")
model = AutoModelForCausalLM.from_pretrained("allenai/open-instruct-gpt4-alpaca-7b")

Notebooks
Google Colab
Kaggle
Local Apps

vLLM

How to use allenai/open-instruct-gpt4-alpaca-7b with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "allenai/open-instruct-gpt4-alpaca-7b"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "allenai/open-instruct-gpt4-alpaca-7b",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Use Docker

docker model run hf.co/allenai/open-instruct-gpt4-alpaca-7b

SGLang

How to use allenai/open-instruct-gpt4-alpaca-7b 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 "allenai/open-instruct-gpt4-alpaca-7b" \
    --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": "allenai/open-instruct-gpt4-alpaca-7b",
		"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 "allenai/open-instruct-gpt4-alpaca-7b" \
        --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": "allenai/open-instruct-gpt4-alpaca-7b",
		"prompt": "Once upon a time,",
		"max_tokens": 512,
		"temperature": 0.5
	}'

Docker Model Runner
How to use allenai/open-instruct-gpt4-alpaca-7b with Docker Model Runner:
```
docker model run hf.co/allenai/open-instruct-gpt4-alpaca-7b
```

reproducing experiment, looking for some help or advise

by fblgit - opened Mar 18, 2025

Discussion

fblgit

Mar 18, 2025

hi guys, amazing work from an amazing lab.

I am working on a paper, looking to reproduce some causal modeling experiment and I saw the massive amount of score marks provided in the paper of this model, but i cant find the hyperparameters and some material needed to reproduce it exactly.

Can you give me a hand to reproduce this or another from ur collections that you know/had made before and its a good empirical evidence for contrasting an llm performance so we can reproduce & reference & cite ?

much appreciated, thanks!

hamishivi

Ai2 org Mar 19, 2025

Hi! We provide information on replicating the experiments here: https://github.com/allenai/open-instruct/blob/main/docs/tulu1_tulu2.md (note that you need to use an earlier commit of open-instruct for this, since this is older work). I would recommend looking also at our Tulu 3 models (https://huggingface.co/collections/allenai/tulu-3-models-673b8e0dc3512e30e7dc54f5, paper: https://arxiv.org/abs/2411.15124), which are our latest and best sets of models and experiments, with fully open recipes and data for replication.

hamishivi changed discussion status to closed Mar 19, 2025

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