Datasets:

lamooon
/

vocal

+=== onset_v9 (NO CACHE, live MERT) — Sun Apr  5 19:55:23 HKT 2026 ===
+SLURM_JOB_ID=128397 node=gpu-4080-409 gpu=0
+Cleaned up any stale onset_v9 cache
+Filesystem                               Size  Used Avail Use% Mounted on
+10.21.7.193:/dpool/userhome/cs/u3571824  400G  356G   45G  89% /userhome/cs/u3571824
+=== Phase 1: Train onset_v9 (learnable layer weights, MERT live) ===
+Device: cuda
+v9: Learnable MERT layer weights (all 13 layers, LIVE — no cache)
+Train: 318  Val: 39
+Loading MERT-v1-95M (frozen)...
+WARNING: feature_extractor_cqt requires the libray 'nnAudio'
+MERT loaded (94M params, frozen)
+  OnsetV9Dataset: 318 tracks (live MERT, no cache)
+  OnsetV9Dataset: 39 tracks (live MERT, no cache)
+/userhome/cs/u3571824/fyp/muse/backend/ml_models/dev/vocal/train_onset_v9.py:432: FutureWarning: `torch.cuda.amp.GradScaler(args...)` is deprecated. Please use `torch.amp.GradScaler('cuda', args...)` instead.
+  scaler = GradScaler(enabled=(args.fp16 and device == 'cuda'))
+OnsetMERTLearnedLayers: 0.174M params (+13 layer weight logits)
+=== Training onset_v9: 40 epochs (MERT live) ===
+/userhome/cs/u3571824/fyp/muse/backend/ml_models/dev/vocal/train_onset_v9.py:313: FutureWarning: `torch.cuda.amp.autocast(args...)` is deprecated. Please use `torch.amp.autocast('cuda', args...)` instead.
+  with cuda_autocast(enabled=scaler.is_enabled()):
+Traceback (most recent call last):
+  File "/userhome/cs/u3571824/fyp/muse/backend/ml_models/dev/vocal/train_onset_v9.py", line 494, in <module>
+    main()
+  File "/userhome/cs/u3571824/fyp/muse/backend/ml_models/dev/vocal/train_onset_v9.py", line 444, in main
+    train_loss = train_epoch(model, train_loader, optim, sched, scaler, device, focal)
+  File "/userhome/cs/u3571824/fyp/muse/backend/ml_models/dev/vocal/train_onset_v9.py", line 309, in train_epoch
+    for batch in loader:
+  File "/userhome/cs/u3571824/anaconda3/envs/bass/lib/python3.10/site-packages/torch/utils/data/dataloader.py", line 741, in __next__
+    data = self._next_data()
+  File "/userhome/cs/u3571824/anaconda3/envs/bass/lib/python3.10/site-packages/torch/utils/data/dataloader.py", line 801, in _next_data
+    data = self._dataset_fetcher.fetch(index)  # may raise StopIteration
+  File "/userhome/cs/u3571824/anaconda3/envs/bass/lib/python3.10/site-packages/torch/utils/data/_utils/fetch.py", line 54, in fetch
+    data = [self.dataset[idx] for idx in possibly_batched_index]
+  File "/userhome/cs/u3571824/anaconda3/envs/bass/lib/python3.10/site-packages/torch/utils/data/_utils/fetch.py", line 54, in <listcomp>
+    data = [self.dataset[idx] for idx in possibly_batched_index]
+  File "/userhome/cs/u3571824/fyp/muse/backend/ml_models/dev/vocal/train_onset_v9.py", line 210, in __getitem__
+    features = extract_all_layers(track['audio'], self.device)
+  File "/userhome/cs/u3571824/fyp/muse/backend/ml_models/dev/vocal/train_onset_v9.py", line 104, in extract_all_layers
+    out = mert(chunk_gpu, output_hidden_states=True)
+  File "/userhome/cs/u3571824/anaconda3/envs/bass/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1776, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/userhome/cs/u3571824/anaconda3/envs/bass/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1787, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/userhome/cs/u3571824/.cache/huggingface/modules/transformers_modules/m_hyphen_a_hyphen_p/MERT_hyphen_v1_hyphen_95M/12af15fef9d0ac838c3f475bfbbf26d2060dd4f5/modeling_MERT.py", line 112, in forward
+    extract_features = self.feature_extractor(input_values)
+  File "/userhome/cs/u3571824/anaconda3/envs/bass/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1776, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/userhome/cs/u3571824/anaconda3/envs/bass/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1787, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/userhome/cs/u3571824/anaconda3/envs/bass/lib/python3.10/site-packages/transformers/models/hubert/modeling_hubert.py", line 211, in forward
+    hidden_states = conv_layer(hidden_states)
+  File "/userhome/cs/u3571824/anaconda3/envs/bass/lib/python3.10/site-packages/transformers/modeling_layers.py", line 93, in __call__
+    return super().__call__(*args, **kwargs)
+  File "/userhome/cs/u3571824/anaconda3/envs/bass/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1776, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/userhome/cs/u3571824/anaconda3/envs/bass/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1787, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/userhome/cs/u3571824/anaconda3/envs/bass/lib/python3.10/site-packages/transformers/models/hubert/modeling_hubert.py", line 122, in forward
+    hidden_states = self.conv(hidden_states)
+  File "/userhome/cs/u3571824/anaconda3/envs/bass/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1776, in _wrapped_call_impl
+    return self._call_impl(*args, **kwargs)
+  File "/userhome/cs/u3571824/anaconda3/envs/bass/lib/python3.10/site-packages/torch/nn/modules/module.py", line 1787, in _call_impl
+    return forward_call(*args, **kwargs)
+  File "/userhome/cs/u3571824/anaconda3/envs/bass/lib/python3.10/site-packages/torch/nn/modules/conv.py", line 375, in forward
+    return self._conv_forward(input, self.weight, self.bias)
+  File "/userhome/cs/u3571824/anaconda3/envs/bass/lib/python3.10/site-packages/torch/nn/modules/conv.py", line 370, in _conv_forward
+    return F.conv1d(
+RuntimeError: Calculated padded input size per channel: (1). Kernel size: (2). Kernel size can't be greater than actual input size
+FATAL: v9 training failed