Image-Text-to-Text
Transformers
English
ophthalmology
retina
fundus
medical
vision-language
llava
llava-med
Instructions to use ASU-GSL/RetinalGPT with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- Transformers
How to use ASU-GSL/RetinalGPT with Transformers:
# Use a pipeline as a high-level helper from transformers import pipeline pipe = pipeline("image-text-to-text", model="ASU-GSL/RetinalGPT")# pip install -U transformers accelerate # Load model directly from transformers import AutoModel model = AutoModel.from_pretrained("ASU-GSL/RetinalGPT", device_map="auto") - Notebooks
- Google Colab
- Kaggle
- Local Apps Settings
- vLLM
How to use ASU-GSL/RetinalGPT with vLLM:
Install from pip and serve model
# Install vLLM from pip: pip install vllm # Start the vLLM server: vllm serve "ASU-GSL/RetinalGPT" # Call the server using curl (OpenAI-compatible API): curl -X POST "http://localhost:8000/v1/completions" \ -H "Content-Type: application/json" \ --data '{ "model": "ASU-GSL/RetinalGPT", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }'Use Docker
docker model run hf.co/ASU-GSL/RetinalGPT
- SGLang
How to use ASU-GSL/RetinalGPT 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 "ASU-GSL/RetinalGPT" \ --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": "ASU-GSL/RetinalGPT", "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 "ASU-GSL/RetinalGPT" \ --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": "ASU-GSL/RetinalGPT", "prompt": "Once upon a time,", "max_tokens": 512, "temperature": 0.5 }' - Docker Model Runner
How to use ASU-GSL/RetinalGPT with Docker Model Runner:
docker model run hf.co/ASU-GSL/RetinalGPT
Simplify model card
Browse files
README.md
CHANGED
|
@@ -1,6 +1,4 @@
|
|
| 1 |
---
|
| 2 |
-
license: other
|
| 3 |
-
license_name: llama-and-research-only
|
| 4 |
language:
|
| 5 |
- en
|
| 6 |
pipeline_tag: image-text-to-text
|
|
@@ -178,21 +176,7 @@ Instruction data is generated with GPT-4 from structured clinical features: dise
|
|
| 178 |
<img src="figures/figure4.png" alt="Lesion localization and vascular analysis" width="720">
|
| 179 |
</p>
|
| 180 |
|
| 181 |
-
|
| 182 |
-
|
| 183 |
-
**Intended use.** Research on vision–language modeling for retinal imaging, and reproduction of the results in the paper.
|
| 184 |
-
|
| 185 |
-
**Out of scope.** RetinalGPT is **not a medical device**. It must not be used for clinical diagnosis, treatment, screening, or any patient-facing or commercial deployment.
|
| 186 |
-
|
| 187 |
-
**Limitations.**
|
| 188 |
-
- The model can hallucinate findings, diseases, or severity; outputs must be verified by qualified professionals.
|
| 189 |
-
- Vascular metrics are estimates learned from AutoMorph supervision, not measurements. Fractal dimension and vessel density agree well with AutoMorph; vessel width is considerably less reliable.
|
| 190 |
-
- Bounding boxes are approximate and were only evaluated as a preliminary feasibility study.
|
| 191 |
-
- Trained on color fundus photographs in English only; behavior on other modalities (e.g., OCT), devices, and populations is not established.
|
| 192 |
-
|
| 193 |
-
## 📄 License
|
| 194 |
-
|
| 195 |
-
The code is released under the MIT License. The model weights are derived from LLaMA and LLaVA-Med and are subject to their original licenses and terms of use; they are released for **research purposes only**.
|
| 196 |
|
| 197 |
## 📖 Citation
|
| 198 |
|
|
|
|
| 1 |
---
|
|
|
|
|
|
|
| 2 |
language:
|
| 3 |
- en
|
| 4 |
pipeline_tag: image-text-to-text
|
|
|
|
| 176 |
<img src="figures/figure4.png" alt="Lesion localization and vascular analysis" width="720">
|
| 177 |
</p>
|
| 178 |
|
| 179 |
+
> RetinalGPT is released for research purposes.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 180 |
|
| 181 |
## 📖 Citation
|
| 182 |
|