Upload logs/melroformer/melroformer_20260406_020622.log with huggingface_hub

logs/melroformer/melroformer_20260406_020622.log

@@ -1197,3 +1197,27 @@ Uploaded: logs/melroformer/training_log.jsonl
 
 Done. Best val COn=0.2499 → /userhome/cs/u3571824/fyp/muse/backend/ml_models/dev/vocal/outputs/melroformer/best.pt
 Phase 1 complete: /userhome/cs/u3571824/fyp/muse/backend/ml_models/dev/vocal/outputs/melroformer/best.pt
+Uploaded: logs/melroformer/melroformer_20260406_020622.log
+
+=== Phase 2: COn eval — Mel-RoFormer (MIR-ST500) ===
+Loaded melroformer (epoch 32, val_COn=0.2499)
+Loading MelBandRoformer backbone...
+  Checkpoint has 6 transformer layers (indices 0-5)
+  Checkpoint keys: 684, Model keys: 684
+  Weights loaded successfully (strict=True)
+Backbone loaded: 228.2M params (frozen)
+2026-04-06 03:37:03,223 INFO HTTP Request: GET https://huggingface.co/api/datasets/lamooon/vocal/tree/main?recursive=true&expand=false "HTTP/1.1 200 OK"
+2026-04-06 03:37:03,270 INFO MIR-ST500 eval ready: 63 tracks in data_eval/processed/audio
+2026-04-06 03:37:03,275 INFO Using directory scan: 63 tracks (manifest had 39)
+[1/63] mir_st500_12
+Traceback (most recent call last):
+  File "<string>", line 54, in <module>
+  File "/userhome/cs/u3571824/fyp/muse/backend/ml_models/dev/vocal/./train_melroformer.py", line 167, in extract_backbone_features
+    raise RuntimeError("Failed to capture backbone features")
+RuntimeError: Failed to capture backbone features
+
+=== Disk after ===
+Filesystem                               Size  Used Avail Use% Mounted on
+10.21.7.193:/dpool/userhome/cs/u3571824  400G  357G   44G  90% /userhome/cs/u3571824
+
+=== ALL DONE — Mon Apr  6 03:37:06 HKT 2026 ===