Instructions to use Yale-BIDS-Chen/LEME-SFT-70B with libraries, inference providers, notebooks, and local apps. Follow these links to get started.

Libraries

How to use Yale-BIDS-Chen/LEME-SFT-70B with Transformers:

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

pipe = pipeline("text-generation", model="Yale-BIDS-Chen/LEME-SFT-70B")
messages = [
    {"role": "user", "content": "Who are you?"},
]
pipe(messages)

# pip install -U transformers accelerate
# Load model directly
from transformers import AutoTokenizer, AutoModelForCausalLM

tokenizer = AutoTokenizer.from_pretrained("Yale-BIDS-Chen/LEME-SFT-70B")
model = AutoModelForCausalLM.from_pretrained("Yale-BIDS-Chen/LEME-SFT-70B", device_map="auto")
messages = [
    {"role": "user", "content": "Who are you?"},
]
inputs = tokenizer.apply_chat_template(
	messages,
	add_generation_prompt=True,
	tokenize=True,
	return_dict=True,
	return_tensors="pt",
).to(model.device)

outputs = model.generate(**inputs, max_new_tokens=256)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:]))

Notebooks
Google Colab
Kaggle
Local Apps Settings

vLLM

How to use Yale-BIDS-Chen/LEME-SFT-70B with vLLM:

Install from pip and serve model

# Install vLLM from pip:
pip install vllm
# Start the vLLM server:
vllm serve "Yale-BIDS-Chen/LEME-SFT-70B"
# Call the server using curl (OpenAI-compatible API):
curl -X POST "http://localhost:8000/v1/chat/completions" \
	-H "Content-Type: application/json" \
	--data '{
		"model": "Yale-BIDS-Chen/LEME-SFT-70B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Use Docker

docker model run hf.co/Yale-BIDS-Chen/LEME-SFT-70B

SGLang

How to use Yale-BIDS-Chen/LEME-SFT-70B 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 "Yale-BIDS-Chen/LEME-SFT-70B" \
    --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": "Yale-BIDS-Chen/LEME-SFT-70B",
		"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 "Yale-BIDS-Chen/LEME-SFT-70B" \
        --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": "Yale-BIDS-Chen/LEME-SFT-70B",
		"messages": [
			{
				"role": "user",
				"content": "What is the capital of France?"
			}
		]
	}'

Docker Model Runner
How to use Yale-BIDS-Chen/LEME-SFT-70B with Docker Model Runner:
```
docker model run hf.co/Yale-BIDS-Chen/LEME-SFT-70B
```
Browse Quantizations to use this model in llama.cpp, Ollama, LM Studio, or any compatible app.

Llama 3 LEME-SFT-70B

LEME-SFT-70B is an ophthalmology language model developed by the Yale BIDS Chen Lab through supervised fine-tuning of Meta-Llama-3-70B-Instruct.

Paper · Code · Training data

Property	Value
Base model	Meta-Llama-3-70B-Instruct
Parameters	70B
Training	Supervised fine-tuning (SFT)
Weights	BF16, safetensors

Training data

The model was fine-tuned on ophthalmology instruction data covering open-ended and multiple-choice question answering, USMLE-style questions, clinical information extraction, and patient–clinician dialogue.

Training data: LEME-Training-Data (sft).

Usage

from transformers import AutoModelForCausalLM, AutoTokenizer

model_id = "Yale-BIDS-Chen/LEME-SFT-70B"
tokenizer = AutoTokenizer.from_pretrained(model_id)
model = AutoModelForCausalLM.from_pretrained(
    model_id,
    torch_dtype="auto",
    device_map="auto",
)

messages = [{"role": "user", "content": "Explain the difference between dry and wet AMD."}]
inputs = tokenizer.apply_chat_template(
    messages, add_generation_prompt=True, return_dict=True, return_tensors="pt"
).to(model.device)
outputs = model.generate(**inputs, max_new_tokens=512, do_sample=False)
print(tokenizer.decode(outputs[0][inputs["input_ids"].shape[-1]:], skip_special_tokens=True))

Contact

For general inquiries, please contact Hyunjae Kim at hyunjae.kim@yale.edu.

License

LEME code, models, data, and other related materials may be used for non-commercial research purposes. Any derivative works must be made publicly available through a public platform, such as GitHub or Hugging Face. For commercial-use inquiries, please contact us (agreements@yale.edu or qingyu.chen@yale.edu). Please see the LICENSE file for the full terms.

Built with Meta Llama 3. The base model is also subject to the Meta Llama 3 Community License and Acceptable Use Policy.

Citation

@article{kim2026leme,
  title={LEME: open large language models for ophthalmology with advanced reasoning and clinical validation},
  author={Kim, Hyunjae and Ai, Xuguang and Srinivasan, Sahana and Gilson, Aidan and Singer, Maxwell B and Pushpanathan, Krithi and Xie, Qianqian and Park, Jungwoo and Applebaum, Serina and Yang, Gabriel Dawei and others},
  journal={npj Digital Medicine},
  year={2026},
  publisher={Nature Publishing Group UK, London}
}