How openmedlab/retfound_mae works
RETFound is a retinal-image foundation model, with this repository chiefly providing supervised fine-tuning and evaluation of pretrained weights. The workflow is: an engineer configures a task and dataset, the entry point builds image loaders and a ViT classifier, the training engine updates it, and evaluation reports disease-classification metrics. The repository also includes a masked-autoencoder architecture representing the self-supervised pretraining approach. The README documents pretraining and adaptation capabilities, but the supplied tree has no pretraining runner; that branch is therefore limited to the architecture. Edges reflect excerpts, indexed references, or README statements.
Main entry points: main_finetune.py Fine-tune entry · engine_finetune.py Train and evaluate · models_mae.py Masked autoencoder · pos_embed.py Position embeddings
Components 12
Fine-tuning workflow
- Fine-tune entry
main_finetune.py - Image datasets
util/datasets.py - Train and evaluate
engine_finetune.py
Model architectures
- ViT classifier
models_vit.py - Masked autoencoder
models_mae.py - Position embeddings
util/pos_embed.py
Training support
- Layer-wise decay
util/lr_decay.py - Learning-rate schedule
util/lr_sched.py - Distributed utilities
util/misc.py
Weights and results
- Pretrained weights
- Metrics and checkpoints
Other components
- Engineer
Connections 14
- Engineer → Fine-tune entry: configures task
README.md - Fine-tune entry → Image datasets: builds loaders
main_finetune.py - Fine-tune entry → ViT classifier: constructs model
main_finetune.py - Fine-tune entry → Position embeddings: interpolates embeddings
main_finetune.py - Fine-tune entry → Layer-wise decay: groups parameters
main_finetune.py - Fine-tune entry → Distributed utilities: sets up training
main_finetune.py - Fine-tune entry → Train and evaluate: runs epochs
main_finetune.py - Fine-tune entry → Pretrained weights: loads fine-tune weights
README.md - Train and evaluate → Learning-rate schedule: adjusts learning rate
engine_finetune.py - Train and evaluate → Distributed utilities: logs and synchronizes
engine_finetune.py - Train and evaluate → Metrics and checkpoints: writes evaluation metrics
engine_finetune.py - Fine-tune entry → Metrics and checkpoints: saves checkpoints
main_finetune.py - Masked autoencoder → Position embeddings: initializes embeddings
models_mae.py - Engineer → Masked autoencoder: uses pretrained approach
README.md