Create save_ema.py

save_ema.py

@@ -0,0 +1,14 @@
+import torch
+input_path = "best_checkpoint.pth"
+output_path = "best_checkpoint_ema.pth" 
+
+state = torch.load(input_path, map_location="cpu", weights_only=False)
+
+ema_state = state["model_ema"]
+
+if hasattr(ema_state, "state_dict"):
+    ema_state = ema_state.state_dict()
+
+torch.save(ema_state, output_path)
+
+print(f"saved EMA weights to {output_path}")