feat: log learning_rate

Former-commit-id: aaf6be4efe43d65e70650ee8c07b81b584a8d70e [formerly c4289255d70c75c72b684886824832ab61df533b]
Former-commit-id: a163c42fa2ca66e32c093424ed8ffdc3b82b5ea5

src/train.py

@@ -17,8 +17,7 @@ class_labels = {
     1: "sphere",
 }
 
-
-def main():
+if __name__ == "__main__":
     # setup logging
     logging.basicConfig(level=logging.INFO, format="%(levelname)s: %(message)s")
 
@@ -41,6 +40,8 @@ def main():
             EPOCHS=5,
             BATCH_SIZE=16,
             LEARNING_RATE=1e-4,
+            WEIGHT_DECAY=1e-8,
+            MOMENTUM=0.9,
             IMG_SIZE=512,
             SPHERES=5,
         ),
@@ -88,7 +89,7 @@ def main():
     ds_train = SphereDataset(image_dir=wandb.config.DIR_TRAIN_IMG, transform=tf_train)
     ds_valid = SphereDataset(image_dir=wandb.config.DIR_VALID_IMG, transform=tf_valid)
 
-    # 2.5 Create subset, if uncommented
+    # 2.5. Create subset, if uncommented
     ds_train = torch.utils.data.Subset(ds_train, list(range(0, len(ds_train), len(ds_train) // 5000)))
     ds_valid = torch.utils.data.Subset(ds_valid, list(range(0, len(ds_valid), len(ds_valid) // 100)))
 
@@ -110,7 +111,12 @@ def main():
     )
 
     # 4. Set up the optimizer, the loss, the learning rate scheduler and the loss scaling for amp
-    optimizer = torch.optim.RMSprop(net.parameters(), lr=wandb.config.LEARNING_RATE, weight_decay=1e-8, momentum=0.9)
+    optimizer = torch.optim.RMSprop(
+        net.parameters(),
+        lr=wandb.config.LEARNING_RATE,
+        weight_decay=wandb.config.WEIGHT_DECAY,
+        momentum=wandb.config.MOMENTUM,
+    )
     scheduler = torch.optim.lr_scheduler.ReduceLROnPlateau(optimizer, "max", patience=2)
     grad_scaler = torch.cuda.amp.GradScaler(enabled=wandb.config.AMP)
     criterion = torch.nn.BCEWithLogitsLoss()
@@ -137,6 +143,14 @@ def main():
     logging.info(f"wandb config:\n{yaml.dump(wandb.config.as_dict())}")
 
     try:
+        # wandb init log
+        # wandb.log(
+        #     {
+        #         "train/learning_rate": scheduler.get_lr(),
+        #     },
+        #     commit=False,
+        # )
+
         for epoch in range(1, wandb.config.EPOCHS + 1):
             with tqdm(total=len(ds_train), desc=f"{epoch}/{wandb.config.EPOCHS}", unit="img") as pbar:
 
@@ -245,6 +259,7 @@ def main():
                 wandb.log(
                     {
                         "predictions": table,
+                        "train/learning_rate": optimizer.state_dict()["param_groups"][0]["lr"],
                         "val/accuracy": accuracy,
                         "val/bce": val_loss,
                         "val/dice": dice,
@@ -276,7 +291,3 @@ def main():
     except KeyboardInterrupt:
         torch.save(net.state_dict(), "INTERRUPTED.pth")
         raise
-
-
-if __name__ == "__main__":
-    main()  # TODO: fix toutes les metrics, loss, accuracy, dice...