Instructions to use facebook/opt-iml-30b with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use facebook/opt-iml-30b with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("text-generation", model="facebook/opt-iml-30b")# Load model directly from transformers import AutoTokenizer, AutoModelForCausalLM tokenizer = AutoTokenizer.from_pretrained("facebook/opt-iml-30b") model = AutoModelForCausalLM.from_pretrained("facebook/opt-iml-30b") - Notebooks
- Google Colab
- Kaggle
- Local Apps
- vLLM
How to use facebook/opt-iml-30b with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "facebook/opt-iml-30b" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "facebook/opt-iml-30b", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker
docker model run hf.co/facebook/opt-iml-30b
- SGLang
How to use facebook/opt-iml-30b 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 "facebook/opt-iml-30b" \ --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": "facebook/opt-iml-30b", "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 "facebook/opt-iml-30b" \ --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": "facebook/opt-iml-30b", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }' - Docker Model Runner
How to use facebook/opt-iml-30b with Docker Model Runner:
docker model run hf.co/facebook/opt-iml-30b
Grammar Error Correction prompt
Hi,
I see the paper evaluates the model on zero-shot grammar error correction - and it was quite strong, evaluating on JFLEG dataset. I have been trying zero-shot gec on passages using a few different prompts, example below, but had no luck so far. It just seems to rewrite the passage each time.
Would you happen to have the prompts used for gec in the paper ?
Thanks!
Darragh.
Rewrite the below essay with correct grammar.
Context : The hardest part of school is getting ready. you wake up go brush your teeth and go to your closet and look at your cloths. after you think you picked a outfit u go look in the mirror and youll either not like it or you look and see a stain. Then you'll have to change. with the online classes you can wear anything and stay home and you wont need to stress about what to wear.
Output:
@darraghd Did you have any progress on this? I would be interested in finding out more about this too, as I didn't have much success with GEC with OPT-IML.
I am also interested to see how we can make this work. I checked the paper, but there are no samples related to the JFLEG dataset.