Upload logs/onset_v7/vocal_onset_v7_20260404_234632.log with huggingface_hub

logs/onset_v7/vocal_onset_v7_20260404_234632.log

@@ -0,0 +1,92 @@
+=== onset_v7 — Sat Apr  4 23:46:32 HKT 2026 ===
+SLURM_JOB_ID=128337 node=gpu-4080-410 gpu=0
+v6 MERT cache: 11321 files (reusing for v7)
+
+=== Phase 1: Train onset_v7 ===
+Device: cuda
+v7: MERT features (768) + temporal diff (768) = 1536-dim input
+proj_dim=128, hidden_dim=64
+Train: 11283  Val: 39
+
+--- Checking MERT cache at /userhome/cs/u3571824/fyp/muse/backend/ml_models/dev/vocal/cache/onset_v6 ---
+Loading MERT-v1-95M (frozen)...
+WARNING: feature_extractor_cqt requires the libray 'nnAudio'
+
+/userhome/cs/u3571824/fyp/muse/backend/ml_models/dev/vocal/train_onset_v7.py:311: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead.
+  scaler = GradScaler(enabled=(args.fp16 and device == 'cuda'))
+/userhome/cs/u3571824/fyp/muse/backend/ml_models/dev/vocal/train_onset_v7.py:197: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead.
+  with cuda_autocast(enabled=scaler.is_enabled()):
+MERT loaded (94M params, frozen)
+  WARN: /userhome/cs/u3571824/fyp/muse/backend/ml_models/dev/vocal/data/processed/audio/mir_st500_162.wav: Calculated padded input size per channel: (1). Kernel size: (2). Kernel size can't be greater than actual input size
+MERT feature extraction: 0 new, 11321 cached, 1 failed
+  WARN: missing cache for /userhome/cs/u3571824/fyp/muse/backend/ml_models/dev/vocal/data/processed/audio/mir_st500_162.wav, skipping
+  OnsetV7Dataset: 11282 tracks (v6 cache)
+  OnsetV7Dataset: 39 tracks (v6 cache)
+OnsetMERTDiffGRU: 0.346M params
+
+=== Training onset_v7: 40 epochs ===
+Ep 1/40 (1014s) lr=1.00e-03 | train_loss=0.0223 | val_f1=0.4469 P=0.5615 R=0.3830
+  * New best: 0.4469
+Ep 2/40 (363s) lr=1.00e-03 | train_loss=0.0210 | val_f1=0.5351 P=0.6160 R=0.4797
+  * New best: 0.5351
+Ep 3/40 (361s) lr=9.98e-04 | train_loss=0.0203 | val_f1=0.5875 P=0.6639 R=0.5382
+  * New best: 0.5875
+Ep 4/40 (353s) lr=9.93e-04 | train_loss=0.0198 | val_f1=0.6011 P=0.6587 R=0.5654
+  * New best: 0.6011
+Ep 5/40 (360s) lr=9.85e-04 | train_loss=0.0195 | val_f1=0.6306 P=0.6980 R=0.5887
+  * New best: 0.6306
+Ep 6/40 (356s) lr=9.73e-04 | train_loss=0.0191 | val_f1=0.5370 P=0.6715 R=0.4610
+Ep 7/40 (357s) lr=9.58e-04 | train_loss=0.0188 | val_f1=0.6031 P=0.6885 R=0.5488
+Ep 8/40 (354s) lr=9.40e-04 | train_loss=0.0183 | val_f1=0.6277 P=0.6635 R=0.6052
+Ep 9/40 (339s) lr=9.19e-04 | train_loss=0.0180 | val_f1=0.6130 P=0.7097 R=0.5514
+Ep 10/40 (340s) lr=8.95e-04 | train_loss=0.0176 | val_f1=0.6156 P=0.7053 R=0.5589
+Ep 11/40 (345s) lr=8.68e-04 | train_loss=0.0172 | val_f1=0.6008 P=0.6845 R=0.5471
+Ep 12/40 (341s) lr=8.39e-04 | train_loss=0.0168 | val_f1=0.6258 P=0.6753 R=0.5977
+  Early stopping at epoch 12
+
+Done. Best val_f1=0.6306  ->  /userhome/cs/u3571824/fyp/muse/backend/ml_models/dev/vocal/outputs/onset_v7/best.pt
+
+
+
+
  .../outputs/onset_v7/best.pt:  40%|███▉      |  551kB / 1.39MB            
+
+
  .../outputs/onset_v7/best.pt:  40%|███▉      |  551kB / 1.39MB            
+
+
+
  .../outputs/onset_v7/best.pt:  40%|███▉      |  551kB / 1.39MB            
+
+
  .../outputs/onset_v7/best.pt:  40%|███▉      |  551kB / 1.39MB            
+
+
  .../outputs/onset_v7/best.pt:  40%|███▉      |  551kB / 1.39MB            
+
+
  .../outputs/onset_v7/best.pt:  40%|███▉      |  551kB / 1.39MB            
+
+
  .../outputs/onset_v7/best.pt:  40%|███▉      |  551kB / 1.39MB            
+
+
  .../outputs/onset_v7/best.pt:  79%|███████▉  | 1.10MB / 1.39MB            
+
+
+
  .../outputs/onset_v7/best.pt:  79%|███████▉  | 1.10MB / 1.39MB            
+
+
  .../outputs/onset_v7/best.pt:  79%|███████▉  | 1.10MB / 1.39MB            
+
+
  .../outputs/onset_v7/best.pt:  79%|███████▉  | 1.10MB / 1.39MB            
+
+
  .../outputs/onset_v7/best.pt:  79%|███████▉  | 1.10MB / 1.39MB            
+
+
  .../outputs/onset_v7/best.pt:  79%|███████▉  | 1.10MB / 1.39MB            
+
+
  .../outputs/onset_v7/best.pt:  79%|███████▉  | 1.10MB / 1.39MB            
+
+
  .../outputs/onset_v7/best.pt:  79%|███████▉  | 1.10MB / 1.39MB            
+
+
  .../outputs/onset_v7/best.pt: 100%|██████████| 1.39MB / 1.39MB            
+
+
+
  .../outputs/onset_v7/best.pt: 100%|██████████| 1.39MB / 1.39MB            
+
+
  .../outputs/onset_v7/best.pt: 100%|██████████| 1.39MB / 1.39MB            
+
+
  .../outputs/onset_v7/best.pt: 100%|██████████| 1.39MB / 1.39MB            
+No files have been modified since last commit. Skipping to prevent empty commit.
+Phase 1 complete: /userhome/cs/u3571824/fyp/muse/backend/ml_models/dev/vocal/outputs/onset_v7/best.pt