|
|
2025-07-10 14:08:47,126 - INFO - Training with parameters: |
|
|
2025-07-10 14:08:47,126 - INFO - Text model: sentence-transformers/paraphrase-multilingual-mpnet-base-v2 |
|
|
2025-07-10 14:08:47,126 - INFO - Audio model: facebook/w2v-bert-2.0 |
|
|
2025-07-10 14:08:47,126 - INFO - Freeze encoders: partial |
|
|
2025-07-10 14:08:47,126 - INFO - Text layers to unfreeze: 3 |
|
|
2025-07-10 14:08:47,126 - INFO - Audio layers to unfreeze: 3 |
|
|
2025-07-10 14:08:47,126 - INFO - Use cross-modal attention: False |
|
|
2025-07-10 14:08:47,127 - INFO - Use attentive pooling: False |
|
|
2025-07-10 14:08:47,127 - INFO - Use word-level alignment: True |
|
|
2025-07-10 14:08:47,127 - INFO - Batch size: 48 |
|
|
2025-07-10 14:08:47,127 - INFO - Gradient accumulation steps: 15 |
|
|
2025-07-10 14:08:47,127 - INFO - Effective batch size: 720 |
|
|
2025-07-10 14:08:47,127 - INFO - Mixed precision training: False |
|
|
2025-07-10 14:08:47,127 - INFO - Learning rate: 0.0008 |
|
|
2025-07-10 14:08:47,127 - INFO - Temperature: 0.1 |
|
|
2025-07-10 14:08:47,127 - INFO - Projection dimension: 768 |
|
|
2025-07-10 14:08:47,127 - INFO - Training samples: 21968 |
|
|
2025-07-10 14:08:47,127 - INFO - Validation samples: 9464 |
|
|
2025-07-10 14:08:47,127 - INFO - Test samples: 9467 |
|
|
2025-07-10 14:08:47,127 - INFO - Max audio length: 480000 samples (30.00 seconds at 16kHz) |
|
|
2025-07-10 14:08:47,127 - INFO - Loading tokenizer and feature extractor... |
|
|
2025-07-10 14:08:49,554 - INFO - Feature extractor output keys: ['input_features', 'attention_mask'] |
|
|
2025-07-10 14:08:49,554 - INFO - Creating datasets... |
|
|
2025-07-10 14:08:49,554 - INFO - Feature extractor output keys: ['input_features', 'attention_mask'] |
|
|
2025-07-10 14:08:49,555 - INFO - Feature extractor output keys: ['input_features', 'attention_mask'] |
|
|
2025-07-10 14:08:49,555 - INFO - Feature extractor output keys: ['input_features', 'attention_mask'] |
|
|
2025-07-10 14:08:49,555 - INFO - Creating data loaders... |
|
|
2025-07-10 14:08:49,555 - INFO - Checking a sample batch... |
|
|
2025-07-10 14:09:17,287 - INFO - input_ids_pos: torch.Size([48, 128]) |
|
|
2025-07-10 14:09:17,287 - INFO - attention_mask_pos: torch.Size([48, 128]) |
|
|
2025-07-10 14:09:17,287 - INFO - input_ids_neg: torch.Size([48, 128]) |
|
|
2025-07-10 14:09:17,287 - INFO - attention_mask_neg: torch.Size([48, 128]) |
|
|
2025-07-10 14:09:17,288 - INFO - input_values: torch.Size([48, 473, 160]) |
|
|
2025-07-10 14:09:17,288 - INFO - attention_mask_audio: torch.Size([48, 473]) |
|
|
2025-07-10 14:09:17,288 - INFO - is_corrupted: torch.Size([48]) |
|
|
2025-07-10 14:09:17,288 - INFO - correctness_scores: torch.Size([48]) |
|
|
2025-07-10 14:09:17,288 - INFO - Initializing model... |
|
|
2025-07-10 14:10:04,370 - INFO - Text encoder hidden dim: 768 |
|
|
2025-07-10 14:10:04,370 - INFO - Audio encoder hidden dim: 1024 |
|
|
2025-07-10 14:10:04,370 - INFO - Partial freezing: unfreezing last 3 text layers and 3 audio layers |
|
|
2025-07-10 14:10:04,371 - INFO - Unfreezing text encoder layer 9 |
|
|
2025-07-10 14:10:04,371 - INFO - Unfreezing text encoder layer 10 |
|
|
2025-07-10 14:10:04,371 - INFO - Unfreezing text encoder layer 11 |
|
|
2025-07-10 14:10:04,372 - INFO - Unfreezing audio encoder layer 21 |
|
|
2025-07-10 14:10:04,372 - INFO - Unfreezing audio encoder layer 22 |
|
|
2025-07-10 14:10:04,372 - INFO - Unfreezing audio encoder layer 23 |
|
|
2025-07-10 14:10:04,482 - INFO - Model initialized with 308,221,186 trainable parameters out of 879,798,082 total |
|
|
2025-07-10 14:10:05,165 - INFO - Using discriminative learning rates: encoder_lr=4e-05, main_lr=0.0008 |
|
|
2025-07-10 14:10:05,165 - INFO - Encoder parameters: 156, Non-encoder parameters: 37 |
|
|
2025-07-10 14:10:05,165 - INFO - Scheduler setup: |
|
|
2025-07-10 14:10:05,165 - INFO - Batches per epoch: 457 |
|
|
2025-07-10 14:10:05,165 - INFO - Accumulation steps: 15 |
|
|
2025-07-10 14:10:05,165 - INFO - Optimizer steps per epoch: 31 |
|
|
2025-07-10 14:10:05,165 - INFO - Total optimizer steps: 930 |
|
|
2025-07-10 14:10:05,165 - INFO - Warmup steps: 1000 |
|
|
2025-07-10 14:10:05,165 - INFO - Validating gradient accumulation setup... |
|
|
2025-07-10 14:10:05,165 - INFO - Validating gradient accumulation with 15 steps... |
|
|
2025-07-10 14:10:25,924 - WARNING - Not enough test batches (10) for accumulation_steps (15) |
|
|
2025-07-10 14:10:25,924 - INFO - Starting training for 30 epochs |
|
|
2025-07-10 14:11:12,619 - ERROR - Error in epoch 1: 'correctness' |
|
|
2025-07-10 14:14:10,031 - INFO - Training with parameters: |
|
|
2025-07-10 14:14:10,031 - INFO - Text model: sentence-transformers/paraphrase-multilingual-mpnet-base-v2 |
|
|
2025-07-10 14:14:10,031 - INFO - Audio model: facebook/w2v-bert-2.0 |
|
|
2025-07-10 14:14:10,031 - INFO - Freeze encoders: partial |
|
|
2025-07-10 14:14:10,031 - INFO - Text layers to unfreeze: 3 |
|
|
2025-07-10 14:14:10,032 - INFO - Audio layers to unfreeze: 3 |
|
|
2025-07-10 14:14:10,032 - INFO - Use cross-modal attention: False |
|
|
2025-07-10 14:14:10,032 - INFO - Use attentive pooling: False |
|
|
2025-07-10 14:14:10,032 - INFO - Use word-level alignment: True |
|
|
2025-07-10 14:14:10,032 - INFO - Batch size: 48 |
|
|
2025-07-10 14:14:10,032 - INFO - Gradient accumulation steps: 15 |
|
|
2025-07-10 14:14:10,032 - INFO - Effective batch size: 720 |
|
|
2025-07-10 14:14:10,032 - INFO - Mixed precision training: False |
|
|
2025-07-10 14:14:10,032 - INFO - Learning rate: 0.0008 |
|
|
2025-07-10 14:14:10,032 - INFO - Temperature: 0.1 |
|
|
2025-07-10 14:14:10,032 - INFO - Projection dimension: 768 |
|
|
2025-07-10 14:14:10,032 - INFO - Training samples: 21968 |
|
|
2025-07-10 14:14:10,032 - INFO - Validation samples: 9464 |
|
|
2025-07-10 14:14:10,032 - INFO - Test samples: 9467 |
|
|
2025-07-10 14:14:10,032 - INFO - Max audio length: 480000 samples (30.00 seconds at 16kHz) |
|
|
2025-07-10 14:14:10,032 - INFO - Loading tokenizer and feature extractor... |
|
|
2025-07-10 14:14:11,069 - INFO - Feature extractor output keys: ['input_features', 'attention_mask'] |
|
|
2025-07-10 14:14:11,069 - INFO - Creating datasets... |
|
|
2025-07-10 14:14:11,070 - INFO - Feature extractor output keys: ['input_features', 'attention_mask'] |
|
|
2025-07-10 14:14:11,070 - INFO - Feature extractor output keys: ['input_features', 'attention_mask'] |
|
|
2025-07-10 14:14:11,070 - INFO - Feature extractor output keys: ['input_features', 'attention_mask'] |
|
|
2025-07-10 14:14:11,070 - INFO - Creating data loaders... |
|
|
2025-07-10 14:14:11,070 - INFO - Checking a sample batch... |
|
|
2025-07-10 14:14:30,482 - INFO - input_ids_pos: torch.Size([48, 128]) |
|
|
2025-07-10 14:14:30,483 - INFO - attention_mask_pos: torch.Size([48, 128]) |
|
|
2025-07-10 14:14:30,483 - INFO - input_ids_neg: torch.Size([48, 128]) |
|
|
2025-07-10 14:14:30,483 - INFO - attention_mask_neg: torch.Size([48, 128]) |
|
|
2025-07-10 14:14:30,483 - INFO - input_values: torch.Size([48, 473, 160]) |
|
|
2025-07-10 14:14:30,483 - INFO - attention_mask_audio: torch.Size([48, 473]) |
|
|
2025-07-10 14:14:30,483 - INFO - is_corrupted: torch.Size([48]) |
|
|
2025-07-10 14:14:30,483 - INFO - correctness_scores: torch.Size([48]) |
|
|
2025-07-10 14:14:30,483 - INFO - Initializing model... |
|
|
2025-07-10 14:14:31,362 - INFO - Text encoder hidden dim: 768 |
|
|
2025-07-10 14:14:31,362 - INFO - Audio encoder hidden dim: 1024 |
|
|
2025-07-10 14:14:31,362 - INFO - Partial freezing: unfreezing last 3 text layers and 3 audio layers |
|
|
2025-07-10 14:14:31,363 - INFO - Unfreezing text encoder layer 9 |
|
|
2025-07-10 14:14:31,363 - INFO - Unfreezing text encoder layer 10 |
|
|
2025-07-10 14:14:31,363 - INFO - Unfreezing text encoder layer 11 |
|
|
2025-07-10 14:14:31,364 - INFO - Unfreezing audio encoder layer 21 |
|
|
2025-07-10 14:14:31,364 - INFO - Unfreezing audio encoder layer 22 |
|
|
2025-07-10 14:14:31,364 - INFO - Unfreezing audio encoder layer 23 |
|
|
2025-07-10 14:14:31,489 - INFO - Model initialized with 308,221,186 trainable parameters out of 879,798,082 total |
|
|
2025-07-10 14:14:32,299 - INFO - Using discriminative learning rates: encoder_lr=4e-05, main_lr=0.0008 |
|
|
2025-07-10 14:14:32,299 - INFO - Encoder parameters: 156, Non-encoder parameters: 37 |
|
|
2025-07-10 14:14:32,299 - INFO - Scheduler setup: |
|
|
2025-07-10 14:14:32,299 - INFO - Batches per epoch: 457 |
|
|
2025-07-10 14:14:32,299 - INFO - Accumulation steps: 15 |
|
|
2025-07-10 14:14:32,299 - INFO - Optimizer steps per epoch: 31 |
|
|
2025-07-10 14:14:32,299 - INFO - Total optimizer steps: 930 |
|
|
2025-07-10 14:14:32,299 - INFO - Warmup steps: 1000 |
|
|
2025-07-10 14:14:32,299 - INFO - Validating gradient accumulation setup... |
|
|
2025-07-10 14:14:32,299 - INFO - Validating gradient accumulation with 15 steps... |
|
|
2025-07-10 14:14:52,293 - WARNING - Not enough test batches (10) for accumulation_steps (15) |
|
|
2025-07-10 14:14:52,294 - INFO - Starting training for 30 epochs |
|
|
2025-07-10 14:27:48,777 - INFO - Epoch 1: Total optimizer steps: 31 |
|
|
2025-07-10 14:31:08,722 - INFO - Validation metrics: |
|
|
2025-07-10 14:31:08,723 - INFO - Loss: 1.0246 |
|
|
2025-07-10 14:31:08,723 - INFO - Average similarity: 0.1445 |
|
|
2025-07-10 14:31:08,723 - INFO - Median similarity: 0.0786 |
|
|
2025-07-10 14:31:08,723 - INFO - Clean sample similarity: 0.1445 |
|
|
2025-07-10 14:31:08,723 - INFO - Corrupted sample similarity: 0.0879 |
|
|
2025-07-10 14:31:08,723 - INFO - Similarity gap (clean - corrupt): 0.0566 |
|
|
2025-07-10 14:31:08,840 - INFO - Epoch 1/30 - Train Loss: 1.2519, Val Loss: 1.0246, Clean Sim: 0.1445, Corrupt Sim: 0.0879, Gap: 0.0566, Time: 976.55s |
|
|
2025-07-10 14:31:08,840 - INFO - New best validation loss: 1.0246 |
|
|
2025-07-10 14:31:14,641 - INFO - New best similarity gap: 0.0566 |
|
|
2025-07-10 14:31:29,121 - INFO - Training with parameters: |
|
|
2025-07-10 14:31:29,121 - INFO - Text model: sentence-transformers/paraphrase-multilingual-mpnet-base-v2 |
|
|
2025-07-10 14:31:29,121 - INFO - Audio model: facebook/w2v-bert-2.0 |
|
|
2025-07-10 14:31:29,121 - INFO - Freeze encoders: partial |
|
|
2025-07-10 14:31:29,121 - INFO - Text layers to unfreeze: 3 |
|
|
2025-07-10 14:31:29,121 - INFO - Audio layers to unfreeze: 3 |
|
|
2025-07-10 14:31:29,121 - INFO - Use cross-modal attention: False |
|
|
2025-07-10 14:31:29,121 - INFO - Use attentive pooling: False |
|
|
2025-07-10 14:31:29,121 - INFO - Use word-level alignment: True |
|
|
2025-07-10 14:31:29,121 - INFO - Batch size: 48 |
|
|
2025-07-10 14:31:29,121 - INFO - Gradient accumulation steps: 15 |
|
|
2025-07-10 14:31:29,121 - INFO - Effective batch size: 720 |
|
|
2025-07-10 14:31:29,121 - INFO - Mixed precision training: False |
|
|
2025-07-10 14:31:29,121 - INFO - Learning rate: 0.0008 |
|
|
2025-07-10 14:31:29,121 - INFO - Temperature: 0.1 |
|
|
2025-07-10 14:31:29,121 - INFO - Projection dimension: 768 |
|
|
2025-07-10 14:31:29,121 - INFO - Training samples: 21968 |
|
|
2025-07-10 14:31:29,121 - INFO - Validation samples: 9464 |
|
|
2025-07-10 14:31:29,121 - INFO - Test samples: 9467 |
|
|
2025-07-10 14:31:29,122 - INFO - Max audio length: 480000 samples (30.00 seconds at 16kHz) |
|
|
2025-07-10 14:31:29,122 - INFO - Loading tokenizer and feature extractor... |
|
|
2025-07-10 14:31:30,216 - INFO - Feature extractor output keys: ['input_features', 'attention_mask'] |
|
|
2025-07-10 14:31:30,216 - INFO - Creating datasets... |
|
|
2025-07-10 14:31:30,216 - INFO - Feature extractor output keys: ['input_features', 'attention_mask'] |
|
|
2025-07-10 14:31:30,216 - INFO - Feature extractor output keys: ['input_features', 'attention_mask'] |
|
|
2025-07-10 14:31:30,217 - INFO - Feature extractor output keys: ['input_features', 'attention_mask'] |
|
|
2025-07-10 14:31:30,217 - INFO - Creating data loaders... |
|
|
2025-07-10 14:31:30,217 - INFO - Checking a sample batch... |
|
|
2025-07-10 14:32:58,374 - INFO - Training with parameters: |
|
|
2025-07-10 14:32:58,374 - INFO - Text model: sentence-transformers/paraphrase-multilingual-mpnet-base-v2 |
|
|
2025-07-10 14:32:58,374 - INFO - Audio model: facebook/w2v-bert-2.0 |
|
|
2025-07-10 14:32:58,374 - INFO - Freeze encoders: partial |
|
|
2025-07-10 14:32:58,374 - INFO - Text layers to unfreeze: 3 |
|
|
2025-07-10 14:32:58,374 - INFO - Audio layers to unfreeze: 3 |
|
|
2025-07-10 14:32:58,374 - INFO - Use cross-modal attention: False |
|
|
2025-07-10 14:32:58,374 - INFO - Use attentive pooling: False |
|
|
2025-07-10 14:32:58,374 - INFO - Use word-level alignment: True |
|
|
2025-07-10 14:32:58,374 - INFO - Batch size: 48 |
|
|
2025-07-10 14:32:58,374 - INFO - Gradient accumulation steps: 15 |
|
|
2025-07-10 14:32:58,374 - INFO - Effective batch size: 720 |
|
|
2025-07-10 14:32:58,374 - INFO - Mixed precision training: False |
|
|
2025-07-10 14:32:58,374 - INFO - Learning rate: 0.0008 |
|
|
2025-07-10 14:32:58,374 - INFO - Temperature: 0.1 |
|
|
2025-07-10 14:32:58,374 - INFO - Projection dimension: 768 |
|
|
2025-07-10 14:32:58,374 - INFO - Training samples: 21968 |
|
|
2025-07-10 14:32:58,374 - INFO - Validation samples: 9464 |
|
|
2025-07-10 14:32:58,374 - INFO - Test samples: 9467 |
|
|
2025-07-10 14:32:58,374 - INFO - Max audio length: 480000 samples (30.00 seconds at 16kHz) |
|
|
2025-07-10 14:32:58,374 - INFO - Loading tokenizer and feature extractor... |
|
|
2025-07-10 14:32:59,342 - INFO - Feature extractor output keys: ['input_features', 'attention_mask'] |
|
|
2025-07-10 14:32:59,342 - INFO - Creating datasets... |
|
|
2025-07-10 14:32:59,342 - INFO - Feature extractor output keys: ['input_features', 'attention_mask'] |
|
|
2025-07-10 14:32:59,342 - INFO - Feature extractor output keys: ['input_features', 'attention_mask'] |
|
|
2025-07-10 14:32:59,343 - INFO - Feature extractor output keys: ['input_features', 'attention_mask'] |
|
|
2025-07-10 14:32:59,343 - INFO - Creating data loaders... |
|
|
2025-07-10 14:32:59,343 - INFO - Checking a sample batch... |
|
|
2025-07-10 14:33:16,418 - INFO - input_ids_pos: torch.Size([48, 128]) |
|
|
2025-07-10 14:33:16,418 - INFO - attention_mask_pos: torch.Size([48, 128]) |
|
|
2025-07-10 14:33:16,418 - INFO - input_ids_neg: torch.Size([48, 128]) |
|
|
2025-07-10 14:33:16,418 - INFO - attention_mask_neg: torch.Size([48, 128]) |
|
|
2025-07-10 14:33:16,419 - INFO - input_values: torch.Size([48, 473, 160]) |
|
|
2025-07-10 14:33:16,420 - INFO - attention_mask_audio: torch.Size([48, 473]) |
|
|
2025-07-10 14:33:16,420 - INFO - is_corrupted: torch.Size([48]) |
|
|
2025-07-10 14:33:16,420 - INFO - correctness_scores: torch.Size([48]) |
|
|
2025-07-10 14:33:16,420 - INFO - Initializing model... |
|
|
2025-07-10 14:33:17,175 - INFO - Text encoder hidden dim: 768 |
|
|
2025-07-10 14:33:17,175 - INFO - Audio encoder hidden dim: 1024 |
|
|
2025-07-10 14:33:17,175 - INFO - Partial freezing: unfreezing last 3 text layers and 3 audio layers |
|
|
2025-07-10 14:33:17,175 - INFO - Unfreezing text encoder layer 9 |
|
|
2025-07-10 14:33:17,175 - INFO - Unfreezing text encoder layer 10 |
|
|
2025-07-10 14:33:17,175 - INFO - Unfreezing text encoder layer 11 |
|
|
2025-07-10 14:33:17,177 - INFO - Unfreezing audio encoder layer 21 |
|
|
2025-07-10 14:33:17,177 - INFO - Unfreezing audio encoder layer 22 |
|
|
2025-07-10 14:33:17,177 - INFO - Unfreezing audio encoder layer 23 |
|
|
2025-07-10 14:33:17,297 - INFO - Model initialized with 308,221,186 trainable parameters out of 879,798,082 total |
|
|
2025-07-10 14:33:18,208 - INFO - Using discriminative learning rates: encoder_lr=4e-05, main_lr=0.0008 |
|
|
2025-07-10 14:33:18,208 - INFO - Encoder parameters: 156, Non-encoder parameters: 38 |
|
|
2025-07-10 14:33:18,208 - INFO - Scheduler setup: |
|
|
2025-07-10 14:33:18,209 - INFO - Batches per epoch: 457 |
|
|
2025-07-10 14:33:18,209 - INFO - Accumulation steps: 15 |
|
|
2025-07-10 14:33:18,209 - INFO - Optimizer steps per epoch: 31 |
|
|
2025-07-10 14:33:18,209 - INFO - Total optimizer steps: 930 |
|
|
2025-07-10 14:33:18,209 - INFO - Warmup steps: 1000 |
|
|
2025-07-10 14:33:18,209 - INFO - Validating gradient accumulation setup... |
|
|
2025-07-10 14:33:18,209 - INFO - Validating gradient accumulation with 15 steps... |
|
|
2025-07-10 14:33:36,503 - WARNING - Not enough test batches (10) for accumulation_steps (15) |
|
|
2025-07-10 14:33:36,503 - INFO - Starting training for 30 epochs |
|
|
2025-07-10 14:35:17,701 - INFO - Training with parameters: |
|
|
2025-07-10 14:35:17,701 - INFO - Text model: sentence-transformers/paraphrase-multilingual-mpnet-base-v2 |
|
|
2025-07-10 14:35:17,701 - INFO - Audio model: facebook/w2v-bert-2.0 |
|
|
2025-07-10 14:35:17,701 - INFO - Freeze encoders: partial |
|
|
2025-07-10 14:35:17,701 - INFO - Text layers to unfreeze: 3 |
|
|
2025-07-10 14:35:17,701 - INFO - Audio layers to unfreeze: 3 |
|
|
2025-07-10 14:35:17,701 - INFO - Use cross-modal attention: False |
|
|
2025-07-10 14:35:17,701 - INFO - Use attentive pooling: False |
|
|
2025-07-10 14:35:17,701 - INFO - Use word-level alignment: True |
|
|
2025-07-10 14:35:17,701 - INFO - Batch size: 48 |
|
|
2025-07-10 14:35:17,701 - INFO - Gradient accumulation steps: 15 |
|
|
2025-07-10 14:35:17,701 - INFO - Effective batch size: 720 |
|
|
2025-07-10 14:35:17,701 - INFO - Mixed precision training: False |
|
|
2025-07-10 14:35:17,701 - INFO - Learning rate: 0.0008 |
|
|
2025-07-10 14:35:17,701 - INFO - Temperature: 0.1 |
|
|
2025-07-10 14:35:17,701 - INFO - Projection dimension: 768 |
|
|
2025-07-10 14:35:17,701 - INFO - Training samples: 21968 |
|
|
2025-07-10 14:35:17,701 - INFO - Validation samples: 9464 |
|
|
2025-07-10 14:35:17,701 - INFO - Test samples: 9467 |
|
|
2025-07-10 14:35:17,701 - INFO - Max audio length: 480000 samples (30.00 seconds at 16kHz) |
|
|
2025-07-10 14:35:17,701 - INFO - Loading tokenizer and feature extractor... |
|
|
2025-07-10 14:35:18,912 - INFO - Feature extractor output keys: ['input_features', 'attention_mask'] |
|
|
2025-07-10 14:35:18,912 - INFO - Creating datasets... |
|
|
2025-07-10 14:35:18,913 - INFO - Feature extractor output keys: ['input_features', 'attention_mask'] |
|
|
2025-07-10 14:35:18,913 - INFO - Feature extractor output keys: ['input_features', 'attention_mask'] |
|
|
2025-07-10 14:35:18,913 - INFO - Feature extractor output keys: ['input_features', 'attention_mask'] |
|
|
2025-07-10 14:35:18,913 - INFO - Creating data loaders... |
|
|
2025-07-10 14:35:18,914 - INFO - Checking a sample batch... |
|
|
2025-07-10 14:35:37,087 - INFO - input_ids_pos: torch.Size([48, 128]) |
|
|
2025-07-10 14:35:37,087 - INFO - attention_mask_pos: torch.Size([48, 128]) |
|
|
2025-07-10 14:35:37,087 - INFO - input_ids_neg: torch.Size([48, 128]) |
|
|
2025-07-10 14:35:37,087 - INFO - attention_mask_neg: torch.Size([48, 128]) |
|
|
2025-07-10 14:35:37,087 - INFO - input_values: torch.Size([48, 473, 160]) |
|
|
2025-07-10 14:35:37,087 - INFO - attention_mask_audio: torch.Size([48, 473]) |
|
|
2025-07-10 14:35:37,087 - INFO - is_corrupted: torch.Size([48]) |
|
|
2025-07-10 14:35:37,087 - INFO - correctness_scores: torch.Size([48]) |
|
|
2025-07-10 14:35:37,087 - INFO - Initializing model... |
|
|
2025-07-10 14:35:37,876 - INFO - Text encoder hidden dim: 768 |
|
|
2025-07-10 14:35:37,876 - INFO - Audio encoder hidden dim: 1024 |
|
|
2025-07-10 14:35:37,876 - INFO - Partial freezing: unfreezing last 3 text layers and 3 audio layers |
|
|
2025-07-10 14:35:37,876 - INFO - Unfreezing text encoder layer 9 |
|
|
2025-07-10 14:35:37,876 - INFO - Unfreezing text encoder layer 10 |
|
|
2025-07-10 14:35:37,876 - INFO - Unfreezing text encoder layer 11 |
|
|
2025-07-10 14:35:37,877 - INFO - Unfreezing audio encoder layer 21 |
|
|
2025-07-10 14:35:37,877 - INFO - Unfreezing audio encoder layer 22 |
|
|
2025-07-10 14:35:37,877 - INFO - Unfreezing audio encoder layer 23 |
|
|
2025-07-10 14:35:37,984 - INFO - Model initialized with 308,221,186 trainable parameters out of 879,798,082 total |
|
|
2025-07-10 14:35:38,812 - INFO - Using discriminative learning rates: encoder_lr=4e-05, main_lr=0.0008 |
|
|
2025-07-10 14:35:38,812 - INFO - Encoder parameters: 156, Non-encoder parameters: 38 |
|
|
2025-07-10 14:35:38,812 - INFO - Checking if loss parameters are in optimizer... |
|
|
2025-07-10 14:35:38,812 - INFO - ✓ log_sigma2_align is in optimizer |
|
|
2025-07-10 14:35:38,813 - INFO - Total parameters in optimizer: 194 |
|
|
2025-07-10 14:35:38,814 - INFO - Model parameters: 193 |
|
|
2025-07-10 14:35:38,814 - INFO - Loss parameters: 1 |
|
|
2025-07-10 14:35:38,814 - INFO - Scheduler setup: |
|
|
2025-07-10 14:35:38,814 - INFO - Batches per epoch: 457 |
|
|
2025-07-10 14:35:38,814 - INFO - Accumulation steps: 15 |
|
|
2025-07-10 14:35:38,814 - INFO - Optimizer steps per epoch: 31 |
|
|
2025-07-10 14:35:38,814 - INFO - Total optimizer steps: 930 |
|
|
2025-07-10 14:35:38,814 - INFO - Warmup steps: 1000 |
|
|
2025-07-10 14:35:38,814 - INFO - Validating gradient accumulation setup... |
|
|
2025-07-10 14:35:38,814 - INFO - Validating gradient accumulation with 15 steps... |
|
|
2025-07-10 14:35:57,487 - WARNING - Not enough test batches (10) for accumulation_steps (15) |
|
|
2025-07-10 14:35:57,487 - INFO - Starting training for 30 epochs |
|
|
2025-07-10 14:43:09,016 - INFO - Training with parameters: |
|
|
2025-07-10 14:43:09,016 - INFO - Text model: sentence-transformers/paraphrase-multilingual-mpnet-base-v2 |
|
|
2025-07-10 14:43:09,016 - INFO - Audio model: facebook/w2v-bert-2.0 |
|
|
2025-07-10 14:43:09,016 - INFO - Freeze encoders: partial |
|
|
2025-07-10 14:43:09,016 - INFO - Text layers to unfreeze: 3 |
|
|
2025-07-10 14:43:09,016 - INFO - Audio layers to unfreeze: 3 |
|
|
2025-07-10 14:43:09,016 - INFO - Use cross-modal attention: False |
|
|
2025-07-10 14:43:09,016 - INFO - Use attentive pooling: False |
|
|
2025-07-10 14:43:09,016 - INFO - Use word-level alignment: True |
|
|
2025-07-10 14:43:09,016 - INFO - Batch size: 48 |
|
|
2025-07-10 14:43:09,017 - INFO - Gradient accumulation steps: 15 |
|
|
2025-07-10 14:43:09,017 - INFO - Effective batch size: 720 |
|
|
2025-07-10 14:43:09,017 - INFO - Mixed precision training: False |
|
|
2025-07-10 14:43:09,017 - INFO - Learning rate: 0.0008 |
|
|
2025-07-10 14:43:09,017 - INFO - Temperature: 0.1 |
|
|
2025-07-10 14:43:09,017 - INFO - Projection dimension: 768 |
|
|
2025-07-10 14:43:09,017 - INFO - Training samples: 21968 |
|
|
2025-07-10 14:43:09,017 - INFO - Validation samples: 9464 |
|
|
2025-07-10 14:43:09,017 - INFO - Test samples: 9467 |
|
|
2025-07-10 14:43:09,017 - INFO - Max audio length: 480000 samples (30.00 seconds at 16kHz) |
|
|
2025-07-10 14:43:09,017 - INFO - Loading tokenizer and feature extractor... |
|
|
2025-07-10 14:43:10,076 - INFO - Feature extractor output keys: ['input_features', 'attention_mask'] |
|
|
2025-07-10 14:43:10,076 - INFO - Creating datasets... |
|
|
2025-07-10 14:43:10,076 - INFO - Feature extractor output keys: ['input_features', 'attention_mask'] |
|
|
2025-07-10 14:43:10,077 - INFO - Feature extractor output keys: ['input_features', 'attention_mask'] |
|
|
2025-07-10 14:43:10,077 - INFO - Feature extractor output keys: ['input_features', 'attention_mask'] |
|
|
2025-07-10 14:43:10,077 - INFO - Creating data loaders... |
|
|
2025-07-10 14:43:10,077 - INFO - Checking a sample batch... |
|
|
2025-07-10 14:43:28,271 - INFO - input_ids_pos: torch.Size([48, 128]) |
|
|
2025-07-10 14:43:28,272 - INFO - attention_mask_pos: torch.Size([48, 128]) |
|
|
2025-07-10 14:43:28,272 - INFO - input_ids_neg: torch.Size([48, 128]) |
|
|
2025-07-10 14:43:28,272 - INFO - attention_mask_neg: torch.Size([48, 128]) |
|
|
2025-07-10 14:43:28,272 - INFO - input_values: torch.Size([48, 473, 160]) |
|
|
2025-07-10 14:43:28,272 - INFO - attention_mask_audio: torch.Size([48, 473]) |
|
|
2025-07-10 14:43:28,272 - INFO - is_corrupted: torch.Size([48]) |
|
|
2025-07-10 14:43:28,272 - INFO - correctness_scores: torch.Size([48]) |
|
|
2025-07-10 14:43:28,272 - INFO - Initializing model... |
|
|
2025-07-10 14:43:29,031 - INFO - Text encoder hidden dim: 768 |
|
|
2025-07-10 14:43:29,031 - INFO - Audio encoder hidden dim: 1024 |
|
|
2025-07-10 14:43:29,031 - INFO - Partial freezing: unfreezing last 3 text layers and 3 audio layers |
|
|
2025-07-10 14:43:29,032 - INFO - Unfreezing text encoder layer 9 |
|
|
2025-07-10 14:43:29,032 - INFO - Unfreezing text encoder layer 10 |
|
|
2025-07-10 14:43:29,032 - INFO - Unfreezing text encoder layer 11 |
|
|
2025-07-10 14:43:29,033 - INFO - Unfreezing audio encoder layer 21 |
|
|
2025-07-10 14:43:29,033 - INFO - Unfreezing audio encoder layer 22 |
|
|
2025-07-10 14:43:29,033 - INFO - Unfreezing audio encoder layer 23 |
|
|
2025-07-10 14:43:29,143 - INFO - Model initialized with 308,221,186 trainable parameters out of 879,798,082 total |
|
|
2025-07-10 14:43:30,051 - INFO - Using discriminative learning rates: encoder_lr=4e-05, main_lr=0.0008 |
|
|
2025-07-10 14:43:30,051 - INFO - Encoder parameters: 156, Non-encoder parameters: 38 |
|
|
2025-07-10 14:43:30,051 - INFO - Checking if loss parameters are in optimizer... |
|
|
2025-07-10 14:43:30,051 - INFO - ✓ log_sigma2_align is in optimizer |
|
|
2025-07-10 14:43:30,051 - INFO - Total parameters in optimizer: 194 |
|
|
2025-07-10 14:43:30,053 - INFO - Model parameters: 193 |
|
|
2025-07-10 14:43:30,053 - INFO - Loss parameters: 1 |
|
|
2025-07-10 14:43:30,053 - INFO - Scheduler setup: |
|
|
2025-07-10 14:43:30,053 - INFO - Batches per epoch: 457 |
|
|
2025-07-10 14:43:30,053 - INFO - Accumulation steps: 15 |
|
|
2025-07-10 14:43:30,053 - INFO - Optimizer steps per epoch: 31 |
|
|
2025-07-10 14:43:30,053 - INFO - Total optimizer steps: 930 |
|
|
2025-07-10 14:43:30,053 - INFO - Warmup steps: 1000 |
|
|
2025-07-10 14:43:30,053 - INFO - Validating gradient accumulation setup... |
|
|
2025-07-10 14:43:30,053 - INFO - Validating gradient accumulation with 15 steps... |
|
|
2025-07-10 14:43:48,685 - WARNING - Not enough test batches (10) for accumulation_steps (15) |
|
|
2025-07-10 14:43:48,685 - INFO - Starting training for 30 epochs |
|
|
2025-07-10 14:44:57,042 - INFO - log_σ² gradient: -0.692917 |
|
|
2025-07-10 14:44:57,219 - INFO - Optimizer step 1: log_σ²=0.000000, weight=1.000000 |
|
|
2025-07-10 14:45:20,255 - INFO - log_σ² gradient: -0.695863 |
|
|
2025-07-10 14:45:20,327 - INFO - Optimizer step 2: log_σ²=0.000001, weight=0.999999 |
|
|
2025-07-10 14:45:43,849 - INFO - log_σ² gradient: -0.692190 |
|
|
2025-07-10 14:45:43,928 - INFO - Optimizer step 3: log_σ²=0.000002, weight=0.999998 |
|
|
2025-07-10 14:46:09,332 - INFO - log_σ² gradient: -0.691818 |
|
|
2025-07-10 14:46:09,410 - INFO - Optimizer step 4: log_σ²=0.000005, weight=0.999995 |
|
|
2025-07-10 14:46:35,717 - INFO - log_σ² gradient: -0.691842 |
|
|
2025-07-10 14:46:35,797 - INFO - Optimizer step 5: log_σ²=0.000008, weight=0.999992 |
|
|
2025-07-10 14:47:01,863 - INFO - log_σ² gradient: -0.688111 |
|
|
2025-07-10 14:47:01,941 - INFO - Optimizer step 6: log_σ²=0.000012, weight=0.999988 |
|
|
2025-07-10 14:47:26,008 - INFO - log_σ² gradient: -0.689173 |
|
|
2025-07-10 14:47:26,079 - INFO - Optimizer step 7: log_σ²=0.000017, weight=0.999983 |
|
|
2025-07-10 14:47:50,936 - INFO - log_σ² gradient: -0.684650 |
|
|
2025-07-10 14:47:51,002 - INFO - Optimizer step 8: log_σ²=0.000022, weight=0.999978 |
|
|
2025-07-10 14:48:15,512 - INFO - log_σ² gradient: -0.680450 |
|
|
2025-07-10 14:48:15,591 - INFO - Optimizer step 9: log_σ²=0.000029, weight=0.999971 |
|
|
2025-07-10 14:48:41,568 - INFO - log_σ² gradient: -0.678141 |
|
|
2025-07-10 14:48:41,643 - INFO - Optimizer step 10: log_σ²=0.000036, weight=0.999964 |
|
|
2025-07-10 14:49:06,737 - INFO - log_σ² gradient: -0.673729 |
|
|
2025-07-10 14:49:06,815 - INFO - Optimizer step 11: log_σ²=0.000044, weight=0.999956 |
|
|
2025-07-10 14:49:31,964 - INFO - log_σ² gradient: -0.671539 |
|
|
2025-07-10 14:49:32,043 - INFO - Optimizer step 12: log_σ²=0.000053, weight=0.999947 |
|
|
2025-07-10 14:49:56,967 - INFO - log_σ² gradient: -0.671065 |
|
|
2025-07-10 14:49:57,036 - INFO - Optimizer step 13: log_σ²=0.000062, weight=0.999938 |
|
|
2025-07-10 14:50:21,460 - INFO - log_σ² gradient: -0.668406 |
|
|
2025-07-10 14:50:21,538 - INFO - Optimizer step 14: log_σ²=0.000073, weight=0.999927 |
|
|
2025-07-10 14:50:46,211 - INFO - log_σ² gradient: -0.664372 |
|
|
2025-07-10 14:50:46,293 - INFO - Optimizer step 15: log_σ²=0.000084, weight=0.999916 |
|
|
2025-07-10 14:51:10,654 - INFO - log_σ² gradient: -0.660559 |
|
|
2025-07-10 14:51:10,730 - INFO - Optimizer step 16: log_σ²=0.000096, weight=0.999904 |
|
|
2025-07-10 14:51:37,535 - INFO - log_σ² gradient: -0.653596 |
|
|
2025-07-10 14:51:37,609 - INFO - Optimizer step 17: log_σ²=0.000108, weight=0.999892 |
|
|
2025-07-10 14:52:00,815 - INFO - log_σ² gradient: -0.652813 |
|
|
2025-07-10 14:52:00,885 - INFO - Optimizer step 18: log_σ²=0.000122, weight=0.999878 |
|
|
2025-07-10 14:52:25,599 - INFO - log_σ² gradient: -0.649483 |
|
|
2025-07-10 14:52:25,670 - INFO - Optimizer step 19: log_σ²=0.000136, weight=0.999864 |
|
|
2025-07-10 14:52:50,183 - INFO - log_σ² gradient: -0.640670 |
|
|
2025-07-10 14:52:50,255 - INFO - Optimizer step 20: log_σ²=0.000151, weight=0.999849 |
|
|
2025-07-10 14:53:15,912 - INFO - log_σ² gradient: -0.634642 |
|
|
2025-07-10 14:53:15,987 - INFO - Optimizer step 21: log_σ²=0.000167, weight=0.999833 |
|
|
2025-07-10 14:53:40,734 - INFO - log_σ² gradient: -0.633710 |
|
|
2025-07-10 14:53:40,812 - INFO - Optimizer step 22: log_σ²=0.000183, weight=0.999817 |
|
|
2025-07-10 14:54:05,851 - INFO - log_σ² gradient: -0.626418 |
|
|
2025-07-10 14:54:05,929 - INFO - Optimizer step 23: log_σ²=0.000200, weight=0.999800 |
|
|
2025-07-10 14:54:29,996 - INFO - log_σ² gradient: -0.617266 |
|
|
2025-07-10 14:54:30,072 - INFO - Optimizer step 24: log_σ²=0.000218, weight=0.999782 |
|
|
2025-07-10 14:54:54,697 - INFO - log_σ² gradient: -0.616343 |
|
|
2025-07-10 14:54:54,771 - INFO - Optimizer step 25: log_σ²=0.000237, weight=0.999763 |
|
|
2025-07-10 14:55:19,165 - INFO - log_σ² gradient: -0.609414 |
|
|
2025-07-10 14:55:19,238 - INFO - Optimizer step 26: log_σ²=0.000256, weight=0.999744 |
|
|
2025-07-10 14:55:41,931 - INFO - log_σ² gradient: -0.602047 |
|
|
2025-07-10 14:55:42,007 - INFO - Optimizer step 27: log_σ²=0.000277, weight=0.999723 |
|
|
2025-07-10 14:56:06,154 - INFO - log_σ² gradient: -0.601584 |
|
|
2025-07-10 14:56:06,226 - INFO - Optimizer step 28: log_σ²=0.000297, weight=0.999703 |
|
|
2025-07-10 14:56:30,010 - INFO - log_σ² gradient: -0.590199 |
|
|
2025-07-10 14:56:30,093 - INFO - Optimizer step 29: log_σ²=0.000319, weight=0.999681 |
|
|
2025-07-10 14:56:52,181 - INFO - log_σ² gradient: -0.592825 |
|
|
2025-07-10 14:56:52,259 - INFO - Optimizer step 30: log_σ²=0.000341, weight=0.999659 |
|
|
2025-07-10 14:57:03,019 - INFO - log_σ² gradient: -0.273802 |
|
|
2025-07-10 14:57:03,081 - INFO - Optimizer step 31: log_σ²=0.000363, weight=0.999637 |
|
|
2025-07-10 14:57:03,246 - INFO - Epoch 1: Total optimizer steps: 31 |
|
|
2025-07-10 15:00:35,021 - INFO - Validation metrics: |
|
|
2025-07-10 15:00:35,022 - INFO - Loss: 1.0307 |
|
|
2025-07-10 15:00:35,022 - INFO - Average similarity: 0.1485 |
|
|
2025-07-10 15:00:35,022 - INFO - Median similarity: 0.0839 |
|
|
2025-07-10 15:00:35,022 - INFO - Clean sample similarity: 0.1485 |
|
|
2025-07-10 15:00:35,022 - INFO - Corrupted sample similarity: 0.0914 |
|
|
2025-07-10 15:00:35,022 - INFO - Similarity gap (clean - corrupt): 0.0571 |
|
|
2025-07-10 15:00:35,132 - INFO - Epoch 1/30 - Train Loss: 1.2621, Val Loss: 1.0307, Clean Sim: 0.1485, Corrupt Sim: 0.0914, Gap: 0.0571, Time: 1006.45s |
|
|
2025-07-10 15:00:35,132 - INFO - New best validation loss: 1.0307 |
|
|
2025-07-10 15:00:42,818 - INFO - New best similarity gap: 0.0571 |
|
|
2025-07-10 15:01:59,961 - INFO - log_σ² gradient: -0.589916 |
|
|
2025-07-10 15:02:00,034 - INFO - Optimizer step 1: log_σ²=0.000386, weight=0.999614 |
|
|
2025-07-10 15:02:26,656 - INFO - log_σ² gradient: -0.579981 |
|
|
2025-07-10 15:02:26,733 - INFO - Optimizer step 2: log_σ²=0.000409, weight=0.999591 |
|
|
2025-07-10 15:02:52,532 - INFO - log_σ² gradient: -0.589532 |
|
|
2025-07-10 15:02:52,599 - INFO - Optimizer step 3: log_σ²=0.000433, weight=0.999567 |
|
|
2025-07-10 15:03:15,543 - INFO - log_σ² gradient: -0.583195 |
|
|
2025-07-10 15:03:15,614 - INFO - Optimizer step 4: log_σ²=0.000459, weight=0.999542 |
|
|
2025-07-10 15:03:39,345 - INFO - log_σ² gradient: -0.589153 |
|
|
2025-07-10 15:03:39,415 - INFO - Optimizer step 5: log_σ²=0.000484, weight=0.999516 |
|
|
2025-07-10 15:04:05,842 - INFO - log_σ² gradient: -0.602421 |
|
|
2025-07-10 15:04:05,920 - INFO - Optimizer step 6: log_σ²=0.000511, weight=0.999489 |
|
|
2025-07-10 15:04:31,197 - INFO - log_σ² gradient: -0.581654 |
|
|
2025-07-10 15:04:31,268 - INFO - Optimizer step 7: log_σ²=0.000538, weight=0.999462 |
|
|
2025-07-10 15:04:57,434 - INFO - log_σ² gradient: -0.590047 |
|
|
2025-07-10 15:04:57,512 - INFO - Optimizer step 8: log_σ²=0.000567, weight=0.999433 |
|
|
2025-07-10 15:05:21,811 - INFO - log_σ² gradient: -0.590955 |
|
|
2025-07-10 15:05:21,885 - INFO - Optimizer step 9: log_σ²=0.000596, weight=0.999404 |
|
|
2025-07-10 15:05:46,005 - INFO - log_σ² gradient: -0.593282 |
|
|
2025-07-10 15:05:46,084 - INFO - Optimizer step 10: log_σ²=0.000626, weight=0.999375 |
|
|
2025-07-10 15:06:11,783 - INFO - log_σ² gradient: -0.590858 |
|
|
2025-07-10 15:06:11,856 - INFO - Optimizer step 11: log_σ²=0.000656, weight=0.999344 |
|
|
2025-07-10 15:06:36,267 - INFO - log_σ² gradient: -0.574171 |
|
|
2025-07-10 15:06:36,349 - INFO - Optimizer step 12: log_σ²=0.000688, weight=0.999313 |
|
|
2025-07-10 15:06:59,907 - INFO - log_σ² gradient: -0.570794 |
|
|
2025-07-10 15:06:59,979 - INFO - Optimizer step 13: log_σ²=0.000720, weight=0.999281 |
|
|
2025-07-10 15:07:24,224 - INFO - log_σ² gradient: -0.575564 |
|
|
2025-07-10 15:07:24,295 - INFO - Optimizer step 14: log_σ²=0.000753, weight=0.999248 |
|
|
2025-07-10 15:07:48,635 - INFO - log_σ² gradient: -0.579865 |
|
|
2025-07-10 15:07:48,713 - INFO - Optimizer step 15: log_σ²=0.000786, weight=0.999214 |
|
|
2025-07-10 15:08:12,965 - INFO - log_σ² gradient: -0.582150 |
|
|
2025-07-10 15:08:13,029 - INFO - Optimizer step 16: log_σ²=0.000821, weight=0.999180 |
|
|
2025-07-10 15:08:37,923 - INFO - log_σ² gradient: -0.591940 |
|
|
2025-07-10 15:08:37,997 - INFO - Optimizer step 17: log_σ²=0.000856, weight=0.999144 |
|
|
2025-07-10 15:09:01,687 - INFO - log_σ² gradient: -0.581033 |
|
|
2025-07-10 15:09:01,762 - INFO - Optimizer step 18: log_σ²=0.000892, weight=0.999109 |
|
|
2025-07-10 15:09:27,017 - INFO - log_σ² gradient: -0.578886 |
|
|
2025-07-10 15:09:27,095 - INFO - Optimizer step 19: log_σ²=0.000929, weight=0.999072 |
|
|
2025-07-10 15:09:51,073 - INFO - log_σ² gradient: -0.582442 |
|
|
2025-07-10 15:09:51,138 - INFO - Optimizer step 20: log_σ²=0.000966, weight=0.999034 |
|
|
2025-07-10 15:10:18,203 - INFO - log_σ² gradient: -0.582159 |
|
|
2025-07-10 15:10:18,282 - INFO - Optimizer step 21: log_σ²=0.001005, weight=0.998996 |
|
|
2025-07-10 15:10:46,044 - INFO - log_σ² gradient: -0.568884 |
|
|
2025-07-10 15:10:46,118 - INFO - Optimizer step 22: log_σ²=0.001044, weight=0.998957 |
|
|
2025-07-10 15:11:10,700 - INFO - log_σ² gradient: -0.579983 |
|
|
2025-07-10 15:11:10,776 - INFO - Optimizer step 23: log_σ²=0.001083, weight=0.998917 |
|
|
2025-07-10 15:11:36,451 - INFO - log_σ² gradient: -0.577635 |
|
|
2025-07-10 15:11:36,528 - INFO - Optimizer step 24: log_σ²=0.001124, weight=0.998877 |
|
|
2025-07-10 15:12:01,944 - INFO - log_σ² gradient: -0.571526 |
|
|
2025-07-10 15:12:02,016 - INFO - Optimizer step 25: log_σ²=0.001165, weight=0.998835 |
|
|
2025-07-10 15:12:26,706 - INFO - log_σ² gradient: -0.570213 |
|
|
2025-07-10 15:12:26,777 - INFO - Optimizer step 26: log_σ²=0.001207, weight=0.998793 |
|
|
2025-07-10 15:12:50,859 - INFO - log_σ² gradient: -0.562888 |
|
|
2025-07-10 15:12:50,927 - INFO - Optimizer step 27: log_σ²=0.001250, weight=0.998751 |
|
|
2025-07-10 15:13:14,390 - INFO - log_σ² gradient: -0.570238 |
|
|
2025-07-10 15:13:14,464 - INFO - Optimizer step 28: log_σ²=0.001294, weight=0.998707 |
|
|
2025-07-10 15:13:38,330 - INFO - log_σ² gradient: -0.586211 |
|
|
2025-07-10 15:13:38,407 - INFO - Optimizer step 29: log_σ²=0.001338, weight=0.998663 |
|
|
2025-07-10 15:14:02,545 - INFO - log_σ² gradient: -0.576393 |
|
|
2025-07-10 15:14:02,615 - INFO - Optimizer step 30: log_σ²=0.001383, weight=0.998618 |
|
|
2025-07-10 15:14:14,044 - INFO - log_σ² gradient: -0.268553 |
|
|
2025-07-10 15:14:14,114 - INFO - Optimizer step 31: log_σ²=0.001427, weight=0.998574 |
|
|
2025-07-10 15:14:14,281 - INFO - Epoch 2: Total optimizer steps: 31 |
|
|
2025-07-10 15:17:42,056 - INFO - Validation metrics: |
|
|
2025-07-10 15:17:42,057 - INFO - Loss: 0.8922 |
|
|
2025-07-10 15:17:42,057 - INFO - Average similarity: 0.5403 |
|
|
2025-07-10 15:17:42,057 - INFO - Median similarity: 0.6270 |
|
|
2025-07-10 15:17:42,057 - INFO - Clean sample similarity: 0.5403 |
|
|
2025-07-10 15:17:42,057 - INFO - Corrupted sample similarity: 0.3360 |
|
|
2025-07-10 15:17:42,057 - INFO - Similarity gap (clean - corrupt): 0.2043 |
|
|
2025-07-10 15:17:42,224 - INFO - Epoch 2/30 - Train Loss: 0.9801, Val Loss: 0.8922, Clean Sim: 0.5403, Corrupt Sim: 0.3360, Gap: 0.2043, Time: 1012.18s |
|
|
2025-07-10 15:17:42,225 - INFO - New best validation loss: 0.8922 |
|
|
2025-07-10 15:17:48,996 - INFO - New best similarity gap: 0.2043 |
|
|
2025-07-10 15:20:51,099 - INFO - Epoch 2 Validation Alignment: Pos=0.097, Neg=0.091, Gap=0.006 |
|
|
2025-07-10 15:22:04,315 - INFO - log_σ² gradient: -0.578207 |
|
|
2025-07-10 15:22:04,395 - INFO - Optimizer step 1: log_σ²=0.001472, weight=0.998529 |
|
|
2025-07-10 15:22:27,156 - INFO - log_σ² gradient: -0.576470 |
|
|
2025-07-10 15:22:27,224 - INFO - Optimizer step 2: log_σ²=0.001517, weight=0.998484 |
|
|
2025-07-10 15:22:49,802 - INFO - log_σ² gradient: -0.580390 |
|
|
2025-07-10 15:22:49,880 - INFO - Optimizer step 3: log_σ²=0.001564, weight=0.998437 |
|
|
2025-07-10 15:23:14,947 - INFO - log_σ² gradient: -0.571025 |
|
|
2025-07-10 15:23:15,025 - INFO - Optimizer step 4: log_σ²=0.001612, weight=0.998390 |
|
|
2025-07-10 15:23:39,787 - INFO - log_σ² gradient: -0.568302 |
|
|
2025-07-10 15:23:39,862 - INFO - Optimizer step 5: log_σ²=0.001660, weight=0.998341 |
|
|
2025-07-10 15:24:05,250 - INFO - log_σ² gradient: -0.569555 |
|
|
2025-07-10 15:24:05,320 - INFO - Optimizer step 6: log_σ²=0.001710, weight=0.998292 |
|
|
2025-07-10 15:24:31,361 - INFO - log_σ² gradient: -0.573536 |
|
|
2025-07-10 15:24:31,434 - INFO - Optimizer step 7: log_σ²=0.001760, weight=0.998242 |
|
|
2025-07-10 15:24:56,142 - INFO - log_σ² gradient: -0.572885 |
|
|
2025-07-10 15:24:56,223 - INFO - Optimizer step 8: log_σ²=0.001811, weight=0.998191 |
|
|
2025-07-10 15:25:20,126 - INFO - log_σ² gradient: -0.558845 |
|
|
2025-07-10 15:25:20,202 - INFO - Optimizer step 9: log_σ²=0.001863, weight=0.998139 |
|
|
2025-07-10 15:25:43,713 - INFO - log_σ² gradient: -0.560663 |
|
|
2025-07-10 15:25:43,784 - INFO - Optimizer step 10: log_σ²=0.001916, weight=0.998086 |
|
|
2025-07-10 15:26:07,361 - INFO - log_σ² gradient: -0.563766 |
|
|
2025-07-10 15:26:07,432 - INFO - Optimizer step 11: log_σ²=0.001969, weight=0.998033 |
|
|
2025-07-10 15:26:31,649 - INFO - log_σ² gradient: -0.565848 |
|
|
2025-07-10 15:26:31,712 - INFO - Optimizer step 12: log_σ²=0.002024, weight=0.997978 |
|
|
2025-07-10 15:26:56,443 - INFO - log_σ² gradient: -0.576774 |
|
|
2025-07-10 15:26:56,518 - INFO - Optimizer step 13: log_σ²=0.002079, weight=0.997923 |
|
|
2025-07-10 15:27:18,895 - INFO - log_σ² gradient: -0.576104 |
|
|
2025-07-10 15:27:18,970 - INFO - Optimizer step 14: log_σ²=0.002135, weight=0.997867 |
|
|
2025-07-10 15:27:44,059 - INFO - log_σ² gradient: -0.570861 |
|
|
2025-07-10 15:27:44,123 - INFO - Optimizer step 15: log_σ²=0.002192, weight=0.997810 |
|
|
2025-07-10 15:28:07,333 - INFO - log_σ² gradient: -0.568999 |
|
|
2025-07-10 15:28:07,399 - INFO - Optimizer step 16: log_σ²=0.002250, weight=0.997752 |
|
|
2025-07-10 15:28:32,532 - INFO - log_σ² gradient: -0.562618 |
|
|
2025-07-10 15:28:32,599 - INFO - Optimizer step 17: log_σ²=0.002309, weight=0.997694 |
|
|
2025-07-10 15:28:57,183 - INFO - log_σ² gradient: -0.555216 |
|
|
2025-07-10 15:28:57,256 - INFO - Optimizer step 18: log_σ²=0.002368, weight=0.997635 |
|
|
2025-07-10 15:29:21,199 - INFO - log_σ² gradient: -0.571507 |
|
|
2025-07-10 15:29:21,265 - INFO - Optimizer step 19: log_σ²=0.002428, weight=0.997575 |
|
|
2025-07-10 15:29:45,535 - INFO - log_σ² gradient: -0.574533 |
|
|
2025-07-10 15:29:45,610 - INFO - Optimizer step 20: log_σ²=0.002490, weight=0.997514 |
|
|
2025-07-10 15:30:09,763 - INFO - log_σ² gradient: -0.555783 |
|
|
2025-07-10 15:30:09,841 - INFO - Optimizer step 21: log_σ²=0.002551, weight=0.997452 |
|
|
2025-07-10 15:30:33,760 - INFO - log_σ² gradient: -0.570753 |
|
|
2025-07-10 15:30:33,828 - INFO - Optimizer step 22: log_σ²=0.002614, weight=0.997389 |
|
|
2025-07-10 15:30:58,346 - INFO - log_σ² gradient: -0.559311 |
|
|
2025-07-10 15:30:58,422 - INFO - Optimizer step 23: log_σ²=0.002677, weight=0.997326 |
|
|
2025-07-10 15:31:21,723 - INFO - log_σ² gradient: -0.563625 |
|
|
2025-07-10 15:31:21,787 - INFO - Optimizer step 24: log_σ²=0.002742, weight=0.997262 |
|
|
2025-07-10 15:31:45,719 - INFO - log_σ² gradient: -0.571170 |
|
|
2025-07-10 15:31:45,791 - INFO - Optimizer step 25: log_σ²=0.002807, weight=0.997197 |
|
|
2025-07-10 15:32:10,102 - INFO - log_σ² gradient: -0.558868 |
|
|
2025-07-10 15:32:10,178 - INFO - Optimizer step 26: log_σ²=0.002872, weight=0.997132 |
|
|
2025-07-10 15:32:35,765 - INFO - log_σ² gradient: -0.562836 |
|
|
2025-07-10 15:32:35,836 - INFO - Optimizer step 27: log_σ²=0.002939, weight=0.997065 |
|
|
2025-07-10 15:33:00,001 - INFO - log_σ² gradient: -0.571679 |
|
|
2025-07-10 15:33:00,072 - INFO - Optimizer step 28: log_σ²=0.003006, weight=0.996998 |
|
|
2025-07-10 15:33:23,522 - INFO - log_σ² gradient: -0.559068 |
|
|
2025-07-10 15:33:23,598 - INFO - Optimizer step 29: log_σ²=0.003074, weight=0.996930 |
|
|
2025-07-10 15:33:47,226 - INFO - log_σ² gradient: -0.550306 |
|
|
2025-07-10 15:33:47,297 - INFO - Optimizer step 30: log_σ²=0.003143, weight=0.996862 |
|
|
2025-07-10 15:33:58,792 - INFO - log_σ² gradient: -0.268167 |
|
|
2025-07-10 15:33:58,871 - INFO - Optimizer step 31: log_σ²=0.003209, weight=0.996796 |
|
|
2025-07-10 15:33:59,040 - INFO - Epoch 3: Total optimizer steps: 31 |
|
|
2025-07-10 15:37:19,614 - INFO - Validation metrics: |
|
|
2025-07-10 15:37:19,614 - INFO - Loss: 0.8290 |
|
|
2025-07-10 15:37:19,614 - INFO - Average similarity: 0.6453 |
|
|
2025-07-10 15:37:19,614 - INFO - Median similarity: 0.8567 |
|
|
2025-07-10 15:37:19,614 - INFO - Clean sample similarity: 0.6453 |
|
|
2025-07-10 15:37:19,614 - INFO - Corrupted sample similarity: 0.4012 |
|
|
2025-07-10 15:37:19,614 - INFO - Similarity gap (clean - corrupt): 0.2441 |
|
|
2025-07-10 15:37:19,742 - INFO - Epoch 3/30 - Train Loss: 0.8919, Val Loss: 0.8290, Clean Sim: 0.6453, Corrupt Sim: 0.4012, Gap: 0.2441, Time: 988.64s |
|
|
2025-07-10 15:37:19,742 - INFO - New best validation loss: 0.8290 |
|
|
2025-07-10 15:37:26,344 - INFO - New best similarity gap: 0.2441 |
|
|
2025-07-10 15:38:45,108 - INFO - log_σ² gradient: -0.559397 |
|
|
2025-07-10 15:38:45,184 - INFO - Optimizer step 1: log_σ²=0.003277, weight=0.996729 |
|
|
2025-07-10 15:39:09,176 - INFO - log_σ² gradient: -0.572586 |
|
|
2025-07-10 15:39:09,250 - INFO - Optimizer step 2: log_σ²=0.003345, weight=0.996660 |
|
|
2025-07-10 15:39:33,980 - INFO - log_σ² gradient: -0.565342 |
|
|
2025-07-10 15:39:34,054 - INFO - Optimizer step 3: log_σ²=0.003415, weight=0.996591 |
|
|
2025-07-10 15:39:58,662 - INFO - log_σ² gradient: -0.563123 |
|
|
2025-07-10 15:39:58,735 - INFO - Optimizer step 4: log_σ²=0.003485, weight=0.996521 |
|
|
2025-07-10 15:40:22,668 - INFO - log_σ² gradient: -0.564826 |
|
|
2025-07-10 15:40:22,736 - INFO - Optimizer step 5: log_σ²=0.003557, weight=0.996449 |
|
|
2025-07-10 15:40:47,343 - INFO - log_σ² gradient: -0.566771 |
|
|
2025-07-10 15:40:47,415 - INFO - Optimizer step 6: log_σ²=0.003630, weight=0.996377 |
|
|
2025-07-10 15:41:12,866 - INFO - log_σ² gradient: -0.567160 |
|
|
2025-07-10 15:41:12,930 - INFO - Optimizer step 7: log_σ²=0.003703, weight=0.996303 |
|
|
2025-07-10 15:41:37,897 - INFO - log_σ² gradient: -0.564071 |
|
|
2025-07-10 15:41:37,970 - INFO - Optimizer step 8: log_σ²=0.003778, weight=0.996229 |
|
|
2025-07-10 15:42:02,262 - INFO - log_σ² gradient: -0.551527 |
|
|
2025-07-10 15:42:02,340 - INFO - Optimizer step 9: log_σ²=0.003854, weight=0.996154 |
|
|
2025-07-10 15:42:26,405 - INFO - log_σ² gradient: -0.566299 |
|
|
2025-07-10 15:42:26,475 - INFO - Optimizer step 10: log_σ²=0.003930, weight=0.996078 |
|
|
2025-07-10 15:42:49,803 - INFO - log_σ² gradient: -0.569711 |
|
|
2025-07-10 15:42:49,869 - INFO - Optimizer step 11: log_σ²=0.004007, weight=0.996001 |
|
|
2025-07-10 15:43:15,239 - INFO - log_σ² gradient: -0.554122 |
|
|
2025-07-10 15:43:15,310 - INFO - Optimizer step 12: log_σ²=0.004086, weight=0.995923 |
|
|
2025-07-10 15:43:40,085 - INFO - log_σ² gradient: -0.555365 |
|
|
2025-07-10 15:43:40,155 - INFO - Optimizer step 13: log_σ²=0.004165, weight=0.995844 |
|
|
2025-07-10 15:44:04,897 - INFO - log_σ² gradient: -0.571109 |
|
|
2025-07-10 15:44:04,974 - INFO - Optimizer step 14: log_σ²=0.004245, weight=0.995764 |
|
|
2025-07-10 15:44:29,819 - INFO - log_σ² gradient: -0.564269 |
|
|
2025-07-10 15:44:29,892 - INFO - Optimizer step 15: log_σ²=0.004326, weight=0.995684 |
|
|
2025-07-10 15:44:55,117 - INFO - log_σ² gradient: -0.568096 |
|
|
2025-07-10 15:44:55,189 - INFO - Optimizer step 16: log_σ²=0.004408, weight=0.995602 |
|
|
2025-07-10 15:45:20,110 - INFO - log_σ² gradient: -0.565259 |
|
|
2025-07-10 15:45:20,183 - INFO - Optimizer step 17: log_σ²=0.004490, weight=0.995520 |
|
|
2025-07-10 15:45:46,055 - INFO - log_σ² gradient: -0.555401 |
|
|
2025-07-10 15:45:46,129 - INFO - Optimizer step 18: log_σ²=0.004574, weight=0.995437 |
|
|
2025-07-10 15:46:11,909 - INFO - log_σ² gradient: -0.548517 |
|
|
2025-07-10 15:46:11,987 - INFO - Optimizer step 19: log_σ²=0.004658, weight=0.995353 |
|
|
2025-07-10 15:46:34,856 - INFO - log_σ² gradient: -0.554247 |
|
|
2025-07-10 15:46:34,926 - INFO - Optimizer step 20: log_σ²=0.004743, weight=0.995268 |
|
|
2025-07-10 15:46:58,586 - INFO - log_σ² gradient: -0.558437 |
|
|
2025-07-10 15:46:58,653 - INFO - Optimizer step 21: log_σ²=0.004829, weight=0.995183 |
|
|
2025-07-10 15:47:21,986 - INFO - log_σ² gradient: -0.563944 |
|
|
2025-07-10 15:47:22,062 - INFO - Optimizer step 22: log_σ²=0.004915, weight=0.995097 |
|
|
2025-07-10 15:47:46,871 - INFO - log_σ² gradient: -0.569697 |
|
|
2025-07-10 15:47:46,936 - INFO - Optimizer step 23: log_σ²=0.005003, weight=0.995009 |
|
|
2025-07-10 15:48:10,787 - INFO - log_σ² gradient: -0.561981 |
|
|
2025-07-10 15:48:10,858 - INFO - Optimizer step 24: log_σ²=0.005092, weight=0.994921 |
|
|
2025-07-10 15:48:34,605 - INFO - log_σ² gradient: -0.563767 |
|
|
2025-07-10 15:48:34,679 - INFO - Optimizer step 25: log_σ²=0.005181, weight=0.994833 |
|
|
2025-07-10 15:48:57,793 - INFO - log_σ² gradient: -0.572381 |
|
|
2025-07-10 15:48:57,865 - INFO - Optimizer step 26: log_σ²=0.005271, weight=0.994743 |
|
|
2025-07-10 15:49:22,882 - INFO - log_σ² gradient: -0.557778 |
|
|
2025-07-10 15:49:22,957 - INFO - Optimizer step 27: log_σ²=0.005362, weight=0.994652 |
|
|
2025-07-10 15:49:47,629 - INFO - log_σ² gradient: -0.550403 |
|
|
2025-07-10 15:49:47,700 - INFO - Optimizer step 28: log_σ²=0.005454, weight=0.994561 |
|
|
2025-07-10 15:50:12,363 - INFO - log_σ² gradient: -0.558082 |
|
|
2025-07-10 15:50:12,433 - INFO - Optimizer step 29: log_σ²=0.005546, weight=0.994469 |
|
|
2025-07-10 15:50:35,027 - INFO - log_σ² gradient: -0.563257 |
|
|
2025-07-10 15:50:35,105 - INFO - Optimizer step 30: log_σ²=0.005640, weight=0.994376 |
|
|
2025-07-10 15:50:46,220 - INFO - log_σ² gradient: -0.271958 |
|
|
2025-07-10 15:50:46,287 - INFO - Optimizer step 31: log_σ²=0.005729, weight=0.994287 |
|
|
2025-07-10 15:50:46,498 - INFO - Epoch 4: Total optimizer steps: 31 |
|
|
2025-07-10 15:54:05,705 - INFO - Validation metrics: |
|
|
2025-07-10 15:54:05,705 - INFO - Loss: 0.8083 |
|
|
2025-07-10 15:54:05,705 - INFO - Average similarity: 0.6493 |
|
|
2025-07-10 15:54:05,705 - INFO - Median similarity: 0.8758 |
|
|
2025-07-10 15:54:05,705 - INFO - Clean sample similarity: 0.6493 |
|
|
2025-07-10 15:54:05,705 - INFO - Corrupted sample similarity: 0.3713 |
|
|
2025-07-10 15:54:05,705 - INFO - Similarity gap (clean - corrupt): 0.2780 |
|
|
2025-07-10 15:54:05,831 - INFO - Epoch 4/30 - Train Loss: 0.8501, Val Loss: 0.8083, Clean Sim: 0.6493, Corrupt Sim: 0.3713, Gap: 0.2780, Time: 991.63s |
|
|
2025-07-10 15:54:05,832 - INFO - New best validation loss: 0.8083 |
|
|
2025-07-10 15:54:11,918 - INFO - New best similarity gap: 0.2780 |
|
|
2025-07-10 15:56:58,215 - INFO - Epoch 4 Validation Alignment: Pos=0.113, Neg=0.100, Gap=0.013 |
|
|
2025-07-10 15:58:06,574 - INFO - log_σ² gradient: -0.569854 |
|
|
2025-07-10 15:58:06,646 - INFO - Optimizer step 1: log_σ²=0.005820, weight=0.994197 |
|
|
2025-07-10 15:58:30,177 - INFO - log_σ² gradient: -0.559292 |
|
|
2025-07-10 15:58:30,253 - INFO - Optimizer step 2: log_σ²=0.005912, weight=0.994105 |
|
|
2025-07-10 15:58:54,829 - INFO - log_σ² gradient: -0.558458 |
|
|
2025-07-10 15:58:54,905 - INFO - Optimizer step 3: log_σ²=0.006006, weight=0.994012 |
|
|
2025-07-10 15:59:20,262 - INFO - log_σ² gradient: -0.556274 |
|
|
2025-07-10 15:59:20,341 - INFO - Optimizer step 4: log_σ²=0.006100, weight=0.993919 |
|
|
2025-07-10 15:59:43,766 - INFO - log_σ² gradient: -0.553982 |
|
|
2025-07-10 15:59:43,840 - INFO - Optimizer step 5: log_σ²=0.006195, weight=0.993824 |
|
|
2025-07-10 16:00:08,879 - INFO - log_σ² gradient: -0.558455 |
|
|
2025-07-10 16:00:08,955 - INFO - Optimizer step 6: log_σ²=0.006292, weight=0.993728 |
|
|
2025-07-10 16:00:32,328 - INFO - log_σ² gradient: -0.561167 |
|
|
2025-07-10 16:00:32,391 - INFO - Optimizer step 7: log_σ²=0.006389, weight=0.993631 |
|
|
2025-07-10 16:00:56,801 - INFO - log_σ² gradient: -0.558181 |
|
|
2025-07-10 16:00:56,880 - INFO - Optimizer step 8: log_σ²=0.006488, weight=0.993533 |
|
|
2025-07-10 16:01:20,369 - INFO - log_σ² gradient: -0.562367 |
|
|
2025-07-10 16:01:20,441 - INFO - Optimizer step 9: log_σ²=0.006587, weight=0.993435 |
|
|
2025-07-10 16:01:43,359 - INFO - log_σ² gradient: -0.562431 |
|
|
2025-07-10 16:01:43,427 - INFO - Optimizer step 10: log_σ²=0.006688, weight=0.993335 |
|
|
2025-07-10 16:02:07,447 - INFO - log_σ² gradient: -0.556597 |
|
|
2025-07-10 16:02:07,518 - INFO - Optimizer step 11: log_σ²=0.006789, weight=0.993234 |
|
|
2025-07-10 16:02:30,562 - INFO - log_σ² gradient: -0.544444 |
|
|
2025-07-10 16:02:30,634 - INFO - Optimizer step 12: log_σ²=0.006891, weight=0.993132 |
|
|
2025-07-10 16:02:54,322 - INFO - log_σ² gradient: -0.548111 |
|
|
2025-07-10 16:02:54,394 - INFO - Optimizer step 13: log_σ²=0.006994, weight=0.993030 |
|
|
2025-07-10 16:03:20,207 - INFO - log_σ² gradient: -0.556608 |
|
|
2025-07-10 16:03:20,287 - INFO - Optimizer step 14: log_σ²=0.007098, weight=0.992927 |
|
|
2025-07-10 16:03:44,093 - INFO - log_σ² gradient: -0.549005 |
|
|
2025-07-10 16:03:44,164 - INFO - Optimizer step 15: log_σ²=0.007203, weight=0.992823 |
|
|
2025-07-10 16:04:08,498 - INFO - log_σ² gradient: -0.547460 |
|
|
2025-07-10 16:04:08,572 - INFO - Optimizer step 16: log_σ²=0.007308, weight=0.992719 |
|
|
2025-07-10 16:04:31,481 - INFO - log_σ² gradient: -0.558622 |
|
|
2025-07-10 16:04:31,542 - INFO - Optimizer step 17: log_σ²=0.007414, weight=0.992613 |
|
|
2025-07-10 16:04:55,148 - INFO - log_σ² gradient: -0.560603 |
|
|
2025-07-10 16:04:55,220 - INFO - Optimizer step 18: log_σ²=0.007522, weight=0.992507 |
|
|
2025-07-10 16:05:18,903 - INFO - log_σ² gradient: -0.546133 |
|
|
2025-07-10 16:05:18,978 - INFO - Optimizer step 19: log_σ²=0.007630, weight=0.992399 |
|
|
2025-07-10 16:05:41,872 - INFO - log_σ² gradient: -0.552031 |
|
|
2025-07-10 16:05:41,943 - INFO - Optimizer step 20: log_σ²=0.007738, weight=0.992291 |
|
|
2025-07-10 16:06:05,442 - INFO - log_σ² gradient: -0.545588 |
|
|
2025-07-10 16:06:05,513 - INFO - Optimizer step 21: log_σ²=0.007848, weight=0.992183 |
|
|
2025-07-10 16:06:32,189 - INFO - log_σ² gradient: -0.559624 |
|
|
2025-07-10 16:06:32,264 - INFO - Optimizer step 22: log_σ²=0.007958, weight=0.992073 |
|
|
2025-07-10 16:06:55,749 - INFO - log_σ² gradient: -0.548875 |
|
|
2025-07-10 16:06:55,820 - INFO - Optimizer step 23: log_σ²=0.008070, weight=0.991963 |
|
|
2025-07-10 16:07:19,891 - INFO - log_σ² gradient: -0.561421 |
|
|
2025-07-10 16:07:19,965 - INFO - Optimizer step 24: log_σ²=0.008182, weight=0.991852 |
|
|
2025-07-10 16:07:45,055 - INFO - log_σ² gradient: -0.558281 |
|
|
2025-07-10 16:07:45,134 - INFO - Optimizer step 25: log_σ²=0.008295, weight=0.991739 |
|
|
2025-07-10 16:08:09,691 - INFO - log_σ² gradient: -0.552824 |
|
|
2025-07-10 16:08:09,763 - INFO - Optimizer step 26: log_σ²=0.008409, weight=0.991626 |
|
|
2025-07-10 16:08:34,987 - INFO - log_σ² gradient: -0.549287 |
|
|
2025-07-10 16:08:35,062 - INFO - Optimizer step 27: log_σ²=0.008523, weight=0.991513 |
|
|
2025-07-10 16:09:00,271 - INFO - log_σ² gradient: -0.545912 |
|
|
2025-07-10 16:09:00,347 - INFO - Optimizer step 28: log_σ²=0.008639, weight=0.991398 |
|
|
2025-07-10 16:09:25,166 - INFO - log_σ² gradient: -0.559556 |
|
|
2025-07-10 16:09:25,248 - INFO - Optimizer step 29: log_σ²=0.008755, weight=0.991283 |
|
|
2025-07-10 16:09:48,499 - INFO - log_σ² gradient: -0.553742 |
|
|
2025-07-10 16:09:48,582 - INFO - Optimizer step 30: log_σ²=0.008872, weight=0.991167 |
|
|
2025-07-10 16:10:00,481 - INFO - log_σ² gradient: -0.264338 |
|
|
2025-07-10 16:10:00,555 - INFO - Optimizer step 31: log_σ²=0.008984, weight=0.991056 |
|
|
2025-07-10 16:10:00,745 - INFO - Epoch 5: Total optimizer steps: 31 |
|
|
2025-07-10 16:13:20,132 - INFO - Validation metrics: |
|
|
2025-07-10 16:13:20,132 - INFO - Loss: 0.9651 |
|
|
2025-07-10 16:13:20,132 - INFO - Average similarity: 0.8196 |
|
|
2025-07-10 16:13:20,132 - INFO - Median similarity: 0.9827 |
|
|
2025-07-10 16:13:20,132 - INFO - Clean sample similarity: 0.8196 |
|
|
2025-07-10 16:13:20,132 - INFO - Corrupted sample similarity: 0.5286 |
|
|
2025-07-10 16:13:20,133 - INFO - Similarity gap (clean - corrupt): 0.2909 |
|
|
2025-07-10 16:13:20,264 - INFO - Epoch 5/30 - Train Loss: 0.8336, Val Loss: 0.9651, Clean Sim: 0.8196, Corrupt Sim: 0.5286, Gap: 0.2909, Time: 982.05s |
|
|
2025-07-10 16:13:20,264 - INFO - New best similarity gap: 0.2909 |
|
|
2025-07-10 16:14:35,436 - INFO - log_σ² gradient: -0.568017 |
|
|
2025-07-10 16:14:35,502 - INFO - Optimizer step 1: log_σ²=0.009098, weight=0.990943 |
|
|
2025-07-10 16:14:58,405 - INFO - log_σ² gradient: -0.566842 |
|
|
2025-07-10 16:14:58,475 - INFO - Optimizer step 2: log_σ²=0.009213, weight=0.990829 |
|
|
2025-07-10 16:15:22,146 - INFO - log_σ² gradient: -0.559937 |
|
|
2025-07-10 16:15:22,221 - INFO - Optimizer step 3: log_σ²=0.009330, weight=0.990714 |
|
|
2025-07-10 16:15:46,716 - INFO - log_σ² gradient: -0.560017 |
|
|
2025-07-10 16:15:46,792 - INFO - Optimizer step 4: log_σ²=0.009448, weight=0.990597 |
|
|
2025-07-10 16:16:11,065 - INFO - log_σ² gradient: -0.550409 |
|
|
2025-07-10 16:16:11,138 - INFO - Optimizer step 5: log_σ²=0.009567, weight=0.990479 |
|
|
2025-07-10 16:16:34,886 - INFO - log_σ² gradient: -0.557680 |
|
|
2025-07-10 16:16:34,958 - INFO - Optimizer step 6: log_σ²=0.009687, weight=0.990360 |
|
|
2025-07-10 16:16:59,261 - INFO - log_σ² gradient: -0.564544 |
|
|
2025-07-10 16:16:59,340 - INFO - Optimizer step 7: log_σ²=0.009808, weight=0.990240 |
|
|
2025-07-10 16:17:24,180 - INFO - log_σ² gradient: -0.555164 |
|
|
2025-07-10 16:17:24,251 - INFO - Optimizer step 8: log_σ²=0.009931, weight=0.990118 |
|
|
2025-07-10 16:17:51,754 - INFO - log_σ² gradient: -0.558366 |
|
|
2025-07-10 16:17:51,836 - INFO - Optimizer step 9: log_σ²=0.010055, weight=0.989996 |
|
|
2025-07-10 16:18:16,000 - INFO - log_σ² gradient: -0.558722 |
|
|
2025-07-10 16:18:16,072 - INFO - Optimizer step 10: log_σ²=0.010179, weight=0.989872 |
|
|
2025-07-10 16:18:39,749 - INFO - log_σ² gradient: -0.559033 |
|
|
2025-07-10 16:18:39,831 - INFO - Optimizer step 11: log_σ²=0.010305, weight=0.989748 |
|
|
2025-07-10 16:19:03,405 - INFO - log_σ² gradient: -0.557560 |
|
|
2025-07-10 16:19:03,476 - INFO - Optimizer step 12: log_σ²=0.010432, weight=0.989622 |
|
|
2025-07-10 16:19:27,427 - INFO - log_σ² gradient: -0.561552 |
|
|
2025-07-10 16:19:27,506 - INFO - Optimizer step 13: log_σ²=0.010560, weight=0.989496 |
|
|
2025-07-10 16:19:51,122 - INFO - log_σ² gradient: -0.553774 |
|
|
2025-07-10 16:19:51,194 - INFO - Optimizer step 14: log_σ²=0.010689, weight=0.989368 |
|
|
2025-07-10 16:20:16,341 - INFO - log_σ² gradient: -0.553102 |
|
|
2025-07-10 16:20:16,413 - INFO - Optimizer step 15: log_σ²=0.010818, weight=0.989240 |
|
|
2025-07-10 16:20:39,785 - INFO - log_σ² gradient: -0.565564 |
|
|
2025-07-10 16:20:39,857 - INFO - Optimizer step 16: log_σ²=0.010949, weight=0.989111 |
|
|
2025-07-10 16:21:03,546 - INFO - log_σ² gradient: -0.559742 |
|
|
2025-07-10 16:21:03,626 - INFO - Optimizer step 17: log_σ²=0.011081, weight=0.988981 |
|
|
2025-07-10 16:21:27,397 - INFO - log_σ² gradient: -0.545218 |
|
|
2025-07-10 16:21:27,475 - INFO - Optimizer step 18: log_σ²=0.011213, weight=0.988850 |
|
|
2025-07-10 16:21:52,589 - INFO - log_σ² gradient: -0.553981 |
|
|
2025-07-10 16:21:52,667 - INFO - Optimizer step 19: log_σ²=0.011346, weight=0.988718 |
|
|
2025-07-10 16:22:15,783 - INFO - log_σ² gradient: -0.549238 |
|
|
2025-07-10 16:22:15,855 - INFO - Optimizer step 20: log_σ²=0.011480, weight=0.988586 |
|
|
2025-07-10 16:22:39,039 - INFO - log_σ² gradient: -0.551211 |
|
|
2025-07-10 16:22:39,114 - INFO - Optimizer step 21: log_σ²=0.011615, weight=0.988453 |
|
|
2025-07-10 16:23:04,616 - INFO - log_σ² gradient: -0.567630 |
|
|
2025-07-10 16:23:04,687 - INFO - Optimizer step 22: log_σ²=0.011750, weight=0.988318 |
|
|
2025-07-10 16:23:29,274 - INFO - log_σ² gradient: -0.552309 |
|
|
2025-07-10 16:23:29,346 - INFO - Optimizer step 23: log_σ²=0.011887, weight=0.988183 |
|
|
2025-07-10 16:23:53,292 - INFO - log_σ² gradient: -0.552122 |
|
|
2025-07-10 16:23:53,366 - INFO - Optimizer step 24: log_σ²=0.012024, weight=0.988048 |
|
|
2025-07-10 16:24:16,599 - INFO - log_σ² gradient: -0.549074 |
|
|
2025-07-10 16:24:16,671 - INFO - Optimizer step 25: log_σ²=0.012162, weight=0.987911 |
|
|
2025-07-10 16:24:41,921 - INFO - log_σ² gradient: -0.547727 |
|
|
2025-07-10 16:24:42,000 - INFO - Optimizer step 26: log_σ²=0.012301, weight=0.987774 |
|
|
2025-07-10 16:25:06,168 - INFO - log_σ² gradient: -0.555164 |
|
|
2025-07-10 16:25:06,239 - INFO - Optimizer step 27: log_σ²=0.012441, weight=0.987636 |
|
|
2025-07-10 16:25:29,957 - INFO - log_σ² gradient: -0.566764 |
|
|
2025-07-10 16:25:30,031 - INFO - Optimizer step 28: log_σ²=0.012581, weight=0.987497 |
|
|
2025-07-10 16:25:54,327 - INFO - log_σ² gradient: -0.556580 |
|
|
2025-07-10 16:25:54,401 - INFO - Optimizer step 29: log_σ²=0.012723, weight=0.987358 |
|
|
2025-07-10 16:26:17,781 - INFO - log_σ² gradient: -0.560674 |
|
|
2025-07-10 16:26:17,847 - INFO - Optimizer step 30: log_σ²=0.012866, weight=0.987217 |
|
|
2025-07-10 16:26:29,105 - INFO - log_σ² gradient: -0.258570 |
|
|
2025-07-10 16:26:29,177 - INFO - Optimizer step 31: log_σ²=0.013002, weight=0.987083 |
|
|
2025-07-10 16:26:29,398 - INFO - Epoch 6: Total optimizer steps: 31 |
|
|
2025-07-10 16:29:45,006 - INFO - Validation metrics: |
|
|
2025-07-10 16:29:45,007 - INFO - Loss: 0.7663 |
|
|
2025-07-10 16:29:45,007 - INFO - Average similarity: 0.6200 |
|
|
2025-07-10 16:29:45,007 - INFO - Median similarity: 0.8402 |
|
|
2025-07-10 16:29:45,007 - INFO - Clean sample similarity: 0.6200 |
|
|
2025-07-10 16:29:45,007 - INFO - Corrupted sample similarity: 0.3285 |
|
|
2025-07-10 16:29:45,007 - INFO - Similarity gap (clean - corrupt): 0.2915 |
|
|
2025-07-10 16:29:45,111 - INFO - Epoch 6/30 - Train Loss: 0.8161, Val Loss: 0.7663, Clean Sim: 0.6200, Corrupt Sim: 0.3285, Gap: 0.2915, Time: 978.76s |
|
|
2025-07-10 16:29:45,111 - INFO - New best validation loss: 0.7663 |
|
|
2025-07-10 16:29:51,216 - INFO - New best similarity gap: 0.2915 |
|
|
2025-07-10 16:32:38,608 - INFO - Epoch 6 Validation Alignment: Pos=0.121, Neg=0.100, Gap=0.021 |
|
|
2025-07-10 16:33:45,367 - INFO - log_σ² gradient: -0.548974 |
|
|
2025-07-10 16:33:45,438 - INFO - Optimizer step 1: log_σ²=0.013139, weight=0.986947 |
|
|
2025-07-10 16:34:08,560 - INFO - log_σ² gradient: -0.540829 |
|
|
2025-07-10 16:34:08,634 - INFO - Optimizer step 2: log_σ²=0.013277, weight=0.986810 |
|
|
2025-07-10 16:34:32,645 - INFO - log_σ² gradient: -0.551093 |
|
|
2025-07-10 16:34:32,723 - INFO - Optimizer step 3: log_σ²=0.013417, weight=0.986672 |
|
|
2025-07-10 16:34:55,589 - INFO - log_σ² gradient: -0.557224 |
|
|
2025-07-10 16:34:55,663 - INFO - Optimizer step 4: log_σ²=0.013559, weight=0.986533 |
|
|
2025-07-10 16:35:19,933 - INFO - log_σ² gradient: -0.548364 |
|
|
2025-07-10 16:35:20,009 - INFO - Optimizer step 5: log_σ²=0.013701, weight=0.986392 |
|
|
2025-07-10 16:35:44,443 - INFO - log_σ² gradient: -0.542394 |
|
|
2025-07-10 16:35:44,505 - INFO - Optimizer step 6: log_σ²=0.013844, weight=0.986251 |
|
|
2025-07-10 16:36:09,619 - INFO - log_σ² gradient: -0.547385 |
|
|
2025-07-10 16:36:09,694 - INFO - Optimizer step 7: log_σ²=0.013989, weight=0.986108 |
|
|
2025-07-10 16:36:36,044 - INFO - log_σ² gradient: -0.560439 |
|
|
2025-07-10 16:36:36,117 - INFO - Optimizer step 8: log_σ²=0.014135, weight=0.985965 |
|
|
2025-07-10 16:37:00,513 - INFO - log_σ² gradient: -0.546296 |
|
|
2025-07-10 16:37:00,592 - INFO - Optimizer step 9: log_σ²=0.014282, weight=0.985820 |
|
|
2025-07-10 16:37:25,935 - INFO - log_σ² gradient: -0.554143 |
|
|
2025-07-10 16:37:26,011 - INFO - Optimizer step 10: log_σ²=0.014430, weight=0.985674 |
|
|
2025-07-10 16:37:50,437 - INFO - log_σ² gradient: -0.540571 |
|
|
2025-07-10 16:37:50,506 - INFO - Optimizer step 11: log_σ²=0.014579, weight=0.985527 |
|
|
2025-07-10 16:38:13,937 - INFO - log_σ² gradient: -0.552149 |
|
|
2025-07-10 16:38:14,004 - INFO - Optimizer step 12: log_σ²=0.014729, weight=0.985379 |
|
|
2025-07-10 16:38:37,243 - INFO - log_σ² gradient: -0.551458 |
|
|
2025-07-10 16:38:37,319 - INFO - Optimizer step 13: log_σ²=0.014880, weight=0.985231 |
|
|
2025-07-10 16:39:01,258 - INFO - log_σ² gradient: -0.531494 |
|
|
2025-07-10 16:39:01,325 - INFO - Optimizer step 14: log_σ²=0.015031, weight=0.985081 |
|
|
2025-07-10 16:39:26,230 - INFO - log_σ² gradient: -0.553492 |
|
|
2025-07-10 16:39:26,309 - INFO - Optimizer step 15: log_σ²=0.015184, weight=0.984931 |
|
|
2025-07-10 16:39:50,739 - INFO - log_σ² gradient: -0.544295 |
|
|
2025-07-10 16:39:50,817 - INFO - Optimizer step 16: log_σ²=0.015337, weight=0.984780 |
|
|
2025-07-10 16:40:16,063 - INFO - log_σ² gradient: -0.565899 |
|
|
2025-07-10 16:40:16,134 - INFO - Optimizer step 17: log_σ²=0.015492, weight=0.984628 |
|
|
2025-07-10 16:40:39,321 - INFO - log_σ² gradient: -0.550046 |
|
|
2025-07-10 16:40:39,389 - INFO - Optimizer step 18: log_σ²=0.015647, weight=0.984474 |
|
|
2025-07-10 16:41:05,229 - INFO - log_σ² gradient: -0.552493 |
|
|
2025-07-10 16:41:05,307 - INFO - Optimizer step 19: log_σ²=0.015804, weight=0.984320 |
|
|
2025-07-10 16:41:30,512 - INFO - log_σ² gradient: -0.551766 |
|
|
2025-07-10 16:41:30,590 - INFO - Optimizer step 20: log_σ²=0.015962, weight=0.984165 |
|
|
2025-07-10 16:41:54,817 - INFO - log_σ² gradient: -0.550463 |
|
|
2025-07-10 16:41:54,896 - INFO - Optimizer step 21: log_σ²=0.016120, weight=0.984009 |
|
|
2025-07-10 16:42:18,986 - INFO - log_σ² gradient: -0.540397 |
|
|
2025-07-10 16:42:19,062 - INFO - Optimizer step 22: log_σ²=0.016279, weight=0.983853 |
|
|
2025-07-10 16:42:42,604 - INFO - log_σ² gradient: -0.550544 |
|
|
2025-07-10 16:42:42,672 - INFO - Optimizer step 23: log_σ²=0.016439, weight=0.983695 |
|
|
2025-07-10 16:43:07,275 - INFO - log_σ² gradient: -0.552073 |
|
|
2025-07-10 16:43:07,346 - INFO - Optimizer step 24: log_σ²=0.016600, weight=0.983537 |
|
|
2025-07-10 16:43:32,080 - INFO - log_σ² gradient: -0.555138 |
|
|
2025-07-10 16:43:32,155 - INFO - Optimizer step 25: log_σ²=0.016762, weight=0.983378 |
|
|
2025-07-10 16:43:57,502 - INFO - log_σ² gradient: -0.546422 |
|
|
2025-07-10 16:43:57,574 - INFO - Optimizer step 26: log_σ²=0.016925, weight=0.983218 |
|
|
2025-07-10 16:44:20,806 - INFO - log_σ² gradient: -0.538194 |
|
|
2025-07-10 16:44:20,880 - INFO - Optimizer step 27: log_σ²=0.017088, weight=0.983057 |
|
|
2025-07-10 16:44:44,469 - INFO - log_σ² gradient: -0.537083 |
|
|
2025-07-10 16:44:44,540 - INFO - Optimizer step 28: log_σ²=0.017252, weight=0.982896 |
|
|
2025-07-10 16:45:08,159 - INFO - log_σ² gradient: -0.552016 |
|
|
2025-07-10 16:45:08,233 - INFO - Optimizer step 29: log_σ²=0.017416, weight=0.982734 |
|
|
2025-07-10 16:45:30,333 - INFO - log_σ² gradient: -0.552085 |
|
|
2025-07-10 16:45:30,405 - INFO - Optimizer step 30: log_σ²=0.017582, weight=0.982572 |
|
|
2025-07-10 16:45:40,991 - INFO - log_σ² gradient: -0.260811 |
|
|
2025-07-10 16:45:41,067 - INFO - Optimizer step 31: log_σ²=0.017740, weight=0.982416 |
|
|
2025-07-10 16:45:41,252 - INFO - Epoch 7: Total optimizer steps: 31 |
|
|
2025-07-10 16:48:59,353 - INFO - Validation metrics: |
|
|
2025-07-10 16:48:59,353 - INFO - Loss: 0.7641 |
|
|
2025-07-10 16:48:59,353 - INFO - Average similarity: 0.9188 |
|
|
2025-07-10 16:48:59,353 - INFO - Median similarity: 0.9961 |
|
|
2025-07-10 16:48:59,353 - INFO - Clean sample similarity: 0.9188 |
|
|
2025-07-10 16:48:59,353 - INFO - Corrupted sample similarity: 0.6100 |
|
|
2025-07-10 16:48:59,353 - INFO - Similarity gap (clean - corrupt): 0.3088 |
|
|
2025-07-10 16:48:59,448 - INFO - Epoch 7/30 - Train Loss: 0.7769, Val Loss: 0.7641, Clean Sim: 0.9188, Corrupt Sim: 0.6100, Gap: 0.3088, Time: 980.84s |
|
|
2025-07-10 16:48:59,448 - INFO - New best validation loss: 0.7641 |
|
|
2025-07-10 16:49:05,528 - INFO - New best similarity gap: 0.3088 |
|
|
2025-07-10 16:50:21,456 - INFO - log_σ² gradient: -0.549724 |
|
|
2025-07-10 16:50:21,535 - INFO - Optimizer step 1: log_σ²=0.017900, weight=0.982259 |
|
|
2025-07-10 16:50:46,322 - INFO - log_σ² gradient: -0.543920 |
|
|
2025-07-10 16:50:46,396 - INFO - Optimizer step 2: log_σ²=0.018061, weight=0.982101 |
|
|
2025-07-10 16:51:10,266 - INFO - log_σ² gradient: -0.528869 |
|
|
2025-07-10 16:51:10,341 - INFO - Optimizer step 3: log_σ²=0.018223, weight=0.981942 |
|
|
2025-07-10 16:51:34,600 - INFO - log_σ² gradient: -0.541582 |
|
|
2025-07-10 16:51:34,666 - INFO - Optimizer step 4: log_σ²=0.018387, weight=0.981781 |
|
|
2025-07-10 16:51:56,899 - INFO - log_σ² gradient: -0.548333 |
|
|
2025-07-10 16:51:56,963 - INFO - Optimizer step 5: log_σ²=0.018551, weight=0.981620 |
|
|
2025-07-10 16:52:20,836 - INFO - log_σ² gradient: -0.561278 |
|
|
2025-07-10 16:52:20,914 - INFO - Optimizer step 6: log_σ²=0.018718, weight=0.981456 |
|
|
2025-07-10 16:52:44,669 - INFO - log_σ² gradient: -0.554863 |
|
|
2025-07-10 16:52:44,741 - INFO - Optimizer step 7: log_σ²=0.018886, weight=0.981291 |
|
|
2025-07-10 16:53:08,890 - INFO - log_σ² gradient: -0.550864 |
|
|
2025-07-10 16:53:08,968 - INFO - Optimizer step 8: log_σ²=0.019056, weight=0.981125 |
|
|
2025-07-10 16:53:33,830 - INFO - log_σ² gradient: -0.557946 |
|
|
2025-07-10 16:53:33,901 - INFO - Optimizer step 9: log_σ²=0.019227, weight=0.980957 |
|
|
2025-07-10 16:53:58,879 - INFO - log_σ² gradient: -0.558256 |
|
|
2025-07-10 16:53:58,957 - INFO - Optimizer step 10: log_σ²=0.019399, weight=0.980788 |
|
|
2025-07-10 16:54:23,032 - INFO - log_σ² gradient: -0.540979 |
|
|
2025-07-10 16:54:23,105 - INFO - Optimizer step 11: log_σ²=0.019572, weight=0.980618 |
|
|
2025-07-10 16:54:46,453 - INFO - log_σ² gradient: -0.546736 |
|
|
2025-07-10 16:54:46,525 - INFO - Optimizer step 12: log_σ²=0.019746, weight=0.980447 |
|
|
2025-07-10 16:55:11,138 - INFO - log_σ² gradient: -0.568650 |
|
|
2025-07-10 16:55:11,209 - INFO - Optimizer step 13: log_σ²=0.019922, weight=0.980275 |
|
|
2025-07-10 16:55:35,820 - INFO - log_σ² gradient: -0.543600 |
|
|
2025-07-10 16:55:35,891 - INFO - Optimizer step 14: log_σ²=0.020099, weight=0.980102 |
|
|
2025-07-10 16:55:59,603 - INFO - log_σ² gradient: -0.563959 |
|
|
2025-07-10 16:55:59,677 - INFO - Optimizer step 15: log_σ²=0.020277, weight=0.979927 |
|
|
2025-07-10 16:56:23,570 - INFO - log_σ² gradient: -0.547597 |
|
|
2025-07-10 16:56:23,649 - INFO - Optimizer step 16: log_σ²=0.020456, weight=0.979752 |
|
|
2025-07-10 16:56:48,213 - INFO - log_σ² gradient: -0.549804 |
|
|
2025-07-10 16:56:48,289 - INFO - Optimizer step 17: log_σ²=0.020636, weight=0.979576 |
|
|
2025-07-10 16:57:13,413 - INFO - log_σ² gradient: -0.549053 |
|
|
2025-07-10 16:57:13,486 - INFO - Optimizer step 18: log_σ²=0.020817, weight=0.979398 |
|
|
2025-07-10 16:57:37,625 - INFO - log_σ² gradient: -0.550109 |
|
|
2025-07-10 16:57:37,696 - INFO - Optimizer step 19: log_σ²=0.020998, weight=0.979221 |
|
|
2025-07-10 16:58:02,492 - INFO - log_σ² gradient: -0.536926 |
|
|
2025-07-10 16:58:02,560 - INFO - Optimizer step 20: log_σ²=0.021181, weight=0.979042 |
|
|
2025-07-10 16:58:26,399 - INFO - log_σ² gradient: -0.555423 |
|
|
2025-07-10 16:58:26,477 - INFO - Optimizer step 21: log_σ²=0.021364, weight=0.978863 |
|
|
2025-07-10 16:58:49,819 - INFO - log_σ² gradient: -0.543745 |
|
|
2025-07-10 16:58:49,895 - INFO - Optimizer step 22: log_σ²=0.021548, weight=0.978683 |
|
|
2025-07-10 16:59:14,492 - INFO - log_σ² gradient: -0.545537 |
|
|
2025-07-10 16:59:14,565 - INFO - Optimizer step 23: log_σ²=0.021733, weight=0.978502 |
|
|
2025-07-10 16:59:38,095 - INFO - log_σ² gradient: -0.540812 |
|
|
2025-07-10 16:59:38,174 - INFO - Optimizer step 24: log_σ²=0.021918, weight=0.978320 |
|
|
2025-07-10 17:00:01,576 - INFO - log_σ² gradient: -0.541775 |
|
|
2025-07-10 17:00:01,647 - INFO - Optimizer step 25: log_σ²=0.022104, weight=0.978138 |
|
|
2025-07-10 17:00:28,329 - INFO - log_σ² gradient: -0.546449 |
|
|
2025-07-10 17:00:28,400 - INFO - Optimizer step 26: log_σ²=0.022291, weight=0.977956 |
|
|
2025-07-10 17:00:52,157 - INFO - log_σ² gradient: -0.541647 |
|
|
2025-07-10 17:00:52,236 - INFO - Optimizer step 27: log_σ²=0.022479, weight=0.977772 |
|
|
2025-07-10 17:01:16,625 - INFO - log_σ² gradient: -0.546290 |
|
|
2025-07-10 17:01:16,700 - INFO - Optimizer step 28: log_σ²=0.022667, weight=0.977588 |
|
|
2025-07-10 17:01:41,662 - INFO - log_σ² gradient: -0.536283 |
|
|
2025-07-10 17:01:41,741 - INFO - Optimizer step 29: log_σ²=0.022856, weight=0.977403 |
|
|
2025-07-10 17:02:05,734 - INFO - log_σ² gradient: -0.544419 |
|
|
2025-07-10 17:02:05,813 - INFO - Optimizer step 30: log_σ²=0.023046, weight=0.977218 |
|
|
2025-07-10 17:02:16,351 - INFO - log_σ² gradient: -0.261000 |
|
|
2025-07-10 17:02:16,425 - INFO - Optimizer step 31: log_σ²=0.023227, weight=0.977041 |
|
|
2025-07-10 17:02:16,612 - INFO - Epoch 8: Total optimizer steps: 31 |
|
|
2025-07-10 17:05:38,203 - INFO - Validation metrics: |
|
|
2025-07-10 17:05:38,203 - INFO - Loss: 0.7437 |
|
|
2025-07-10 17:05:38,203 - INFO - Average similarity: 0.6682 |
|
|
2025-07-10 17:05:38,203 - INFO - Median similarity: 0.8904 |
|
|
2025-07-10 17:05:38,203 - INFO - Clean sample similarity: 0.6682 |
|
|
2025-07-10 17:05:38,203 - INFO - Corrupted sample similarity: 0.3007 |
|
|
2025-07-10 17:05:38,203 - INFO - Similarity gap (clean - corrupt): 0.3675 |
|
|
2025-07-10 17:05:38,303 - INFO - Epoch 8/30 - Train Loss: 0.7799, Val Loss: 0.7437, Clean Sim: 0.6682, Corrupt Sim: 0.3007, Gap: 0.3675, Time: 985.85s |
|
|
2025-07-10 17:05:38,304 - INFO - New best validation loss: 0.7437 |
|
|
2025-07-10 17:05:44,522 - INFO - New best similarity gap: 0.3675 |
|
|
2025-07-10 17:08:31,806 - INFO - Epoch 8 Validation Alignment: Pos=0.101, Neg=0.074, Gap=0.027 |
|
|
2025-07-10 17:09:44,564 - INFO - log_σ² gradient: -0.536522 |
|
|
2025-07-10 17:09:44,636 - INFO - Optimizer step 1: log_σ²=0.023409, weight=0.976863 |
|
|
2025-07-10 17:10:08,077 - INFO - log_σ² gradient: -0.543526 |
|
|
2025-07-10 17:10:08,156 - INFO - Optimizer step 2: log_σ²=0.023593, weight=0.976683 |
|
|
2025-07-10 17:10:30,839 - INFO - log_σ² gradient: -0.533749 |
|
|
2025-07-10 17:10:30,918 - INFO - Optimizer step 3: log_σ²=0.023778, weight=0.976502 |
|
|
2025-07-10 17:10:55,908 - INFO - log_σ² gradient: -0.538008 |
|
|
2025-07-10 17:10:55,980 - INFO - Optimizer step 4: log_σ²=0.023965, weight=0.976320 |
|
|
2025-07-10 17:11:20,765 - INFO - log_σ² gradient: -0.544801 |
|
|
2025-07-10 17:11:20,829 - INFO - Optimizer step 5: log_σ²=0.024153, weight=0.976136 |
|
|
2025-07-10 17:11:44,928 - INFO - log_σ² gradient: -0.549975 |
|
|
2025-07-10 17:11:45,010 - INFO - Optimizer step 6: log_σ²=0.024343, weight=0.975951 |
|
|
2025-07-10 17:12:08,952 - INFO - log_σ² gradient: -0.542045 |
|
|
2025-07-10 17:12:09,031 - INFO - Optimizer step 7: log_σ²=0.024534, weight=0.975765 |
|
|
2025-07-10 17:12:32,938 - INFO - log_σ² gradient: -0.549432 |
|
|
2025-07-10 17:12:33,002 - INFO - Optimizer step 8: log_σ²=0.024727, weight=0.975577 |
|
|
2025-07-10 17:12:57,015 - INFO - log_σ² gradient: -0.545405 |
|
|
2025-07-10 17:12:57,088 - INFO - Optimizer step 9: log_σ²=0.024920, weight=0.975388 |
|
|
2025-07-10 17:13:20,788 - INFO - log_σ² gradient: -0.535540 |
|
|
2025-07-10 17:13:20,867 - INFO - Optimizer step 10: log_σ²=0.025115, weight=0.975198 |
|
|
2025-07-10 17:13:45,802 - INFO - log_σ² gradient: -0.548867 |
|
|
2025-07-10 17:13:45,866 - INFO - Optimizer step 11: log_σ²=0.025311, weight=0.975006 |
|
|
2025-07-10 17:14:11,135 - INFO - log_σ² gradient: -0.554270 |
|
|
2025-07-10 17:14:11,212 - INFO - Optimizer step 12: log_σ²=0.025509, weight=0.974814 |
|
|
2025-07-10 17:14:37,615 - INFO - log_σ² gradient: -0.547245 |
|
|
2025-07-10 17:14:37,690 - INFO - Optimizer step 13: log_σ²=0.025708, weight=0.974620 |
|
|
2025-07-10 17:15:00,765 - INFO - log_σ² gradient: -0.542397 |
|
|
2025-07-10 17:15:00,839 - INFO - Optimizer step 14: log_σ²=0.025908, weight=0.974425 |
|
|
2025-07-10 17:15:24,185 - INFO - log_σ² gradient: -0.541244 |
|
|
2025-07-10 17:15:24,253 - INFO - Optimizer step 15: log_σ²=0.026108, weight=0.974230 |
|
|
2025-07-10 17:15:49,143 - INFO - log_σ² gradient: -0.537447 |
|
|
2025-07-10 17:15:49,215 - INFO - Optimizer step 16: log_σ²=0.026310, weight=0.974033 |
|
|
2025-07-10 17:16:12,167 - INFO - log_σ² gradient: -0.542615 |
|
|
2025-07-10 17:16:12,239 - INFO - Optimizer step 17: log_σ²=0.026512, weight=0.973836 |
|
|
2025-07-10 17:16:36,486 - INFO - log_σ² gradient: -0.548041 |
|
|
2025-07-10 17:16:36,564 - INFO - Optimizer step 18: log_σ²=0.026716, weight=0.973638 |
|
|
2025-07-10 17:17:00,170 - INFO - log_σ² gradient: -0.561346 |
|
|
2025-07-10 17:17:00,245 - INFO - Optimizer step 19: log_σ²=0.026921, weight=0.973438 |
|
|
2025-07-10 17:17:24,951 - INFO - log_σ² gradient: -0.537391 |
|
|
2025-07-10 17:17:25,037 - INFO - Optimizer step 20: log_σ²=0.027126, weight=0.973238 |
|
|
2025-07-10 17:17:50,875 - INFO - log_σ² gradient: -0.546383 |
|
|
2025-07-10 17:17:50,946 - INFO - Optimizer step 21: log_σ²=0.027333, weight=0.973037 |
|
|
2025-07-10 17:18:15,318 - INFO - log_σ² gradient: -0.537343 |
|
|
2025-07-10 17:18:15,384 - INFO - Optimizer step 22: log_σ²=0.027541, weight=0.972835 |
|
|
2025-07-10 17:18:38,487 - INFO - log_σ² gradient: -0.534339 |
|
|
2025-07-10 17:18:38,561 - INFO - Optimizer step 23: log_σ²=0.027749, weight=0.972633 |
|
|
2025-07-10 17:19:02,031 - INFO - log_σ² gradient: -0.541069 |
|
|
2025-07-10 17:19:02,103 - INFO - Optimizer step 24: log_σ²=0.027957, weight=0.972430 |
|
|
2025-07-10 17:19:26,465 - INFO - log_σ² gradient: -0.543344 |
|
|
2025-07-10 17:19:26,537 - INFO - Optimizer step 25: log_σ²=0.028167, weight=0.972226 |
|
|
2025-07-10 17:19:50,981 - INFO - log_σ² gradient: -0.553222 |
|
|
2025-07-10 17:19:51,049 - INFO - Optimizer step 26: log_σ²=0.028378, weight=0.972021 |
|
|
2025-07-10 17:20:15,728 - INFO - log_σ² gradient: -0.528853 |
|
|
2025-07-10 17:20:15,802 - INFO - Optimizer step 27: log_σ²=0.028589, weight=0.971816 |
|
|
2025-07-10 17:20:41,130 - INFO - log_σ² gradient: -0.542126 |
|
|
2025-07-10 17:20:41,201 - INFO - Optimizer step 28: log_σ²=0.028801, weight=0.971610 |
|
|
2025-07-10 17:21:03,362 - INFO - log_σ² gradient: -0.536852 |
|
|
2025-07-10 17:21:03,431 - INFO - Optimizer step 29: log_σ²=0.029014, weight=0.971403 |
|
|
2025-07-10 17:21:26,771 - INFO - log_σ² gradient: -0.538833 |
|
|
2025-07-10 17:21:26,843 - INFO - Optimizer step 30: log_σ²=0.029227, weight=0.971196 |
|
|
2025-07-10 17:21:36,675 - INFO - log_σ² gradient: -0.254859 |
|
|
2025-07-10 17:21:36,741 - INFO - Optimizer step 31: log_σ²=0.029430, weight=0.970999 |
|
|
2025-07-10 17:21:36,922 - INFO - Epoch 9: Total optimizer steps: 31 |
|
|
2025-07-10 17:24:56,190 - INFO - Validation metrics: |
|
|
2025-07-10 17:24:56,191 - INFO - Loss: 0.7171 |
|
|
2025-07-10 17:24:56,191 - INFO - Average similarity: 0.7884 |
|
|
2025-07-10 17:24:56,191 - INFO - Median similarity: 0.9629 |
|
|
2025-07-10 17:24:56,191 - INFO - Clean sample similarity: 0.7884 |
|
|
2025-07-10 17:24:56,191 - INFO - Corrupted sample similarity: 0.3944 |
|
|
2025-07-10 17:24:56,191 - INFO - Similarity gap (clean - corrupt): 0.3940 |
|
|
2025-07-10 17:24:56,307 - INFO - Epoch 9/30 - Train Loss: 0.7575, Val Loss: 0.7171, Clean Sim: 0.7884, Corrupt Sim: 0.3944, Gap: 0.3940, Time: 984.50s |
|
|
2025-07-10 17:24:56,307 - INFO - New best validation loss: 0.7171 |
|
|
2025-07-10 17:25:02,421 - INFO - New best similarity gap: 0.3940 |
|
|
2025-07-10 17:26:16,635 - INFO - log_σ² gradient: -0.539628 |
|
|
2025-07-10 17:26:16,707 - INFO - Optimizer step 1: log_σ²=0.029635, weight=0.970800 |
|
|
2025-07-10 17:26:41,090 - INFO - log_σ² gradient: -0.538785 |
|
|
2025-07-10 17:26:41,162 - INFO - Optimizer step 2: log_σ²=0.029842, weight=0.970599 |
|
|
2025-07-10 17:27:06,680 - INFO - log_σ² gradient: -0.538639 |
|
|
2025-07-10 17:27:06,754 - INFO - Optimizer step 3: log_σ²=0.030050, weight=0.970397 |
|
|
2025-07-10 17:27:30,959 - INFO - log_σ² gradient: -0.538436 |
|
|
2025-07-10 17:27:31,030 - INFO - Optimizer step 4: log_σ²=0.030260, weight=0.970193 |
|
|
2025-07-10 17:27:54,271 - INFO - log_σ² gradient: -0.550471 |
|
|
2025-07-10 17:27:54,343 - INFO - Optimizer step 5: log_σ²=0.030472, weight=0.969987 |
|
|
2025-07-10 17:28:17,367 - INFO - log_σ² gradient: -0.532091 |
|
|
2025-07-10 17:28:17,443 - INFO - Optimizer step 6: log_σ²=0.030685, weight=0.969781 |
|
|
2025-07-10 17:28:40,709 - INFO - log_σ² gradient: -0.540206 |
|
|
2025-07-10 17:28:40,774 - INFO - Optimizer step 7: log_σ²=0.030900, weight=0.969573 |
|
|
2025-07-10 17:29:04,903 - INFO - log_σ² gradient: -0.543870 |
|
|
2025-07-10 17:29:04,974 - INFO - Optimizer step 8: log_σ²=0.031116, weight=0.969364 |
|
|
2025-07-10 17:29:29,489 - INFO - log_σ² gradient: -0.543883 |
|
|
2025-07-10 17:29:29,560 - INFO - Optimizer step 9: log_σ²=0.031333, weight=0.969153 |
|
|
2025-07-10 17:29:53,667 - INFO - log_σ² gradient: -0.543740 |
|
|
2025-07-10 17:29:53,747 - INFO - Optimizer step 10: log_σ²=0.031552, weight=0.968941 |
|
|
2025-07-10 17:30:17,193 - INFO - log_σ² gradient: -0.538837 |
|
|
2025-07-10 17:30:17,269 - INFO - Optimizer step 11: log_σ²=0.031771, weight=0.968728 |
|
|
2025-07-10 17:30:40,733 - INFO - log_σ² gradient: -0.547343 |
|
|
2025-07-10 17:30:40,804 - INFO - Optimizer step 12: log_σ²=0.031993, weight=0.968514 |
|
|
2025-07-10 17:31:03,976 - INFO - log_σ² gradient: -0.538469 |
|
|
2025-07-10 17:31:04,048 - INFO - Optimizer step 13: log_σ²=0.032215, weight=0.968299 |
|
|
2025-07-10 17:31:26,564 - INFO - log_σ² gradient: -0.531641 |
|
|
2025-07-10 17:31:26,630 - INFO - Optimizer step 14: log_σ²=0.032438, weight=0.968083 |
|
|
2025-07-10 17:31:50,083 - INFO - log_σ² gradient: -0.535655 |
|
|
2025-07-10 17:31:50,154 - INFO - Optimizer step 15: log_σ²=0.032662, weight=0.967866 |
|
|
2025-07-10 17:32:13,811 - INFO - log_σ² gradient: -0.542713 |
|
|
2025-07-10 17:32:13,875 - INFO - Optimizer step 16: log_σ²=0.032886, weight=0.967648 |
|
|
2025-07-10 17:32:40,813 - INFO - log_σ² gradient: -0.535118 |
|
|
2025-07-10 17:32:40,893 - INFO - Optimizer step 17: log_σ²=0.033112, weight=0.967430 |
|
|
2025-07-10 17:33:04,555 - INFO - log_σ² gradient: -0.545049 |
|
|
2025-07-10 17:33:04,635 - INFO - Optimizer step 18: log_σ²=0.033339, weight=0.967210 |
|
|
2025-07-10 17:33:30,020 - INFO - log_σ² gradient: -0.536629 |
|
|
2025-07-10 17:33:30,091 - INFO - Optimizer step 19: log_σ²=0.033567, weight=0.966990 |
|
|
2025-07-10 17:33:55,474 - INFO - log_σ² gradient: -0.528660 |
|
|
2025-07-10 17:33:55,546 - INFO - Optimizer step 20: log_σ²=0.033795, weight=0.966769 |
|
|
2025-07-10 17:34:20,254 - INFO - log_σ² gradient: -0.527693 |
|
|
2025-07-10 17:34:20,330 - INFO - Optimizer step 21: log_σ²=0.034024, weight=0.966548 |
|
|
2025-07-10 17:34:45,483 - INFO - log_σ² gradient: -0.534387 |
|
|
2025-07-10 17:34:45,559 - INFO - Optimizer step 22: log_σ²=0.034254, weight=0.966326 |
|
|
2025-07-10 17:35:10,531 - INFO - log_σ² gradient: -0.542139 |
|
|
2025-07-10 17:35:10,609 - INFO - Optimizer step 23: log_σ²=0.034485, weight=0.966103 |
|
|
2025-07-10 17:35:34,816 - INFO - log_σ² gradient: -0.543082 |
|
|
2025-07-10 17:35:34,894 - INFO - Optimizer step 24: log_σ²=0.034717, weight=0.965879 |
|
|
2025-07-10 17:35:57,433 - INFO - log_σ² gradient: -0.538778 |
|
|
2025-07-10 17:35:57,500 - INFO - Optimizer step 25: log_σ²=0.034949, weight=0.965654 |
|
|
2025-07-10 17:36:22,872 - INFO - log_σ² gradient: -0.523146 |
|
|
2025-07-10 17:36:22,944 - INFO - Optimizer step 26: log_σ²=0.035182, weight=0.965429 |
|
|
2025-07-10 17:36:47,829 - INFO - log_σ² gradient: -0.539740 |
|
|
2025-07-10 17:36:47,905 - INFO - Optimizer step 27: log_σ²=0.035416, weight=0.965203 |
|
|
2025-07-10 17:37:11,112 - INFO - log_σ² gradient: -0.536670 |
|
|
2025-07-10 17:37:11,184 - INFO - Optimizer step 28: log_σ²=0.035651, weight=0.964977 |
|
|
2025-07-10 17:37:36,293 - INFO - log_σ² gradient: -0.528335 |
|
|
2025-07-10 17:37:36,365 - INFO - Optimizer step 29: log_σ²=0.035887, weight=0.964750 |
|
|
2025-07-10 17:37:58,838 - INFO - log_σ² gradient: -0.532560 |
|
|
2025-07-10 17:37:58,908 - INFO - Optimizer step 30: log_σ²=0.036123, weight=0.964522 |
|
|
2025-07-10 17:38:10,993 - INFO - log_σ² gradient: -0.250389 |
|
|
2025-07-10 17:38:11,060 - INFO - Optimizer step 31: log_σ²=0.036347, weight=0.964305 |
|
|
2025-07-10 17:38:11,216 - INFO - Epoch 10: Total optimizer steps: 31 |
|
|
2025-07-10 17:41:29,454 - INFO - Validation metrics: |
|
|
2025-07-10 17:41:29,454 - INFO - Loss: 0.7129 |
|
|
2025-07-10 17:41:29,454 - INFO - Average similarity: 0.8915 |
|
|
2025-07-10 17:41:29,454 - INFO - Median similarity: 0.9950 |
|
|
2025-07-10 17:41:29,454 - INFO - Clean sample similarity: 0.8915 |
|
|
2025-07-10 17:41:29,454 - INFO - Corrupted sample similarity: 0.5113 |
|
|
2025-07-10 17:41:29,454 - INFO - Similarity gap (clean - corrupt): 0.3802 |
|
|
2025-07-10 17:41:29,561 - INFO - Epoch 10/30 - Train Loss: 0.7421, Val Loss: 0.7129, Clean Sim: 0.8915, Corrupt Sim: 0.5113, Gap: 0.3802, Time: 979.96s |
|
|
2025-07-10 17:41:29,561 - INFO - New best validation loss: 0.7129 |
|
|
2025-07-10 17:44:18,615 - INFO - Epoch 10 Validation Alignment: Pos=0.131, Neg=0.098, Gap=0.033 |
|
|
2025-07-10 17:45:25,597 - INFO - log_σ² gradient: -0.532370 |
|
|
2025-07-10 17:45:25,676 - INFO - Optimizer step 1: log_σ²=0.036574, weight=0.964087 |
|
|
2025-07-10 17:45:46,563 - INFO - log_σ² gradient: -0.536865 |
|
|
2025-07-10 17:45:46,634 - INFO - Optimizer step 2: log_σ²=0.036803, weight=0.963866 |
|
|
2025-07-10 17:46:11,020 - INFO - log_σ² gradient: -0.533114 |
|
|
2025-07-10 17:46:11,092 - INFO - Optimizer step 3: log_σ²=0.037033, weight=0.963645 |
|
|
2025-07-10 17:46:33,921 - INFO - log_σ² gradient: -0.534954 |
|
|
2025-07-10 17:46:33,987 - INFO - Optimizer step 4: log_σ²=0.037265, weight=0.963421 |
|
|
2025-07-10 17:46:58,647 - INFO - log_σ² gradient: -0.542484 |
|
|
2025-07-10 17:46:58,721 - INFO - Optimizer step 5: log_σ²=0.037499, weight=0.963196 |
|
|
2025-07-10 17:47:22,338 - INFO - log_σ² gradient: -0.532388 |
|
|
2025-07-10 17:47:22,414 - INFO - Optimizer step 6: log_σ²=0.037734, weight=0.962969 |
|
|
2025-07-10 17:47:45,982 - INFO - log_σ² gradient: -0.533090 |
|
|
2025-07-10 17:47:46,046 - INFO - Optimizer step 7: log_σ²=0.037971, weight=0.962741 |
|
|
2025-07-10 17:48:09,596 - INFO - log_σ² gradient: -0.530777 |
|
|
2025-07-10 17:48:09,668 - INFO - Optimizer step 8: log_σ²=0.038209, weight=0.962512 |
|
|
2025-07-10 17:48:33,082 - INFO - log_σ² gradient: -0.540906 |
|
|
2025-07-10 17:48:33,161 - INFO - Optimizer step 9: log_σ²=0.038449, weight=0.962281 |
|
|
2025-07-10 17:48:58,081 - INFO - log_σ² gradient: -0.528409 |
|
|
2025-07-10 17:48:58,159 - INFO - Optimizer step 10: log_σ²=0.038689, weight=0.962050 |
|
|
2025-07-10 17:49:23,566 - INFO - log_σ² gradient: -0.526553 |
|
|
2025-07-10 17:49:23,646 - INFO - Optimizer step 11: log_σ²=0.038931, weight=0.961817 |
|
|
2025-07-10 17:49:46,959 - INFO - log_σ² gradient: -0.523638 |
|
|
2025-07-10 17:49:47,030 - INFO - Optimizer step 12: log_σ²=0.039173, weight=0.961584 |
|
|
2025-07-10 17:50:11,777 - INFO - log_σ² gradient: -0.522973 |
|
|
2025-07-10 17:50:11,849 - INFO - Optimizer step 13: log_σ²=0.039416, weight=0.961351 |
|
|
2025-07-10 17:50:35,788 - INFO - log_σ² gradient: -0.515179 |
|
|
2025-07-10 17:50:35,855 - INFO - Optimizer step 14: log_σ²=0.039659, weight=0.961117 |
|
|
2025-07-10 17:51:00,531 - INFO - log_σ² gradient: -0.535419 |
|
|
2025-07-10 17:51:00,606 - INFO - Optimizer step 15: log_σ²=0.039904, weight=0.960882 |
|
|
2025-07-10 17:51:25,037 - INFO - log_σ² gradient: -0.531236 |
|
|
2025-07-10 17:51:25,104 - INFO - Optimizer step 16: log_σ²=0.040150, weight=0.960646 |
|
|
2025-07-10 17:51:47,948 - INFO - log_σ² gradient: -0.528703 |
|
|
2025-07-10 17:51:48,022 - INFO - Optimizer step 17: log_σ²=0.040397, weight=0.960409 |
|
|
2025-07-10 17:52:13,450 - INFO - log_σ² gradient: -0.525497 |
|
|
2025-07-10 17:52:13,525 - INFO - Optimizer step 18: log_σ²=0.040644, weight=0.960171 |
|
|
2025-07-10 17:52:38,621 - INFO - log_σ² gradient: -0.527384 |
|
|
2025-07-10 17:52:38,699 - INFO - Optimizer step 19: log_σ²=0.040893, weight=0.959932 |
|
|
2025-07-10 17:53:04,153 - INFO - log_σ² gradient: -0.529396 |
|
|
2025-07-10 17:53:04,225 - INFO - Optimizer step 20: log_σ²=0.041142, weight=0.959693 |
|
|
2025-07-10 17:53:28,704 - INFO - log_σ² gradient: -0.540948 |
|
|
2025-07-10 17:53:28,778 - INFO - Optimizer step 21: log_σ²=0.041393, weight=0.959452 |
|
|
2025-07-10 17:53:53,551 - INFO - log_σ² gradient: -0.547490 |
|
|
2025-07-10 17:53:53,625 - INFO - Optimizer step 22: log_σ²=0.041645, weight=0.959210 |
|
|
2025-07-10 17:54:17,985 - INFO - log_σ² gradient: -0.539977 |
|
|
2025-07-10 17:54:18,065 - INFO - Optimizer step 23: log_σ²=0.041899, weight=0.958966 |
|
|
2025-07-10 17:54:42,793 - INFO - log_σ² gradient: -0.532642 |
|
|
2025-07-10 17:54:42,867 - INFO - Optimizer step 24: log_σ²=0.042154, weight=0.958722 |
|
|
2025-07-10 17:55:06,844 - INFO - log_σ² gradient: -0.531817 |
|
|
2025-07-10 17:55:06,920 - INFO - Optimizer step 25: log_σ²=0.042410, weight=0.958477 |
|
|
2025-07-10 17:55:31,417 - INFO - log_σ² gradient: -0.531791 |
|
|
2025-07-10 17:55:31,490 - INFO - Optimizer step 26: log_σ²=0.042666, weight=0.958231 |
|
|
2025-07-10 17:55:54,796 - INFO - log_σ² gradient: -0.536748 |
|
|
2025-07-10 17:55:54,869 - INFO - Optimizer step 27: log_σ²=0.042924, weight=0.957984 |
|
|
2025-07-10 17:56:19,380 - INFO - log_σ² gradient: -0.529381 |
|
|
2025-07-10 17:56:19,458 - INFO - Optimizer step 28: log_σ²=0.043182, weight=0.957737 |
|
|
2025-07-10 17:56:42,612 - INFO - log_σ² gradient: -0.521823 |
|
|
2025-07-10 17:56:42,684 - INFO - Optimizer step 29: log_σ²=0.043441, weight=0.957489 |
|
|
2025-07-10 17:57:04,361 - INFO - log_σ² gradient: -0.532169 |
|
|
2025-07-10 17:57:04,431 - INFO - Optimizer step 30: log_σ²=0.043700, weight=0.957241 |
|
|
2025-07-10 17:57:15,730 - INFO - log_σ² gradient: -0.239791 |
|
|
2025-07-10 17:57:15,803 - INFO - Optimizer step 31: log_σ²=0.043946, weight=0.957005 |
|
|
2025-07-10 17:57:15,992 - INFO - Epoch 11: Total optimizer steps: 31 |
|
|
2025-07-10 18:00:34,595 - INFO - Validation metrics: |
|
|
2025-07-10 18:00:34,595 - INFO - Loss: 0.6939 |
|
|
2025-07-10 18:00:34,595 - INFO - Average similarity: 0.8448 |
|
|
2025-07-10 18:00:34,595 - INFO - Median similarity: 0.9873 |
|
|
2025-07-10 18:00:34,595 - INFO - Clean sample similarity: 0.8448 |
|
|
2025-07-10 18:00:34,595 - INFO - Corrupted sample similarity: 0.4299 |
|
|
2025-07-10 18:00:34,595 - INFO - Similarity gap (clean - corrupt): 0.4149 |
|
|
2025-07-10 18:00:34,708 - INFO - Epoch 11/30 - Train Loss: 0.7224, Val Loss: 0.6939, Clean Sim: 0.8448, Corrupt Sim: 0.4299, Gap: 0.4149, Time: 976.09s |
|
|
2025-07-10 18:00:34,709 - INFO - New best validation loss: 0.6939 |
|
|
2025-07-10 18:00:40,745 - INFO - New best similarity gap: 0.4149 |
|
|
2025-07-10 18:01:55,851 - INFO - log_σ² gradient: -0.519495 |
|
|
2025-07-10 18:01:55,922 - INFO - Optimizer step 1: log_σ²=0.044194, weight=0.956768 |
|
|
2025-07-10 18:02:19,837 - INFO - log_σ² gradient: -0.523688 |
|
|
2025-07-10 18:02:19,908 - INFO - Optimizer step 2: log_σ²=0.044444, weight=0.956529 |
|
|
2025-07-10 18:02:45,202 - INFO - log_σ² gradient: -0.531184 |
|
|
2025-07-10 18:02:45,278 - INFO - Optimizer step 3: log_σ²=0.044696, weight=0.956288 |
|
|
2025-07-10 18:03:10,366 - INFO - log_σ² gradient: -0.525087 |
|
|
2025-07-10 18:03:10,441 - INFO - Optimizer step 4: log_σ²=0.044949, weight=0.956046 |
|
|
2025-07-10 18:03:35,100 - INFO - log_σ² gradient: -0.535427 |
|
|
2025-07-10 18:03:35,176 - INFO - Optimizer step 5: log_σ²=0.045205, weight=0.955802 |
|
|
2025-07-10 18:03:58,890 - INFO - log_σ² gradient: -0.529216 |
|
|
2025-07-10 18:03:58,958 - INFO - Optimizer step 6: log_σ²=0.045462, weight=0.955556 |
|
|
2025-07-10 18:04:24,011 - INFO - log_σ² gradient: -0.527409 |
|
|
2025-07-10 18:04:24,083 - INFO - Optimizer step 7: log_σ²=0.045720, weight=0.955309 |
|
|
2025-07-10 18:04:46,180 - INFO - log_σ² gradient: -0.528864 |
|
|
2025-07-10 18:04:46,244 - INFO - Optimizer step 8: log_σ²=0.045980, weight=0.955061 |
|
|
2025-07-10 18:05:10,567 - INFO - log_σ² gradient: -0.533822 |
|
|
2025-07-10 18:05:10,633 - INFO - Optimizer step 9: log_σ²=0.046242, weight=0.954811 |
|
|
2025-07-10 18:05:33,740 - INFO - log_σ² gradient: -0.534954 |
|
|
2025-07-10 18:05:33,813 - INFO - Optimizer step 10: log_σ²=0.046505, weight=0.954560 |
|
|
2025-07-10 18:05:57,869 - INFO - log_σ² gradient: -0.540507 |
|
|
2025-07-10 18:05:57,943 - INFO - Optimizer step 11: log_σ²=0.046770, weight=0.954307 |
|
|
2025-07-10 18:06:21,418 - INFO - log_σ² gradient: -0.536083 |
|
|
2025-07-10 18:06:21,492 - INFO - Optimizer step 12: log_σ²=0.047037, weight=0.954052 |
|
|
2025-07-10 18:06:46,272 - INFO - log_σ² gradient: -0.523205 |
|
|
2025-07-10 18:06:46,343 - INFO - Optimizer step 13: log_σ²=0.047304, weight=0.953797 |
|
|
2025-07-10 18:07:09,358 - INFO - log_σ² gradient: -0.522129 |
|
|
2025-07-10 18:07:09,420 - INFO - Optimizer step 14: log_σ²=0.047572, weight=0.953541 |
|
|
2025-07-10 18:07:32,801 - INFO - log_σ² gradient: -0.513043 |
|
|
2025-07-10 18:07:32,880 - INFO - Optimizer step 15: log_σ²=0.047841, weight=0.953286 |
|
|
2025-07-10 18:07:58,963 - INFO - log_σ² gradient: -0.527191 |
|
|
2025-07-10 18:07:59,034 - INFO - Optimizer step 16: log_σ²=0.048110, weight=0.953029 |
|
|
2025-07-10 18:08:21,910 - INFO - log_σ² gradient: -0.526414 |
|
|
2025-07-10 18:08:21,977 - INFO - Optimizer step 17: log_σ²=0.048381, weight=0.952771 |
|
|
2025-07-10 18:08:46,244 - INFO - log_σ² gradient: -0.533497 |
|
|
2025-07-10 18:08:46,319 - INFO - Optimizer step 18: log_σ²=0.048652, weight=0.952512 |
|
|
2025-07-10 18:09:09,538 - INFO - log_σ² gradient: -0.530888 |
|
|
2025-07-10 18:09:09,610 - INFO - Optimizer step 19: log_σ²=0.048925, weight=0.952252 |
|
|
2025-07-10 18:09:33,391 - INFO - log_σ² gradient: -0.524023 |
|
|
2025-07-10 18:09:33,462 - INFO - Optimizer step 20: log_σ²=0.049199, weight=0.951992 |
|
|
2025-07-10 18:09:57,827 - INFO - log_σ² gradient: -0.530203 |
|
|
2025-07-10 18:09:57,901 - INFO - Optimizer step 21: log_σ²=0.049474, weight=0.951730 |
|
|
2025-07-10 18:10:22,016 - INFO - log_σ² gradient: -0.524581 |
|
|
2025-07-10 18:10:22,092 - INFO - Optimizer step 22: log_σ²=0.049749, weight=0.951468 |
|
|
2025-07-10 18:10:46,244 - INFO - log_σ² gradient: -0.526881 |
|
|
2025-07-10 18:10:46,319 - INFO - Optimizer step 23: log_σ²=0.050026, weight=0.951205 |
|
|
2025-07-10 18:11:12,108 - INFO - log_σ² gradient: -0.529148 |
|
|
2025-07-10 18:11:12,187 - INFO - Optimizer step 24: log_σ²=0.050303, weight=0.950941 |
|
|
2025-07-10 18:11:37,237 - INFO - log_σ² gradient: -0.520859 |
|
|
2025-07-10 18:11:37,312 - INFO - Optimizer step 25: log_σ²=0.050581, weight=0.950677 |
|
|
2025-07-10 18:12:00,210 - INFO - log_σ² gradient: -0.514845 |
|
|
2025-07-10 18:12:00,278 - INFO - Optimizer step 26: log_σ²=0.050859, weight=0.950413 |
|
|
2025-07-10 18:12:24,695 - INFO - log_σ² gradient: -0.517809 |
|
|
2025-07-10 18:12:24,773 - INFO - Optimizer step 27: log_σ²=0.051138, weight=0.950148 |
|
|
2025-07-10 18:12:50,012 - INFO - log_σ² gradient: -0.525186 |
|
|
2025-07-10 18:12:50,084 - INFO - Optimizer step 28: log_σ²=0.051417, weight=0.949882 |
|
|
2025-07-10 18:13:13,976 - INFO - log_σ² gradient: -0.519054 |
|
|
2025-07-10 18:13:14,054 - INFO - Optimizer step 29: log_σ²=0.051698, weight=0.949616 |
|
|
2025-07-10 18:13:35,945 - INFO - log_σ² gradient: -0.521077 |
|
|
2025-07-10 18:13:36,020 - INFO - Optimizer step 30: log_σ²=0.051978, weight=0.949349 |
|
|
2025-07-10 18:13:46,495 - INFO - log_σ² gradient: -0.242576 |
|
|
2025-07-10 18:13:46,567 - INFO - Optimizer step 31: log_σ²=0.052245, weight=0.949096 |
|
|
2025-07-10 18:13:46,732 - INFO - Epoch 12: Total optimizer steps: 31 |
|
|
2025-07-10 18:17:05,832 - INFO - Validation metrics: |
|
|
2025-07-10 18:17:05,832 - INFO - Loss: 0.6799 |
|
|
2025-07-10 18:17:05,833 - INFO - Average similarity: 0.8511 |
|
|
2025-07-10 18:17:05,833 - INFO - Median similarity: 0.9933 |
|
|
2025-07-10 18:17:05,833 - INFO - Clean sample similarity: 0.8511 |
|
|
2025-07-10 18:17:05,833 - INFO - Corrupted sample similarity: 0.4250 |
|
|
2025-07-10 18:17:05,833 - INFO - Similarity gap (clean - corrupt): 0.4261 |
|
|
2025-07-10 18:17:05,949 - INFO - Epoch 12/30 - Train Loss: 0.7097, Val Loss: 0.6799, Clean Sim: 0.8511, Corrupt Sim: 0.4250, Gap: 0.4261, Time: 977.98s |
|
|
2025-07-10 18:17:05,949 - INFO - New best validation loss: 0.6799 |
|
|
2025-07-10 18:17:12,980 - INFO - New best similarity gap: 0.4261 |
|
|
2025-07-10 18:20:00,387 - INFO - Epoch 12 Validation Alignment: Pos=0.136, Neg=0.089, Gap=0.047 |
|
|
2025-07-10 18:21:06,984 - INFO - log_σ² gradient: -0.523952 |
|
|
2025-07-10 18:21:07,055 - INFO - Optimizer step 1: log_σ²=0.052514, weight=0.948841 |
|
|
2025-07-10 18:21:31,256 - INFO - log_σ² gradient: -0.527925 |
|
|
2025-07-10 18:21:31,335 - INFO - Optimizer step 2: log_σ²=0.052786, weight=0.948583 |
|
|
2025-07-10 18:21:54,234 - INFO - log_σ² gradient: -0.521696 |
|
|
2025-07-10 18:21:54,301 - INFO - Optimizer step 3: log_σ²=0.053059, weight=0.948324 |
|
|
2025-07-10 18:22:17,515 - INFO - log_σ² gradient: -0.505590 |
|
|
2025-07-10 18:22:17,595 - INFO - Optimizer step 4: log_σ²=0.053334, weight=0.948064 |
|
|
2025-07-10 18:22:43,052 - INFO - log_σ² gradient: -0.525409 |
|
|
2025-07-10 18:22:43,128 - INFO - Optimizer step 5: log_σ²=0.053610, weight=0.947802 |
|
|
2025-07-10 18:23:05,925 - INFO - log_σ² gradient: -0.525438 |
|
|
2025-07-10 18:23:06,001 - INFO - Optimizer step 6: log_σ²=0.053888, weight=0.947538 |
|
|
2025-07-10 18:23:30,321 - INFO - log_σ² gradient: -0.517129 |
|
|
2025-07-10 18:23:30,400 - INFO - Optimizer step 7: log_σ²=0.054168, weight=0.947273 |
|
|
2025-07-10 18:23:56,183 - INFO - log_σ² gradient: -0.526283 |
|
|
2025-07-10 18:23:56,258 - INFO - Optimizer step 8: log_σ²=0.054449, weight=0.947007 |
|
|
2025-07-10 18:24:20,687 - INFO - log_σ² gradient: -0.523213 |
|
|
2025-07-10 18:24:20,761 - INFO - Optimizer step 9: log_σ²=0.054732, weight=0.946739 |
|
|
2025-07-10 18:24:43,717 - INFO - log_σ² gradient: -0.525440 |
|
|
2025-07-10 18:24:43,796 - INFO - Optimizer step 10: log_σ²=0.055016, weight=0.946470 |
|
|
2025-07-10 18:25:09,593 - INFO - log_σ² gradient: -0.511373 |
|
|
2025-07-10 18:25:09,664 - INFO - Optimizer step 11: log_σ²=0.055301, weight=0.946200 |
|
|
2025-07-10 18:25:32,037 - INFO - log_σ² gradient: -0.527059 |
|
|
2025-07-10 18:25:32,113 - INFO - Optimizer step 12: log_σ²=0.055588, weight=0.945929 |
|
|
2025-07-10 18:25:57,489 - INFO - log_σ² gradient: -0.525321 |
|
|
2025-07-10 18:25:57,555 - INFO - Optimizer step 13: log_σ²=0.055876, weight=0.945657 |
|
|
2025-07-10 18:26:20,095 - INFO - log_σ² gradient: -0.522609 |
|
|
2025-07-10 18:26:20,159 - INFO - Optimizer step 14: log_σ²=0.056165, weight=0.945383 |
|
|
2025-07-10 18:26:46,145 - INFO - log_σ² gradient: -0.518836 |
|
|
2025-07-10 18:26:46,216 - INFO - Optimizer step 15: log_σ²=0.056455, weight=0.945109 |
|
|
2025-07-10 18:27:10,708 - INFO - log_σ² gradient: -0.519566 |
|
|
2025-07-10 18:27:10,791 - INFO - Optimizer step 16: log_σ²=0.056746, weight=0.944834 |
|
|
2025-07-10 18:27:36,853 - INFO - log_σ² gradient: -0.512649 |
|
|
2025-07-10 18:27:36,929 - INFO - Optimizer step 17: log_σ²=0.057038, weight=0.944558 |
|
|
2025-07-10 18:27:59,842 - INFO - log_σ² gradient: -0.520647 |
|
|
2025-07-10 18:27:59,910 - INFO - Optimizer step 18: log_σ²=0.057331, weight=0.944282 |
|
|
2025-07-10 18:28:26,174 - INFO - log_σ² gradient: -0.523872 |
|
|
2025-07-10 18:28:26,256 - INFO - Optimizer step 19: log_σ²=0.057625, weight=0.944004 |
|
|
2025-07-10 18:28:50,690 - INFO - log_σ² gradient: -0.514081 |
|
|
2025-07-10 18:28:50,766 - INFO - Optimizer step 20: log_σ²=0.057920, weight=0.943726 |
|
|
2025-07-10 18:29:17,179 - INFO - log_σ² gradient: -0.513943 |
|
|
2025-07-10 18:29:17,262 - INFO - Optimizer step 21: log_σ²=0.058215, weight=0.943447 |
|
|
2025-07-10 18:29:41,807 - INFO - log_σ² gradient: -0.518854 |
|
|
2025-07-10 18:29:41,870 - INFO - Optimizer step 22: log_σ²=0.058511, weight=0.943168 |
|
|
2025-07-10 18:30:06,019 - INFO - log_σ² gradient: -0.530385 |
|
|
2025-07-10 18:30:06,095 - INFO - Optimizer step 23: log_σ²=0.058809, weight=0.942887 |
|
|
2025-07-10 18:30:31,631 - INFO - log_σ² gradient: -0.518099 |
|
|
2025-07-10 18:30:31,705 - INFO - Optimizer step 24: log_σ²=0.059107, weight=0.942606 |
|
|
2025-07-10 18:30:53,522 - INFO - log_σ² gradient: -0.524785 |
|
|
2025-07-10 18:30:53,587 - INFO - Optimizer step 25: log_σ²=0.059407, weight=0.942323 |
|
|
2025-07-10 18:31:18,211 - INFO - log_σ² gradient: -0.516661 |
|
|
2025-07-10 18:31:18,289 - INFO - Optimizer step 26: log_σ²=0.059707, weight=0.942040 |
|
|
2025-07-10 18:31:41,792 - INFO - log_σ² gradient: -0.512995 |
|
|
2025-07-10 18:31:41,870 - INFO - Optimizer step 27: log_σ²=0.060008, weight=0.941757 |
|
|
2025-07-10 18:32:03,524 - INFO - log_σ² gradient: -0.518172 |
|
|
2025-07-10 18:32:03,591 - INFO - Optimizer step 28: log_σ²=0.060310, weight=0.941473 |
|
|
2025-07-10 18:32:27,842 - INFO - log_σ² gradient: -0.530053 |
|
|
2025-07-10 18:32:27,909 - INFO - Optimizer step 29: log_σ²=0.060613, weight=0.941187 |
|
|
2025-07-10 18:32:51,204 - INFO - log_σ² gradient: -0.523498 |
|
|
2025-07-10 18:32:51,272 - INFO - Optimizer step 30: log_σ²=0.060917, weight=0.940901 |
|
|
2025-07-10 18:33:03,076 - INFO - log_σ² gradient: -0.233310 |
|
|
2025-07-10 18:33:03,155 - INFO - Optimizer step 31: log_σ²=0.061206, weight=0.940630 |
|
|
2025-07-10 18:33:03,319 - INFO - Epoch 13: Total optimizer steps: 31 |
|
|
2025-07-10 18:36:22,587 - INFO - Validation metrics: |
|
|
2025-07-10 18:36:22,587 - INFO - Loss: 0.6622 |
|
|
2025-07-10 18:36:22,587 - INFO - Average similarity: 0.8385 |
|
|
2025-07-10 18:36:22,587 - INFO - Median similarity: 0.9816 |
|
|
2025-07-10 18:36:22,587 - INFO - Clean sample similarity: 0.8385 |
|
|
2025-07-10 18:36:22,587 - INFO - Corrupted sample similarity: 0.3917 |
|
|
2025-07-10 18:36:22,587 - INFO - Similarity gap (clean - corrupt): 0.4468 |
|
|
2025-07-10 18:36:22,695 - INFO - Epoch 13/30 - Train Loss: 0.7026, Val Loss: 0.6622, Clean Sim: 0.8385, Corrupt Sim: 0.3917, Gap: 0.4468, Time: 982.31s |
|
|
2025-07-10 18:36:22,696 - INFO - New best validation loss: 0.6622 |
|
|
2025-07-10 18:36:28,764 - INFO - New best similarity gap: 0.4468 |
|
|
2025-07-10 18:37:46,275 - INFO - log_σ² gradient: -0.514517 |
|
|
2025-07-10 18:37:46,349 - INFO - Optimizer step 1: log_σ²=0.061496, weight=0.940356 |
|
|
2025-07-10 18:38:09,478 - INFO - log_σ² gradient: -0.515819 |
|
|
2025-07-10 18:38:09,557 - INFO - Optimizer step 2: log_σ²=0.061789, weight=0.940081 |
|
|
2025-07-10 18:38:33,753 - INFO - log_σ² gradient: -0.514219 |
|
|
2025-07-10 18:38:33,831 - INFO - Optimizer step 3: log_σ²=0.062084, weight=0.939804 |
|
|
2025-07-10 18:38:57,097 - INFO - log_σ² gradient: -0.520156 |
|
|
2025-07-10 18:38:57,161 - INFO - Optimizer step 4: log_σ²=0.062380, weight=0.939525 |
|
|
2025-07-10 18:39:21,278 - INFO - log_σ² gradient: -0.518849 |
|
|
2025-07-10 18:39:21,344 - INFO - Optimizer step 5: log_σ²=0.062679, weight=0.939245 |
|
|
2025-07-10 18:39:44,345 - INFO - log_σ² gradient: -0.517957 |
|
|
2025-07-10 18:39:44,417 - INFO - Optimizer step 6: log_σ²=0.062979, weight=0.938963 |
|
|
2025-07-10 18:40:06,573 - INFO - log_σ² gradient: -0.516069 |
|
|
2025-07-10 18:40:06,641 - INFO - Optimizer step 7: log_σ²=0.063281, weight=0.938679 |
|
|
2025-07-10 18:40:31,703 - INFO - log_σ² gradient: -0.522988 |
|
|
2025-07-10 18:40:31,775 - INFO - Optimizer step 8: log_σ²=0.063585, weight=0.938394 |
|
|
2025-07-10 18:40:57,901 - INFO - log_σ² gradient: -0.517776 |
|
|
2025-07-10 18:40:57,978 - INFO - Optimizer step 9: log_σ²=0.063890, weight=0.938108 |
|
|
2025-07-10 18:41:21,679 - INFO - log_σ² gradient: -0.509147 |
|
|
2025-07-10 18:41:21,743 - INFO - Optimizer step 10: log_σ²=0.064196, weight=0.937821 |
|
|
2025-07-10 18:41:45,055 - INFO - log_σ² gradient: -0.530793 |
|
|
2025-07-10 18:41:45,126 - INFO - Optimizer step 11: log_σ²=0.064505, weight=0.937532 |
|
|
2025-07-10 18:42:09,301 - INFO - log_σ² gradient: -0.570221 |
|
|
2025-07-10 18:42:09,373 - INFO - Optimizer step 12: log_σ²=0.064817, weight=0.937239 |
|
|
2025-07-10 18:42:34,851 - INFO - log_σ² gradient: -0.565037 |
|
|
2025-07-10 18:42:34,926 - INFO - Optimizer step 13: log_σ²=0.065134, weight=0.936942 |
|
|
2025-07-10 18:42:58,575 - INFO - log_σ² gradient: -0.536559 |
|
|
2025-07-10 18:42:58,643 - INFO - Optimizer step 14: log_σ²=0.065452, weight=0.936644 |
|
|
2025-07-10 18:43:22,947 - INFO - log_σ² gradient: -0.532557 |
|
|
2025-07-10 18:43:23,018 - INFO - Optimizer step 15: log_σ²=0.065771, weight=0.936345 |
|
|
2025-07-10 18:43:46,217 - INFO - log_σ² gradient: -0.547007 |
|
|
2025-07-10 18:43:46,296 - INFO - Optimizer step 16: log_σ²=0.066093, weight=0.936044 |
|
|
2025-07-10 18:44:12,249 - INFO - log_σ² gradient: -0.553911 |
|
|
2025-07-10 18:44:12,323 - INFO - Optimizer step 17: log_σ²=0.066417, weight=0.935740 |
|
|
2025-07-10 18:44:35,973 - INFO - log_σ² gradient: -0.551877 |
|
|
2025-07-10 18:44:36,049 - INFO - Optimizer step 18: log_σ²=0.066744, weight=0.935435 |
|
|
2025-07-10 18:45:00,796 - INFO - log_σ² gradient: -0.548452 |
|
|
2025-07-10 18:45:00,866 - INFO - Optimizer step 19: log_σ²=0.067072, weight=0.935128 |
|
|
2025-07-10 18:45:26,877 - INFO - log_σ² gradient: -0.534554 |
|
|
2025-07-10 18:45:26,956 - INFO - Optimizer step 20: log_σ²=0.067401, weight=0.934820 |
|
|
2025-07-10 18:45:50,813 - INFO - log_σ² gradient: -0.570899 |
|
|
2025-07-10 18:45:50,889 - INFO - Optimizer step 21: log_σ²=0.067733, weight=0.934510 |
|
|
2025-07-10 18:46:15,094 - INFO - log_σ² gradient: -0.696415 |
|
|
2025-07-10 18:46:15,169 - INFO - Optimizer step 22: log_σ²=0.068076, weight=0.934190 |
|
|
2025-07-10 18:46:39,895 - INFO - log_σ² gradient: -0.690586 |
|
|
2025-07-10 18:46:39,961 - INFO - Optimizer step 23: log_σ²=0.068427, weight=0.933861 |
|
|
2025-07-10 18:47:04,892 - INFO - log_σ² gradient: -0.627980 |
|
|
2025-07-10 18:47:04,968 - INFO - Optimizer step 24: log_σ²=0.068784, weight=0.933529 |
|
|
2025-07-10 18:47:29,475 - INFO - log_σ² gradient: -0.602425 |
|
|
2025-07-10 18:47:29,554 - INFO - Optimizer step 25: log_σ²=0.069142, weight=0.933194 |
|
|
2025-07-10 18:47:54,317 - INFO - log_σ² gradient: -0.581720 |
|
|
2025-07-10 18:47:54,390 - INFO - Optimizer step 26: log_σ²=0.069502, weight=0.932858 |
|
|
2025-07-10 18:48:17,644 - INFO - log_σ² gradient: -0.594081 |
|
|
2025-07-10 18:48:17,715 - INFO - Optimizer step 27: log_σ²=0.069864, weight=0.932521 |
|
|
2025-07-10 18:48:42,754 - INFO - log_σ² gradient: -0.549293 |
|
|
2025-07-10 18:48:42,836 - INFO - Optimizer step 28: log_σ²=0.070225, weight=0.932184 |
|
|
2025-07-10 18:49:05,163 - INFO - log_σ² gradient: -0.565008 |
|
|
2025-07-10 18:49:05,242 - INFO - Optimizer step 29: log_σ²=0.070586, weight=0.931848 |
|
|
2025-07-10 18:49:28,100 - INFO - log_σ² gradient: -0.558473 |
|
|
2025-07-10 18:49:28,172 - INFO - Optimizer step 30: log_σ²=0.070947, weight=0.931512 |
|
|
2025-07-10 18:49:39,250 - INFO - log_σ² gradient: -0.247689 |
|
|
2025-07-10 18:49:39,322 - INFO - Optimizer step 31: log_σ²=0.071288, weight=0.931193 |
|
|
2025-07-10 18:49:39,499 - INFO - Epoch 14: Total optimizer steps: 31 |
|
|
2025-07-10 18:52:58,595 - INFO - Validation metrics: |
|
|
2025-07-10 18:52:58,595 - INFO - Loss: 0.7023 |
|
|
2025-07-10 18:52:58,595 - INFO - Average similarity: 0.8296 |
|
|
2025-07-10 18:52:58,595 - INFO - Median similarity: 0.9777 |
|
|
2025-07-10 18:52:58,595 - INFO - Clean sample similarity: 0.8296 |
|
|
2025-07-10 18:52:58,595 - INFO - Corrupted sample similarity: 0.4273 |
|
|
2025-07-10 18:52:58,595 - INFO - Similarity gap (clean - corrupt): 0.4024 |
|
|
2025-07-10 18:52:58,685 - INFO - Epoch 14/30 - Train Loss: 0.8791, Val Loss: 0.7023, Clean Sim: 0.8296, Corrupt Sim: 0.4273, Gap: 0.4024, Time: 982.95s |
|
|
2025-07-10 18:55:39,909 - INFO - Epoch 14 Validation Alignment: Pos=0.145, Neg=0.103, Gap=0.042 |
|
|
2025-07-10 18:56:47,972 - INFO - log_σ² gradient: -0.513230 |
|
|
2025-07-10 18:56:48,044 - INFO - Optimizer step 1: log_σ²=0.071629, weight=0.930876 |
|
|
2025-07-10 18:57:13,115 - INFO - log_σ² gradient: -0.524490 |
|
|
2025-07-10 18:57:13,194 - INFO - Optimizer step 2: log_σ²=0.071970, weight=0.930559 |
|
|
2025-07-10 18:57:38,051 - INFO - log_σ² gradient: -0.530914 |
|
|
2025-07-10 18:57:38,117 - INFO - Optimizer step 3: log_σ²=0.072311, weight=0.930242 |
|
|
2025-07-10 18:58:01,971 - INFO - log_σ² gradient: -0.534616 |
|
|
2025-07-10 18:58:02,040 - INFO - Optimizer step 4: log_σ²=0.072653, weight=0.929924 |
|
|
2025-07-10 18:58:23,917 - INFO - log_σ² gradient: -0.537152 |
|
|
2025-07-10 18:58:23,988 - INFO - Optimizer step 5: log_σ²=0.072995, weight=0.929605 |
|
|
2025-07-10 18:58:47,613 - INFO - log_σ² gradient: -0.519909 |
|
|
2025-07-10 18:58:47,685 - INFO - Optimizer step 6: log_σ²=0.073338, weight=0.929287 |
|
|
2025-07-10 18:59:11,126 - INFO - log_σ² gradient: -0.516431 |
|
|
2025-07-10 18:59:11,205 - INFO - Optimizer step 7: log_σ²=0.073680, weight=0.928969 |
|
|
2025-07-10 18:59:35,180 - INFO - log_σ² gradient: -0.514404 |
|
|
2025-07-10 18:59:35,259 - INFO - Optimizer step 8: log_σ²=0.074022, weight=0.928651 |
|
|
2025-07-10 18:59:59,723 - INFO - log_σ² gradient: -0.530065 |
|
|
2025-07-10 18:59:59,794 - INFO - Optimizer step 9: log_σ²=0.074365, weight=0.928333 |
|
|
2025-07-10 19:00:24,323 - INFO - log_σ² gradient: -0.521644 |
|
|
2025-07-10 19:00:24,396 - INFO - Optimizer step 10: log_σ²=0.074708, weight=0.928015 |
|
|
2025-07-10 19:00:47,232 - INFO - log_σ² gradient: -0.514367 |
|
|
2025-07-10 19:00:47,310 - INFO - Optimizer step 11: log_σ²=0.075051, weight=0.927696 |
|
|
2025-07-10 19:01:11,458 - INFO - log_σ² gradient: -0.515804 |
|
|
2025-07-10 19:01:11,537 - INFO - Optimizer step 12: log_σ²=0.075394, weight=0.927378 |
|
|
2025-07-10 19:01:35,817 - INFO - log_σ² gradient: -0.513997 |
|
|
2025-07-10 19:01:35,890 - INFO - Optimizer step 13: log_σ²=0.075736, weight=0.927061 |
|
|
2025-07-10 19:01:58,935 - INFO - log_σ² gradient: -0.510152 |
|
|
2025-07-10 19:01:59,014 - INFO - Optimizer step 14: log_σ²=0.076079, weight=0.926743 |
|
|
2025-07-10 19:02:23,974 - INFO - log_σ² gradient: -0.508527 |
|
|
2025-07-10 19:02:24,056 - INFO - Optimizer step 15: log_σ²=0.076422, weight=0.926426 |
|
|
2025-07-10 19:02:48,648 - INFO - log_σ² gradient: -0.506742 |
|
|
2025-07-10 19:02:48,716 - INFO - Optimizer step 16: log_σ²=0.076764, weight=0.926109 |
|
|
2025-07-10 19:03:13,569 - INFO - log_σ² gradient: -0.516821 |
|
|
2025-07-10 19:03:13,649 - INFO - Optimizer step 17: log_σ²=0.077107, weight=0.925791 |
|
|
2025-07-10 19:03:37,476 - INFO - log_σ² gradient: -0.508687 |
|
|
2025-07-10 19:03:37,550 - INFO - Optimizer step 18: log_σ²=0.077449, weight=0.925474 |
|
|
2025-07-10 19:04:02,904 - INFO - log_σ² gradient: -0.502677 |
|
|
2025-07-10 19:04:02,976 - INFO - Optimizer step 19: log_σ²=0.077792, weight=0.925157 |
|
|
2025-07-10 19:04:25,858 - INFO - log_σ² gradient: -0.512708 |
|
|
2025-07-10 19:04:25,930 - INFO - Optimizer step 20: log_σ²=0.078135, weight=0.924839 |
|
|
2025-07-10 19:04:48,814 - INFO - log_σ² gradient: -0.501101 |
|
|
2025-07-10 19:04:48,882 - INFO - Optimizer step 21: log_σ²=0.078478, weight=0.924522 |
|
|
2025-07-10 19:05:13,739 - INFO - log_σ² gradient: -0.495576 |
|
|
2025-07-10 19:05:13,809 - INFO - Optimizer step 22: log_σ²=0.078820, weight=0.924206 |
|
|
2025-07-10 19:05:37,545 - INFO - log_σ² gradient: -0.519050 |
|
|
2025-07-10 19:05:37,624 - INFO - Optimizer step 23: log_σ²=0.079164, weight=0.923889 |
|
|
2025-07-10 19:06:01,089 - INFO - log_σ² gradient: -0.501411 |
|
|
2025-07-10 19:06:01,168 - INFO - Optimizer step 24: log_σ²=0.079507, weight=0.923571 |
|
|
2025-07-10 19:06:25,550 - INFO - log_σ² gradient: -0.506722 |
|
|
2025-07-10 19:06:25,619 - INFO - Optimizer step 25: log_σ²=0.079851, weight=0.923254 |
|
|
2025-07-10 19:06:50,154 - INFO - log_σ² gradient: -0.501817 |
|
|
2025-07-10 19:06:50,228 - INFO - Optimizer step 26: log_σ²=0.080195, weight=0.922936 |
|
|
2025-07-10 19:07:13,382 - INFO - log_σ² gradient: -0.506251 |
|
|
2025-07-10 19:07:13,450 - INFO - Optimizer step 27: log_σ²=0.080540, weight=0.922618 |
|
|
2025-07-10 19:07:37,619 - INFO - log_σ² gradient: -0.500636 |
|
|
2025-07-10 19:07:37,687 - INFO - Optimizer step 28: log_σ²=0.080884, weight=0.922301 |
|
|
2025-07-10 19:07:59,675 - INFO - log_σ² gradient: -0.514477 |
|
|
2025-07-10 19:07:59,746 - INFO - Optimizer step 29: log_σ²=0.081230, weight=0.921982 |
|
|
2025-07-10 19:08:25,522 - INFO - log_σ² gradient: -0.508771 |
|
|
2025-07-10 19:08:25,600 - INFO - Optimizer step 30: log_σ²=0.081576, weight=0.921662 |
|
|
2025-07-10 19:08:36,755 - INFO - log_σ² gradient: -0.236138 |
|
|
2025-07-10 19:08:36,826 - INFO - Optimizer step 31: log_σ²=0.081905, weight=0.921359 |
|
|
2025-07-10 19:08:36,990 - INFO - Epoch 15: Total optimizer steps: 31 |
|
|
2025-07-10 19:11:56,891 - INFO - Validation metrics: |
|
|
2025-07-10 19:11:56,892 - INFO - Loss: 0.6371 |
|
|
2025-07-10 19:11:56,892 - INFO - Average similarity: 0.7917 |
|
|
2025-07-10 19:11:56,892 - INFO - Median similarity: 0.9793 |
|
|
2025-07-10 19:11:56,892 - INFO - Clean sample similarity: 0.7917 |
|
|
2025-07-10 19:11:56,892 - INFO - Corrupted sample similarity: 0.3418 |
|
|
2025-07-10 19:11:56,892 - INFO - Similarity gap (clean - corrupt): 0.4500 |
|
|
2025-07-10 19:11:57,014 - INFO - Epoch 15/30 - Train Loss: 0.6891, Val Loss: 0.6371, Clean Sim: 0.7917, Corrupt Sim: 0.3418, Gap: 0.4500, Time: 977.11s |
|
|
2025-07-10 19:11:57,014 - INFO - New best validation loss: 0.6371 |
|
|
2025-07-10 19:12:03,241 - INFO - New best similarity gap: 0.4500 |
|
|
2025-07-10 19:13:23,886 - INFO - log_σ² gradient: -0.506652 |
|
|
2025-07-10 19:13:23,957 - INFO - Optimizer step 1: log_σ²=0.082237, weight=0.921054 |
|
|
2025-07-10 19:13:46,717 - INFO - log_σ² gradient: -0.489263 |
|
|
2025-07-10 19:13:46,788 - INFO - Optimizer step 2: log_σ²=0.082569, weight=0.920748 |
|
|
2025-07-10 19:14:10,814 - INFO - log_σ² gradient: -0.501242 |
|
|
2025-07-10 19:14:10,886 - INFO - Optimizer step 3: log_σ²=0.082903, weight=0.920440 |
|
|
2025-07-10 19:14:35,182 - INFO - log_σ² gradient: -0.496462 |
|
|
2025-07-10 19:14:35,255 - INFO - Optimizer step 4: log_σ²=0.083239, weight=0.920131 |
|
|
2025-07-10 19:14:59,320 - INFO - log_σ² gradient: -0.510647 |
|
|
2025-07-10 19:14:59,391 - INFO - Optimizer step 5: log_σ²=0.083577, weight=0.919820 |
|
|
2025-07-10 19:15:22,730 - INFO - log_σ² gradient: -0.498962 |
|
|
2025-07-10 19:15:22,794 - INFO - Optimizer step 6: log_σ²=0.083916, weight=0.919508 |
|
|
2025-07-10 19:15:46,890 - INFO - log_σ² gradient: -0.496972 |
|
|
2025-07-10 19:15:46,969 - INFO - Optimizer step 7: log_σ²=0.084257, weight=0.919195 |
|
|
2025-07-10 19:16:10,380 - INFO - log_σ² gradient: -0.494926 |
|
|
2025-07-10 19:16:10,454 - INFO - Optimizer step 8: log_σ²=0.084599, weight=0.918881 |
|
|
2025-07-10 19:16:33,676 - INFO - log_σ² gradient: -0.493579 |
|
|
2025-07-10 19:16:33,750 - INFO - Optimizer step 9: log_σ²=0.084941, weight=0.918566 |
|
|
2025-07-10 19:16:57,819 - INFO - log_σ² gradient: -0.495823 |
|
|
2025-07-10 19:16:57,897 - INFO - Optimizer step 10: log_σ²=0.085285, weight=0.918251 |
|
|
2025-07-10 19:17:21,001 - INFO - log_σ² gradient: -0.495975 |
|
|
2025-07-10 19:17:21,074 - INFO - Optimizer step 11: log_σ²=0.085629, weight=0.917934 |
|
|
2025-07-10 19:17:44,815 - INFO - log_σ² gradient: -0.498009 |
|
|
2025-07-10 19:17:44,887 - INFO - Optimizer step 12: log_σ²=0.085975, weight=0.917617 |
|
|
2025-07-10 19:18:08,789 - INFO - log_σ² gradient: -0.498226 |
|
|
2025-07-10 19:18:08,872 - INFO - Optimizer step 13: log_σ²=0.086322, weight=0.917299 |
|
|
2025-07-10 19:18:32,798 - INFO - log_σ² gradient: -0.494703 |
|
|
2025-07-10 19:18:32,870 - INFO - Optimizer step 14: log_σ²=0.086670, weight=0.916980 |
|
|
2025-07-10 19:18:57,459 - INFO - log_σ² gradient: -0.490732 |
|
|
2025-07-10 19:18:57,535 - INFO - Optimizer step 15: log_σ²=0.087018, weight=0.916660 |
|
|
2025-07-10 19:19:20,985 - INFO - log_σ² gradient: -0.495979 |
|
|
2025-07-10 19:19:21,065 - INFO - Optimizer step 16: log_σ²=0.087368, weight=0.916340 |
|
|
2025-07-10 19:19:46,330 - INFO - log_σ² gradient: -0.506442 |
|
|
2025-07-10 19:19:46,409 - INFO - Optimizer step 17: log_σ²=0.087719, weight=0.916019 |
|
|
2025-07-10 19:20:10,729 - INFO - log_σ² gradient: -0.493309 |
|
|
2025-07-10 19:20:10,803 - INFO - Optimizer step 18: log_σ²=0.088070, weight=0.915697 |
|
|
2025-07-10 19:20:35,447 - INFO - log_σ² gradient: -0.507596 |
|
|
2025-07-10 19:20:35,518 - INFO - Optimizer step 19: log_σ²=0.088424, weight=0.915373 |
|
|
2025-07-10 19:20:59,936 - INFO - log_σ² gradient: -0.498363 |
|
|
2025-07-10 19:20:59,999 - INFO - Optimizer step 20: log_σ²=0.088778, weight=0.915049 |
|
|
2025-07-10 19:21:23,122 - INFO - log_σ² gradient: -0.500010 |
|
|
2025-07-10 19:21:23,195 - INFO - Optimizer step 21: log_σ²=0.089133, weight=0.914724 |
|
|
2025-07-10 19:21:47,175 - INFO - log_σ² gradient: -0.491464 |
|
|
2025-07-10 19:21:47,247 - INFO - Optimizer step 22: log_σ²=0.089489, weight=0.914398 |
|
|
2025-07-10 19:22:13,267 - INFO - log_σ² gradient: -0.501515 |
|
|
2025-07-10 19:22:13,333 - INFO - Optimizer step 23: log_σ²=0.089846, weight=0.914072 |
|
|
2025-07-10 19:22:38,313 - INFO - log_σ² gradient: -0.508589 |
|
|
2025-07-10 19:22:38,392 - INFO - Optimizer step 24: log_σ²=0.090205, weight=0.913744 |
|
|
2025-07-10 19:23:02,944 - INFO - log_σ² gradient: -0.496165 |
|
|
2025-07-10 19:23:03,014 - INFO - Optimizer step 25: log_σ²=0.090564, weight=0.913416 |
|
|
2025-07-10 19:23:28,555 - INFO - log_σ² gradient: -0.506895 |
|
|
2025-07-10 19:23:28,630 - INFO - Optimizer step 26: log_σ²=0.090925, weight=0.913086 |
|
|
2025-07-10 19:23:52,267 - INFO - log_σ² gradient: -0.491906 |
|
|
2025-07-10 19:23:52,346 - INFO - Optimizer step 27: log_σ²=0.091286, weight=0.912757 |
|
|
2025-07-10 19:24:14,793 - INFO - log_σ² gradient: -0.494699 |
|
|
2025-07-10 19:24:14,859 - INFO - Optimizer step 28: log_σ²=0.091648, weight=0.912426 |
|
|
2025-07-10 19:24:40,006 - INFO - log_σ² gradient: -0.482729 |
|
|
2025-07-10 19:24:40,082 - INFO - Optimizer step 29: log_σ²=0.092009, weight=0.912097 |
|
|
2025-07-10 19:25:03,713 - INFO - log_σ² gradient: -0.492009 |
|
|
2025-07-10 19:25:03,789 - INFO - Optimizer step 30: log_σ²=0.092371, weight=0.911767 |
|
|
2025-07-10 19:25:15,146 - INFO - log_σ² gradient: -0.230564 |
|
|
2025-07-10 19:25:15,216 - INFO - Optimizer step 31: log_σ²=0.092715, weight=0.911454 |
|
|
2025-07-10 19:25:15,401 - INFO - Epoch 16: Total optimizer steps: 31 |
|
|
2025-07-10 19:28:34,567 - INFO - Validation metrics: |
|
|
2025-07-10 19:28:34,567 - INFO - Loss: 0.6358 |
|
|
2025-07-10 19:28:34,567 - INFO - Average similarity: 0.9004 |
|
|
2025-07-10 19:28:34,567 - INFO - Median similarity: 0.9926 |
|
|
2025-07-10 19:28:34,567 - INFO - Clean sample similarity: 0.9004 |
|
|
2025-07-10 19:28:34,567 - INFO - Corrupted sample similarity: 0.4322 |
|
|
2025-07-10 19:28:34,567 - INFO - Similarity gap (clean - corrupt): 0.4682 |
|
|
2025-07-10 19:28:34,670 - INFO - Epoch 16/30 - Train Loss: 0.6572, Val Loss: 0.6358, Clean Sim: 0.9004, Corrupt Sim: 0.4322, Gap: 0.4682, Time: 978.77s |
|
|
2025-07-10 19:28:34,670 - INFO - New best validation loss: 0.6358 |
|
|
2025-07-10 19:28:40,841 - INFO - New best similarity gap: 0.4682 |
|
|
2025-07-10 19:31:28,701 - INFO - Epoch 16 Validation Alignment: Pos=0.151, Neg=0.092, Gap=0.059 |
|
|
2025-07-10 19:32:35,736 - INFO - log_σ² gradient: -0.487571 |
|
|
2025-07-10 19:32:35,811 - INFO - Optimizer step 1: log_σ²=0.093060, weight=0.911139 |
|
|
2025-07-10 19:33:02,025 - INFO - log_σ² gradient: -0.490278 |
|
|
2025-07-10 19:33:02,103 - INFO - Optimizer step 2: log_σ²=0.093408, weight=0.910822 |
|
|
2025-07-10 19:33:27,664 - INFO - log_σ² gradient: -0.493809 |
|
|
2025-07-10 19:33:27,740 - INFO - Optimizer step 3: log_σ²=0.093758, weight=0.910503 |
|
|
2025-07-10 19:33:51,307 - INFO - log_σ² gradient: -0.494680 |
|
|
2025-07-10 19:33:51,386 - INFO - Optimizer step 4: log_σ²=0.094111, weight=0.910182 |
|
|
2025-07-10 19:34:15,967 - INFO - log_σ² gradient: -0.492143 |
|
|
2025-07-10 19:34:16,042 - INFO - Optimizer step 5: log_σ²=0.094465, weight=0.909859 |
|
|
2025-07-10 19:34:40,855 - INFO - log_σ² gradient: -0.490081 |
|
|
2025-07-10 19:34:40,923 - INFO - Optimizer step 6: log_σ²=0.094821, weight=0.909536 |
|
|
2025-07-10 19:35:05,279 - INFO - log_σ² gradient: -0.482258 |
|
|
2025-07-10 19:35:05,358 - INFO - Optimizer step 7: log_σ²=0.095178, weight=0.909211 |
|
|
2025-07-10 19:35:32,051 - INFO - log_σ² gradient: -0.498510 |
|
|
2025-07-10 19:35:32,126 - INFO - Optimizer step 8: log_σ²=0.095537, weight=0.908885 |
|
|
2025-07-10 19:35:54,970 - INFO - log_σ² gradient: -0.491812 |
|
|
2025-07-10 19:35:55,049 - INFO - Optimizer step 9: log_σ²=0.095898, weight=0.908557 |
|
|
2025-07-10 19:36:19,932 - INFO - log_σ² gradient: -0.483600 |
|
|
2025-07-10 19:36:20,010 - INFO - Optimizer step 10: log_σ²=0.096259, weight=0.908228 |
|
|
2025-07-10 19:36:43,679 - INFO - log_σ² gradient: -0.487231 |
|
|
2025-07-10 19:36:43,750 - INFO - Optimizer step 11: log_σ²=0.096622, weight=0.907899 |
|
|
2025-07-10 19:37:06,219 - INFO - log_σ² gradient: -0.497391 |
|
|
2025-07-10 19:37:06,292 - INFO - Optimizer step 12: log_σ²=0.096986, weight=0.907568 |
|
|
2025-07-10 19:37:30,900 - INFO - log_σ² gradient: -0.495075 |
|
|
2025-07-10 19:37:30,972 - INFO - Optimizer step 13: log_σ²=0.097352, weight=0.907236 |
|
|
2025-07-10 19:37:55,217 - INFO - log_σ² gradient: -0.492056 |
|
|
2025-07-10 19:37:55,286 - INFO - Optimizer step 14: log_σ²=0.097720, weight=0.906903 |
|
|
2025-07-10 19:38:18,996 - INFO - log_σ² gradient: -0.483765 |
|
|
2025-07-10 19:38:19,071 - INFO - Optimizer step 15: log_σ²=0.098088, weight=0.906569 |
|
|
2025-07-10 19:38:42,826 - INFO - log_σ² gradient: -0.496189 |
|
|
2025-07-10 19:38:42,897 - INFO - Optimizer step 16: log_σ²=0.098457, weight=0.906235 |
|
|
2025-07-10 19:39:06,946 - INFO - log_σ² gradient: -0.488257 |
|
|
2025-07-10 19:39:07,020 - INFO - Optimizer step 17: log_σ²=0.098828, weight=0.905899 |
|
|
2025-07-10 19:39:31,050 - INFO - log_σ² gradient: -0.487438 |
|
|
2025-07-10 19:39:31,122 - INFO - Optimizer step 18: log_σ²=0.099199, weight=0.905563 |
|
|
2025-07-10 19:39:54,577 - INFO - log_σ² gradient: -0.483195 |
|
|
2025-07-10 19:39:54,645 - INFO - Optimizer step 19: log_σ²=0.099570, weight=0.905226 |
|
|
2025-07-10 19:40:19,837 - INFO - log_σ² gradient: -0.498274 |
|
|
2025-07-10 19:40:19,909 - INFO - Optimizer step 20: log_σ²=0.099944, weight=0.904888 |
|
|
2025-07-10 19:40:42,831 - INFO - log_σ² gradient: -0.483536 |
|
|
2025-07-10 19:40:42,907 - INFO - Optimizer step 21: log_σ²=0.100318, weight=0.904550 |
|
|
2025-07-10 19:41:07,502 - INFO - log_σ² gradient: -0.487251 |
|
|
2025-07-10 19:41:07,578 - INFO - Optimizer step 22: log_σ²=0.100692, weight=0.904211 |
|
|
2025-07-10 19:41:31,060 - INFO - log_σ² gradient: -0.485231 |
|
|
2025-07-10 19:41:31,129 - INFO - Optimizer step 23: log_σ²=0.101067, weight=0.903872 |
|
|
2025-07-10 19:41:56,068 - INFO - log_σ² gradient: -0.489868 |
|
|
2025-07-10 19:41:56,146 - INFO - Optimizer step 24: log_σ²=0.101444, weight=0.903532 |
|
|
2025-07-10 19:42:20,236 - INFO - log_σ² gradient: -0.497105 |
|
|
2025-07-10 19:42:20,291 - INFO - Optimizer step 25: log_σ²=0.101821, weight=0.903191 |
|
|
2025-07-10 19:42:45,024 - INFO - log_σ² gradient: -0.495480 |
|
|
2025-07-10 19:42:45,095 - INFO - Optimizer step 26: log_σ²=0.102200, weight=0.902849 |
|
|
2025-07-10 19:43:08,111 - INFO - log_σ² gradient: -0.484709 |
|
|
2025-07-10 19:43:08,180 - INFO - Optimizer step 27: log_σ²=0.102580, weight=0.902506 |
|
|
2025-07-10 19:43:35,092 - INFO - log_σ² gradient: -0.488335 |
|
|
2025-07-10 19:43:35,163 - INFO - Optimizer step 28: log_σ²=0.102960, weight=0.902163 |
|
|
2025-07-10 19:43:59,191 - INFO - log_σ² gradient: -0.490414 |
|
|
2025-07-10 19:43:59,271 - INFO - Optimizer step 29: log_σ²=0.103342, weight=0.901819 |
|
|
2025-07-10 19:44:22,121 - INFO - log_σ² gradient: -0.486700 |
|
|
2025-07-10 19:44:22,185 - INFO - Optimizer step 30: log_σ²=0.103724, weight=0.901474 |
|
|
2025-07-10 19:44:33,132 - INFO - log_σ² gradient: -0.227770 |
|
|
2025-07-10 19:44:33,208 - INFO - Optimizer step 31: log_σ²=0.104086, weight=0.901148 |
|
|
2025-07-10 19:44:33,384 - INFO - Epoch 17: Total optimizer steps: 31 |
|
|
2025-07-10 19:47:52,105 - INFO - Validation metrics: |
|
|
2025-07-10 19:47:52,106 - INFO - Loss: 0.6148 |
|
|
2025-07-10 19:47:52,106 - INFO - Average similarity: 0.7811 |
|
|
2025-07-10 19:47:52,106 - INFO - Median similarity: 0.9655 |
|
|
2025-07-10 19:47:52,106 - INFO - Clean sample similarity: 0.7811 |
|
|
2025-07-10 19:47:52,106 - INFO - Corrupted sample similarity: 0.2877 |
|
|
2025-07-10 19:47:52,106 - INFO - Similarity gap (clean - corrupt): 0.4933 |
|
|
2025-07-10 19:47:52,218 - INFO - Epoch 17/30 - Train Loss: 0.6461, Val Loss: 0.6148, Clean Sim: 0.7811, Corrupt Sim: 0.2877, Gap: 0.4933, Time: 983.52s |
|
|
2025-07-10 19:47:52,219 - INFO - New best validation loss: 0.6148 |
|
|
2025-07-10 19:47:58,345 - INFO - New best similarity gap: 0.4933 |
|
|
2025-07-10 19:49:12,633 - INFO - log_σ² gradient: -0.493563 |
|
|
2025-07-10 19:49:12,711 - INFO - Optimizer step 1: log_σ²=0.104452, weight=0.900818 |
|
|
2025-07-10 19:49:37,367 - INFO - log_σ² gradient: -0.489635 |
|
|
2025-07-10 19:49:37,433 - INFO - Optimizer step 2: log_σ²=0.104820, weight=0.900486 |
|
|
2025-07-10 19:50:00,713 - INFO - log_σ² gradient: -0.490291 |
|
|
2025-07-10 19:50:00,792 - INFO - Optimizer step 3: log_σ²=0.105191, weight=0.900153 |
|
|
2025-07-10 19:50:24,554 - INFO - log_σ² gradient: -0.489970 |
|
|
2025-07-10 19:50:24,625 - INFO - Optimizer step 4: log_σ²=0.105564, weight=0.899817 |
|
|
2025-07-10 19:50:48,378 - INFO - log_σ² gradient: -0.477850 |
|
|
2025-07-10 19:50:48,442 - INFO - Optimizer step 5: log_σ²=0.105939, weight=0.899480 |
|
|
2025-07-10 19:51:12,057 - INFO - log_σ² gradient: -0.483001 |
|
|
2025-07-10 19:51:12,135 - INFO - Optimizer step 6: log_σ²=0.106315, weight=0.899141 |
|
|
2025-07-10 19:51:36,160 - INFO - log_σ² gradient: -0.490630 |
|
|
2025-07-10 19:51:36,239 - INFO - Optimizer step 7: log_σ²=0.106693, weight=0.898801 |
|
|
2025-07-10 19:52:02,073 - INFO - log_σ² gradient: -0.490138 |
|
|
2025-07-10 19:52:02,145 - INFO - Optimizer step 8: log_σ²=0.107073, weight=0.898460 |
|
|
2025-07-10 19:52:25,592 - INFO - log_σ² gradient: -0.479919 |
|
|
2025-07-10 19:52:25,668 - INFO - Optimizer step 9: log_σ²=0.107454, weight=0.898117 |
|
|
2025-07-10 19:52:51,012 - INFO - log_σ² gradient: -0.485070 |
|
|
2025-07-10 19:52:51,083 - INFO - Optimizer step 10: log_σ²=0.107837, weight=0.897774 |
|
|
2025-07-10 19:53:14,363 - INFO - log_σ² gradient: -0.491695 |
|
|
2025-07-10 19:53:14,438 - INFO - Optimizer step 11: log_σ²=0.108222, weight=0.897429 |
|
|
2025-07-10 19:53:39,113 - INFO - log_σ² gradient: -0.487314 |
|
|
2025-07-10 19:53:39,185 - INFO - Optimizer step 12: log_σ²=0.108607, weight=0.897083 |
|
|
2025-07-10 19:54:03,490 - INFO - log_σ² gradient: -0.487364 |
|
|
2025-07-10 19:54:03,561 - INFO - Optimizer step 13: log_σ²=0.108995, weight=0.896735 |
|
|
2025-07-10 19:54:27,497 - INFO - log_σ² gradient: -0.479694 |
|
|
2025-07-10 19:54:27,563 - INFO - Optimizer step 14: log_σ²=0.109383, weight=0.896387 |
|
|
2025-07-10 19:54:51,999 - INFO - log_σ² gradient: -0.478104 |
|
|
2025-07-10 19:54:52,074 - INFO - Optimizer step 15: log_σ²=0.109771, weight=0.896039 |
|
|
2025-07-10 19:55:15,836 - INFO - log_σ² gradient: -0.482936 |
|
|
2025-07-10 19:55:15,915 - INFO - Optimizer step 16: log_σ²=0.110161, weight=0.895690 |
|
|
2025-07-10 19:55:39,947 - INFO - log_σ² gradient: -0.485020 |
|
|
2025-07-10 19:55:40,023 - INFO - Optimizer step 17: log_σ²=0.110552, weight=0.895340 |
|
|
2025-07-10 19:56:04,378 - INFO - log_σ² gradient: -0.491121 |
|
|
2025-07-10 19:56:04,449 - INFO - Optimizer step 18: log_σ²=0.110944, weight=0.894989 |
|
|
2025-07-10 19:56:28,739 - INFO - log_σ² gradient: -0.473956 |
|
|
2025-07-10 19:56:28,815 - INFO - Optimizer step 19: log_σ²=0.111336, weight=0.894638 |
|
|
2025-07-10 19:56:52,715 - INFO - log_σ² gradient: -0.489448 |
|
|
2025-07-10 19:56:52,791 - INFO - Optimizer step 20: log_σ²=0.111730, weight=0.894285 |
|
|
2025-07-10 19:57:17,718 - INFO - log_σ² gradient: -0.478794 |
|
|
2025-07-10 19:57:17,790 - INFO - Optimizer step 21: log_σ²=0.112125, weight=0.893933 |
|
|
2025-07-10 19:57:40,416 - INFO - log_σ² gradient: -0.484692 |
|
|
2025-07-10 19:57:40,488 - INFO - Optimizer step 22: log_σ²=0.112520, weight=0.893579 |
|
|
2025-07-10 19:58:02,744 - INFO - log_σ² gradient: -0.485239 |
|
|
2025-07-10 19:58:02,818 - INFO - Optimizer step 23: log_σ²=0.112917, weight=0.893225 |
|
|
2025-07-10 19:58:25,784 - INFO - log_σ² gradient: -0.486221 |
|
|
2025-07-10 19:58:25,853 - INFO - Optimizer step 24: log_σ²=0.113314, weight=0.892870 |
|
|
2025-07-10 19:58:50,232 - INFO - log_σ² gradient: -0.476134 |
|
|
2025-07-10 19:58:50,304 - INFO - Optimizer step 25: log_σ²=0.113712, weight=0.892515 |
|
|
2025-07-10 19:59:14,228 - INFO - log_σ² gradient: -0.477603 |
|
|
2025-07-10 19:59:14,303 - INFO - Optimizer step 26: log_σ²=0.114111, weight=0.892159 |
|
|
2025-07-10 19:59:38,083 - INFO - log_σ² gradient: -0.500643 |
|
|
2025-07-10 19:59:38,151 - INFO - Optimizer step 27: log_σ²=0.114511, weight=0.891802 |
|
|
2025-07-10 20:00:02,263 - INFO - log_σ² gradient: -0.480895 |
|
|
2025-07-10 20:00:02,337 - INFO - Optimizer step 28: log_σ²=0.114913, weight=0.891444 |
|
|
2025-07-10 20:00:25,958 - INFO - log_σ² gradient: -0.471780 |
|
|
2025-07-10 20:00:26,032 - INFO - Optimizer step 29: log_σ²=0.115314, weight=0.891086 |
|
|
2025-07-10 20:00:50,293 - INFO - log_σ² gradient: -0.474412 |
|
|
2025-07-10 20:00:50,371 - INFO - Optimizer step 30: log_σ²=0.115715, weight=0.890729 |
|
|
2025-07-10 20:01:00,714 - INFO - log_σ² gradient: -0.225541 |
|
|
2025-07-10 20:01:00,785 - INFO - Optimizer step 31: log_σ²=0.116096, weight=0.890390 |
|
|
2025-07-10 20:01:00,976 - INFO - Epoch 18: Total optimizer steps: 31 |
|
|
2025-07-10 20:04:17,858 - INFO - Validation metrics: |
|
|
2025-07-10 20:04:17,858 - INFO - Loss: 0.6189 |
|
|
2025-07-10 20:04:17,858 - INFO - Average similarity: 0.9612 |
|
|
2025-07-10 20:04:17,858 - INFO - Median similarity: 0.9989 |
|
|
2025-07-10 20:04:17,858 - INFO - Clean sample similarity: 0.9612 |
|
|
2025-07-10 20:04:17,858 - INFO - Corrupted sample similarity: 0.5316 |
|
|
2025-07-10 20:04:17,858 - INFO - Similarity gap (clean - corrupt): 0.4297 |
|
|
2025-07-10 20:04:17,957 - INFO - Epoch 18/30 - Train Loss: 0.6364, Val Loss: 0.6189, Clean Sim: 0.9612, Corrupt Sim: 0.5316, Gap: 0.4297, Time: 972.62s |
|
|
2025-07-10 20:06:55,914 - INFO - Epoch 18 Validation Alignment: Pos=0.151, Neg=0.086, Gap=0.065 |
|
|
2025-07-10 20:08:05,318 - INFO - log_σ² gradient: -0.475905 |
|
|
2025-07-10 20:08:05,389 - INFO - Optimizer step 1: log_σ²=0.116480, weight=0.890048 |
|
|
2025-07-10 20:08:29,453 - INFO - log_σ² gradient: -0.484003 |
|
|
2025-07-10 20:08:29,521 - INFO - Optimizer step 2: log_σ²=0.116866, weight=0.889704 |
|
|
2025-07-10 20:08:54,333 - INFO - log_σ² gradient: -0.475975 |
|
|
2025-07-10 20:08:54,412 - INFO - Optimizer step 3: log_σ²=0.117255, weight=0.889359 |
|
|
2025-07-10 20:09:18,968 - INFO - log_σ² gradient: -0.477796 |
|
|
2025-07-10 20:09:19,042 - INFO - Optimizer step 4: log_σ²=0.117645, weight=0.889011 |
|
|
2025-07-10 20:09:43,845 - INFO - log_σ² gradient: -0.485441 |
|
|
2025-07-10 20:09:43,919 - INFO - Optimizer step 5: log_σ²=0.118039, weight=0.888662 |
|
|
2025-07-10 20:10:08,127 - INFO - log_σ² gradient: -0.474462 |
|
|
2025-07-10 20:10:08,198 - INFO - Optimizer step 6: log_σ²=0.118433, weight=0.888311 |
|
|
2025-07-10 20:10:33,525 - INFO - log_σ² gradient: -0.473434 |
|
|
2025-07-10 20:10:33,606 - INFO - Optimizer step 7: log_σ²=0.118830, weight=0.887959 |
|
|
2025-07-10 20:10:56,151 - INFO - log_σ² gradient: -0.480096 |
|
|
2025-07-10 20:10:56,223 - INFO - Optimizer step 8: log_σ²=0.119228, weight=0.887606 |
|
|
2025-07-10 20:11:21,660 - INFO - log_σ² gradient: -0.474620 |
|
|
2025-07-10 20:11:21,742 - INFO - Optimizer step 9: log_σ²=0.119627, weight=0.887251 |
|
|
2025-07-10 20:11:47,036 - INFO - log_σ² gradient: -0.471927 |
|
|
2025-07-10 20:11:47,112 - INFO - Optimizer step 10: log_σ²=0.120027, weight=0.886896 |
|
|
2025-07-10 20:12:10,079 - INFO - log_σ² gradient: -0.477728 |
|
|
2025-07-10 20:12:10,155 - INFO - Optimizer step 11: log_σ²=0.120429, weight=0.886540 |
|
|
2025-07-10 20:12:33,955 - INFO - log_σ² gradient: -0.478776 |
|
|
2025-07-10 20:12:34,031 - INFO - Optimizer step 12: log_σ²=0.120833, weight=0.886182 |
|
|
2025-07-10 20:12:57,845 - INFO - log_σ² gradient: -0.474007 |
|
|
2025-07-10 20:12:57,919 - INFO - Optimizer step 13: log_σ²=0.121237, weight=0.885824 |
|
|
2025-07-10 20:13:21,686 - INFO - log_σ² gradient: -0.474185 |
|
|
2025-07-10 20:13:21,765 - INFO - Optimizer step 14: log_σ²=0.121643, weight=0.885465 |
|
|
2025-07-10 20:13:46,807 - INFO - log_σ² gradient: -0.474659 |
|
|
2025-07-10 20:13:46,881 - INFO - Optimizer step 15: log_σ²=0.122049, weight=0.885105 |
|
|
2025-07-10 20:14:11,190 - INFO - log_σ² gradient: -0.476732 |
|
|
2025-07-10 20:14:11,255 - INFO - Optimizer step 16: log_σ²=0.122457, weight=0.884744 |
|
|
2025-07-10 20:14:34,143 - INFO - log_σ² gradient: -0.475776 |
|
|
2025-07-10 20:14:34,210 - INFO - Optimizer step 17: log_σ²=0.122866, weight=0.884382 |
|
|
2025-07-10 20:14:56,957 - INFO - log_σ² gradient: -0.484050 |
|
|
2025-07-10 20:14:57,028 - INFO - Optimizer step 18: log_σ²=0.123276, weight=0.884019 |
|
|
2025-07-10 20:15:21,248 - INFO - log_σ² gradient: -0.490429 |
|
|
2025-07-10 20:15:21,316 - INFO - Optimizer step 19: log_σ²=0.123689, weight=0.883654 |
|
|
2025-07-10 20:15:45,862 - INFO - log_σ² gradient: -0.467047 |
|
|
2025-07-10 20:15:45,933 - INFO - Optimizer step 20: log_σ²=0.124102, weight=0.883290 |
|
|
2025-07-10 20:16:09,949 - INFO - log_σ² gradient: -0.478839 |
|
|
2025-07-10 20:16:10,017 - INFO - Optimizer step 21: log_σ²=0.124516, weight=0.882924 |
|
|
2025-07-10 20:16:34,959 - INFO - log_σ² gradient: -0.476612 |
|
|
2025-07-10 20:16:35,031 - INFO - Optimizer step 22: log_σ²=0.124931, weight=0.882558 |
|
|
2025-07-10 20:16:59,299 - INFO - log_σ² gradient: -0.477970 |
|
|
2025-07-10 20:16:59,372 - INFO - Optimizer step 23: log_σ²=0.125347, weight=0.882190 |
|
|
2025-07-10 20:17:24,775 - INFO - log_σ² gradient: -0.472383 |
|
|
2025-07-10 20:17:24,842 - INFO - Optimizer step 24: log_σ²=0.125764, weight=0.881823 |
|
|
2025-07-10 20:17:48,198 - INFO - log_σ² gradient: -0.470216 |
|
|
2025-07-10 20:17:48,274 - INFO - Optimizer step 25: log_σ²=0.126181, weight=0.881455 |
|
|
2025-07-10 20:18:11,506 - INFO - log_σ² gradient: -0.480273 |
|
|
2025-07-10 20:18:11,569 - INFO - Optimizer step 26: log_σ²=0.126599, weight=0.881087 |
|
|
2025-07-10 20:18:36,739 - INFO - log_σ² gradient: -0.467456 |
|
|
2025-07-10 20:18:36,810 - INFO - Optimizer step 27: log_σ²=0.127017, weight=0.880718 |
|
|
2025-07-10 20:19:00,412 - INFO - log_σ² gradient: -0.467239 |
|
|
2025-07-10 20:19:00,490 - INFO - Optimizer step 28: log_σ²=0.127436, weight=0.880350 |
|
|
2025-07-10 20:19:24,759 - INFO - log_σ² gradient: -0.470192 |
|
|
2025-07-10 20:19:24,825 - INFO - Optimizer step 29: log_σ²=0.127855, weight=0.879981 |
|
|
2025-07-10 20:19:46,910 - INFO - log_σ² gradient: -0.481645 |
|
|
2025-07-10 20:19:46,974 - INFO - Optimizer step 30: log_σ²=0.128276, weight=0.879611 |
|
|
2025-07-10 20:19:57,222 - INFO - log_σ² gradient: -0.228506 |
|
|
2025-07-10 20:19:57,301 - INFO - Optimizer step 31: log_σ²=0.128675, weight=0.879259 |
|
|
2025-07-10 20:19:57,457 - INFO - Epoch 19: Total optimizer steps: 31 |
|
|
2025-07-10 20:23:14,790 - INFO - Validation metrics: |
|
|
2025-07-10 20:23:14,790 - INFO - Loss: 0.5889 |
|
|
2025-07-10 20:23:14,790 - INFO - Average similarity: 0.8652 |
|
|
2025-07-10 20:23:14,790 - INFO - Median similarity: 0.9890 |
|
|
2025-07-10 20:23:14,790 - INFO - Clean sample similarity: 0.8652 |
|
|
2025-07-10 20:23:14,790 - INFO - Corrupted sample similarity: 0.3400 |
|
|
2025-07-10 20:23:14,790 - INFO - Similarity gap (clean - corrupt): 0.5253 |
|
|
2025-07-10 20:23:14,904 - INFO - Epoch 19/30 - Train Loss: 0.6202, Val Loss: 0.5889, Clean Sim: 0.8652, Corrupt Sim: 0.3400, Gap: 0.5253, Time: 978.99s |
|
|
2025-07-10 20:23:14,904 - INFO - New best validation loss: 0.5889 |
|
|
2025-07-10 20:23:20,961 - INFO - New best similarity gap: 0.5253 |
|
|
2025-07-10 20:24:37,566 - INFO - log_σ² gradient: -0.468961 |
|
|
2025-07-10 20:24:37,644 - INFO - Optimizer step 1: log_σ²=0.129078, weight=0.878906 |
|
|
2025-07-10 20:24:59,907 - INFO - log_σ² gradient: -0.468636 |
|
|
2025-07-10 20:24:59,978 - INFO - Optimizer step 2: log_σ²=0.129482, weight=0.878550 |
|
|
2025-07-10 20:25:24,250 - INFO - log_σ² gradient: -0.479279 |
|
|
2025-07-10 20:25:24,329 - INFO - Optimizer step 3: log_σ²=0.129890, weight=0.878192 |
|
|
2025-07-10 20:25:48,439 - INFO - log_σ² gradient: -0.478745 |
|
|
2025-07-10 20:25:48,510 - INFO - Optimizer step 4: log_σ²=0.130300, weight=0.877832 |
|
|
2025-07-10 20:26:13,323 - INFO - log_σ² gradient: -0.471934 |
|
|
2025-07-10 20:26:13,394 - INFO - Optimizer step 5: log_σ²=0.130713, weight=0.877470 |
|
|
2025-07-10 20:26:38,091 - INFO - log_σ² gradient: -0.471823 |
|
|
2025-07-10 20:26:38,162 - INFO - Optimizer step 6: log_σ²=0.131127, weight=0.877106 |
|
|
2025-07-10 20:27:01,492 - INFO - log_σ² gradient: -0.466067 |
|
|
2025-07-10 20:27:01,567 - INFO - Optimizer step 7: log_σ²=0.131543, weight=0.876742 |
|
|
2025-07-10 20:27:25,022 - INFO - log_σ² gradient: -0.471086 |
|
|
2025-07-10 20:27:25,093 - INFO - Optimizer step 8: log_σ²=0.131960, weight=0.876376 |
|
|
2025-07-10 20:27:49,902 - INFO - log_σ² gradient: -0.473326 |
|
|
2025-07-10 20:27:49,972 - INFO - Optimizer step 9: log_σ²=0.132379, weight=0.876009 |
|
|
2025-07-10 20:28:14,329 - INFO - log_σ² gradient: -0.470796 |
|
|
2025-07-10 20:28:14,405 - INFO - Optimizer step 10: log_σ²=0.132800, weight=0.875640 |
|
|
2025-07-10 20:28:36,492 - INFO - log_σ² gradient: -0.482196 |
|
|
2025-07-10 20:28:36,560 - INFO - Optimizer step 11: log_σ²=0.133223, weight=0.875270 |
|
|
2025-07-10 20:29:01,661 - INFO - log_σ² gradient: -0.474403 |
|
|
2025-07-10 20:29:01,732 - INFO - Optimizer step 12: log_σ²=0.133648, weight=0.874898 |
|
|
2025-07-10 20:29:26,214 - INFO - log_σ² gradient: -0.466456 |
|
|
2025-07-10 20:29:26,287 - INFO - Optimizer step 13: log_σ²=0.134073, weight=0.874526 |
|
|
2025-07-10 20:29:53,166 - INFO - log_σ² gradient: -0.473468 |
|
|
2025-07-10 20:29:53,245 - INFO - Optimizer step 14: log_σ²=0.134500, weight=0.874153 |
|
|
2025-07-10 20:30:19,072 - INFO - log_σ² gradient: -0.470650 |
|
|
2025-07-10 20:30:19,146 - INFO - Optimizer step 15: log_σ²=0.134928, weight=0.873779 |
|
|
2025-07-10 20:30:43,720 - INFO - log_σ² gradient: -0.459560 |
|
|
2025-07-10 20:30:43,796 - INFO - Optimizer step 16: log_σ²=0.135355, weight=0.873405 |
|
|
2025-07-10 20:31:09,610 - INFO - log_σ² gradient: -0.467218 |
|
|
2025-07-10 20:31:09,683 - INFO - Optimizer step 17: log_σ²=0.135784, weight=0.873031 |
|
|
2025-07-10 20:31:32,833 - INFO - log_σ² gradient: -0.470880 |
|
|
2025-07-10 20:31:32,905 - INFO - Optimizer step 18: log_σ²=0.136214, weight=0.872656 |
|
|
2025-07-10 20:31:55,226 - INFO - log_σ² gradient: -0.473222 |
|
|
2025-07-10 20:31:55,298 - INFO - Optimizer step 19: log_σ²=0.136645, weight=0.872279 |
|
|
2025-07-10 20:32:18,204 - INFO - log_σ² gradient: -0.468703 |
|
|
2025-07-10 20:32:18,283 - INFO - Optimizer step 20: log_σ²=0.137078, weight=0.871903 |
|
|
2025-07-10 20:32:42,524 - INFO - log_σ² gradient: -0.467667 |
|
|
2025-07-10 20:32:42,596 - INFO - Optimizer step 21: log_σ²=0.137510, weight=0.871525 |
|
|
2025-07-10 20:33:06,604 - INFO - log_σ² gradient: -0.466641 |
|
|
2025-07-10 20:33:06,676 - INFO - Optimizer step 22: log_σ²=0.137944, weight=0.871147 |
|
|
2025-07-10 20:33:30,981 - INFO - log_σ² gradient: -0.469925 |
|
|
2025-07-10 20:33:31,052 - INFO - Optimizer step 23: log_σ²=0.138379, weight=0.870769 |
|
|
2025-07-10 20:33:55,918 - INFO - log_σ² gradient: -0.477843 |
|
|
2025-07-10 20:33:55,992 - INFO - Optimizer step 24: log_σ²=0.138815, weight=0.870389 |
|
|
2025-07-10 20:34:18,912 - INFO - log_σ² gradient: -0.460365 |
|
|
2025-07-10 20:34:18,978 - INFO - Optimizer step 25: log_σ²=0.139251, weight=0.870009 |
|
|
2025-07-10 20:34:41,713 - INFO - log_σ² gradient: -0.468959 |
|
|
2025-07-10 20:34:41,781 - INFO - Optimizer step 26: log_σ²=0.139689, weight=0.869629 |
|
|
2025-07-10 20:35:05,261 - INFO - log_σ² gradient: -0.474560 |
|
|
2025-07-10 20:35:05,337 - INFO - Optimizer step 27: log_σ²=0.140127, weight=0.869248 |
|
|
2025-07-10 20:35:29,157 - INFO - log_σ² gradient: -0.477944 |
|
|
2025-07-10 20:35:29,229 - INFO - Optimizer step 28: log_σ²=0.140567, weight=0.868865 |
|
|
2025-07-10 20:35:53,718 - INFO - log_σ² gradient: -0.470703 |
|
|
2025-07-10 20:35:53,786 - INFO - Optimizer step 29: log_σ²=0.141008, weight=0.868482 |
|
|
2025-07-10 20:36:15,971 - INFO - log_σ² gradient: -0.470436 |
|
|
2025-07-10 20:36:16,039 - INFO - Optimizer step 30: log_σ²=0.141450, weight=0.868098 |
|
|
2025-07-10 20:36:25,966 - INFO - log_σ² gradient: -0.218151 |
|
|
2025-07-10 20:36:26,031 - INFO - Optimizer step 31: log_σ²=0.141870, weight=0.867734 |
|
|
2025-07-10 20:36:26,192 - INFO - Epoch 20: Total optimizer steps: 31 |
|
|
2025-07-10 20:39:42,578 - INFO - Validation metrics: |
|
|
2025-07-10 20:39:42,579 - INFO - Loss: 0.5749 |
|
|
2025-07-10 20:39:42,579 - INFO - Average similarity: 0.8910 |
|
|
2025-07-10 20:39:42,579 - INFO - Median similarity: 0.9922 |
|
|
2025-07-10 20:39:42,579 - INFO - Clean sample similarity: 0.8910 |
|
|
2025-07-10 20:39:42,579 - INFO - Corrupted sample similarity: 0.3512 |
|
|
2025-07-10 20:39:42,579 - INFO - Similarity gap (clean - corrupt): 0.5398 |
|
|
2025-07-10 20:39:42,700 - INFO - Epoch 20/30 - Train Loss: 0.6085, Val Loss: 0.5749, Clean Sim: 0.8910, Corrupt Sim: 0.3512, Gap: 0.5398, Time: 974.82s |
|
|
2025-07-10 20:39:42,700 - INFO - New best validation loss: 0.5749 |
|
|
2025-07-10 20:39:48,686 - INFO - New best similarity gap: 0.5398 |
|
|
2025-07-10 20:42:33,850 - INFO - Epoch 20 Validation Alignment: Pos=0.146, Neg=0.075, Gap=0.072 |
|
|
2025-07-10 20:43:40,899 - INFO - log_σ² gradient: -0.460704 |
|
|
2025-07-10 20:43:40,967 - INFO - Optimizer step 1: log_σ²=0.142291, weight=0.867368 |
|
|
2025-07-10 20:44:03,845 - INFO - log_σ² gradient: -0.463430 |
|
|
2025-07-10 20:44:03,917 - INFO - Optimizer step 2: log_σ²=0.142715, weight=0.867001 |
|
|
2025-07-10 20:44:28,467 - INFO - log_σ² gradient: -0.457900 |
|
|
2025-07-10 20:44:28,544 - INFO - Optimizer step 3: log_σ²=0.143141, weight=0.866632 |
|
|
2025-07-10 20:44:53,182 - INFO - log_σ² gradient: -0.461259 |
|
|
2025-07-10 20:44:53,260 - INFO - Optimizer step 4: log_σ²=0.143569, weight=0.866261 |
|
|
2025-07-10 20:45:16,473 - INFO - log_σ² gradient: -0.464551 |
|
|
2025-07-10 20:45:16,537 - INFO - Optimizer step 5: log_σ²=0.143999, weight=0.865889 |
|
|
2025-07-10 20:45:40,168 - INFO - log_σ² gradient: -0.468401 |
|
|
2025-07-10 20:45:40,244 - INFO - Optimizer step 6: log_σ²=0.144431, weight=0.865515 |
|
|
2025-07-10 20:46:04,260 - INFO - log_σ² gradient: -0.464272 |
|
|
2025-07-10 20:46:04,324 - INFO - Optimizer step 7: log_σ²=0.144865, weight=0.865139 |
|
|
2025-07-10 20:46:29,443 - INFO - log_σ² gradient: -0.451838 |
|
|
2025-07-10 20:46:29,522 - INFO - Optimizer step 8: log_σ²=0.145299, weight=0.864763 |
|
|
2025-07-10 20:46:54,998 - INFO - log_σ² gradient: -0.461550 |
|
|
2025-07-10 20:46:55,070 - INFO - Optimizer step 9: log_σ²=0.145736, weight=0.864386 |
|
|
2025-07-10 20:47:19,142 - INFO - log_σ² gradient: -0.469158 |
|
|
2025-07-10 20:47:19,214 - INFO - Optimizer step 10: log_σ²=0.146174, weight=0.864008 |
|
|
2025-07-10 20:47:43,562 - INFO - log_σ² gradient: -0.464638 |
|
|
2025-07-10 20:47:43,637 - INFO - Optimizer step 11: log_σ²=0.146614, weight=0.863628 |
|
|
2025-07-10 20:48:08,859 - INFO - log_σ² gradient: -0.463304 |
|
|
2025-07-10 20:48:08,935 - INFO - Optimizer step 12: log_σ²=0.147055, weight=0.863247 |
|
|
2025-07-10 20:48:32,589 - INFO - log_σ² gradient: -0.466994 |
|
|
2025-07-10 20:48:32,667 - INFO - Optimizer step 13: log_σ²=0.147498, weight=0.862864 |
|
|
2025-07-10 20:48:57,942 - INFO - log_σ² gradient: -0.462670 |
|
|
2025-07-10 20:48:58,010 - INFO - Optimizer step 14: log_σ²=0.147942, weight=0.862481 |
|
|
2025-07-10 20:49:23,359 - INFO - log_σ² gradient: -0.466207 |
|
|
2025-07-10 20:49:23,431 - INFO - Optimizer step 15: log_σ²=0.148387, weight=0.862097 |
|
|
2025-07-10 20:49:47,033 - INFO - log_σ² gradient: -0.460918 |
|
|
2025-07-10 20:49:47,107 - INFO - Optimizer step 16: log_σ²=0.148834, weight=0.861712 |
|
|
2025-07-10 20:50:10,662 - INFO - log_σ² gradient: -0.461072 |
|
|
2025-07-10 20:50:10,726 - INFO - Optimizer step 17: log_σ²=0.149281, weight=0.861327 |
|
|
2025-07-10 20:50:34,373 - INFO - log_σ² gradient: -0.465137 |
|
|
2025-07-10 20:50:34,449 - INFO - Optimizer step 18: log_σ²=0.149730, weight=0.860940 |
|
|
2025-07-10 20:50:59,389 - INFO - log_σ² gradient: -0.460244 |
|
|
2025-07-10 20:50:59,472 - INFO - Optimizer step 19: log_σ²=0.150179, weight=0.860554 |
|
|
2025-07-10 20:51:23,560 - INFO - log_σ² gradient: -0.466120 |
|
|
2025-07-10 20:51:23,635 - INFO - Optimizer step 20: log_σ²=0.150630, weight=0.860166 |
|
|
2025-07-10 20:51:46,347 - INFO - log_σ² gradient: -0.459669 |
|
|
2025-07-10 20:51:46,419 - INFO - Optimizer step 21: log_σ²=0.151082, weight=0.859777 |
|
|
2025-07-10 20:52:10,355 - INFO - log_σ² gradient: -0.457066 |
|
|
2025-07-10 20:52:10,433 - INFO - Optimizer step 22: log_σ²=0.151534, weight=0.859389 |
|
|
2025-07-10 20:52:33,713 - INFO - log_σ² gradient: -0.460893 |
|
|
2025-07-10 20:52:33,794 - INFO - Optimizer step 23: log_σ²=0.151987, weight=0.859000 |
|
|
2025-07-10 20:52:58,171 - INFO - log_σ² gradient: -0.463902 |
|
|
2025-07-10 20:52:58,236 - INFO - Optimizer step 24: log_σ²=0.152440, weight=0.858610 |
|
|
2025-07-10 20:53:22,176 - INFO - log_σ² gradient: -0.451257 |
|
|
2025-07-10 20:53:22,243 - INFO - Optimizer step 25: log_σ²=0.152894, weight=0.858221 |
|
|
2025-07-10 20:53:47,195 - INFO - log_σ² gradient: -0.454842 |
|
|
2025-07-10 20:53:47,267 - INFO - Optimizer step 26: log_σ²=0.153348, weight=0.857831 |
|
|
2025-07-10 20:54:10,566 - INFO - log_σ² gradient: -0.469370 |
|
|
2025-07-10 20:54:10,640 - INFO - Optimizer step 27: log_σ²=0.153804, weight=0.857440 |
|
|
2025-07-10 20:54:34,788 - INFO - log_σ² gradient: -0.460745 |
|
|
2025-07-10 20:54:34,859 - INFO - Optimizer step 28: log_σ²=0.154261, weight=0.857048 |
|
|
2025-07-10 20:54:58,203 - INFO - log_σ² gradient: -0.459347 |
|
|
2025-07-10 20:54:58,277 - INFO - Optimizer step 29: log_σ²=0.154718, weight=0.856656 |
|
|
2025-07-10 20:55:20,620 - INFO - log_σ² gradient: -0.456804 |
|
|
2025-07-10 20:55:20,685 - INFO - Optimizer step 30: log_σ²=0.155176, weight=0.856264 |
|
|
2025-07-10 20:55:31,291 - INFO - log_σ² gradient: -0.210761 |
|
|
2025-07-10 20:55:31,358 - INFO - Optimizer step 31: log_σ²=0.155610, weight=0.855893 |
|
|
2025-07-10 20:55:31,534 - INFO - Epoch 21: Total optimizer steps: 31 |
|
|
2025-07-10 20:58:48,125 - INFO - Validation metrics: |
|
|
2025-07-10 20:58:48,125 - INFO - Loss: 0.5733 |
|
|
2025-07-10 20:58:48,125 - INFO - Average similarity: 0.8433 |
|
|
2025-07-10 20:58:48,125 - INFO - Median similarity: 0.9900 |
|
|
2025-07-10 20:58:48,125 - INFO - Clean sample similarity: 0.8433 |
|
|
2025-07-10 20:58:48,125 - INFO - Corrupted sample similarity: 0.3093 |
|
|
2025-07-10 20:58:48,125 - INFO - Similarity gap (clean - corrupt): 0.5341 |
|
|
2025-07-10 20:58:48,224 - INFO - Epoch 21/30 - Train Loss: 0.5983, Val Loss: 0.5733, Clean Sim: 0.8433, Corrupt Sim: 0.3093, Gap: 0.5341, Time: 974.37s |
|
|
2025-07-10 20:58:48,224 - INFO - New best validation loss: 0.5733 |
|
|
2025-07-10 21:00:04,099 - INFO - log_σ² gradient: -0.468757 |
|
|
2025-07-10 21:00:04,178 - INFO - Optimizer step 1: log_σ²=0.156048, weight=0.855518 |
|
|
2025-07-10 21:00:29,264 - INFO - log_σ² gradient: -0.453064 |
|
|
2025-07-10 21:00:29,341 - INFO - Optimizer step 2: log_σ²=0.156489, weight=0.855141 |
|
|
2025-07-10 21:00:53,749 - INFO - log_σ² gradient: -0.458401 |
|
|
2025-07-10 21:00:53,823 - INFO - Optimizer step 3: log_σ²=0.156932, weight=0.854762 |
|
|
2025-07-10 21:01:19,705 - INFO - log_σ² gradient: -0.460927 |
|
|
2025-07-10 21:01:19,778 - INFO - Optimizer step 4: log_σ²=0.157378, weight=0.854381 |
|
|
2025-07-10 21:01:45,171 - INFO - log_σ² gradient: -0.456388 |
|
|
2025-07-10 21:01:45,247 - INFO - Optimizer step 5: log_σ²=0.157825, weight=0.853999 |
|
|
2025-07-10 21:02:09,734 - INFO - log_σ² gradient: -0.462545 |
|
|
2025-07-10 21:02:09,813 - INFO - Optimizer step 6: log_σ²=0.158276, weight=0.853614 |
|
|
2025-07-10 21:02:33,249 - INFO - log_σ² gradient: -0.449968 |
|
|
2025-07-10 21:02:33,320 - INFO - Optimizer step 7: log_σ²=0.158727, weight=0.853229 |
|
|
2025-07-10 21:02:56,122 - INFO - log_σ² gradient: -0.449460 |
|
|
2025-07-10 21:02:56,201 - INFO - Optimizer step 8: log_σ²=0.159180, weight=0.852843 |
|
|
2025-07-10 21:03:19,507 - INFO - log_σ² gradient: -0.454469 |
|
|
2025-07-10 21:03:19,585 - INFO - Optimizer step 9: log_σ²=0.159634, weight=0.852456 |
|
|
2025-07-10 21:03:43,690 - INFO - log_σ² gradient: -0.454554 |
|
|
2025-07-10 21:03:43,769 - INFO - Optimizer step 10: log_σ²=0.160090, weight=0.852067 |
|
|
2025-07-10 21:04:07,956 - INFO - log_σ² gradient: -0.450983 |
|
|
2025-07-10 21:04:08,036 - INFO - Optimizer step 11: log_σ²=0.160546, weight=0.851678 |
|
|
2025-07-10 21:04:31,703 - INFO - log_σ² gradient: -0.455505 |
|
|
2025-07-10 21:04:31,783 - INFO - Optimizer step 12: log_σ²=0.161004, weight=0.851288 |
|
|
2025-07-10 21:04:54,859 - INFO - log_σ² gradient: -0.455380 |
|
|
2025-07-10 21:04:54,935 - INFO - Optimizer step 13: log_σ²=0.161464, weight=0.850897 |
|
|
2025-07-10 21:05:20,919 - INFO - log_σ² gradient: -0.448476 |
|
|
2025-07-10 21:05:20,986 - INFO - Optimizer step 14: log_σ²=0.161924, weight=0.850506 |
|
|
2025-07-10 21:05:43,040 - INFO - log_σ² gradient: -0.454150 |
|
|
2025-07-10 21:05:43,114 - INFO - Optimizer step 15: log_σ²=0.162385, weight=0.850114 |
|
|
2025-07-10 21:06:07,147 - INFO - log_σ² gradient: -0.450357 |
|
|
2025-07-10 21:06:07,225 - INFO - Optimizer step 16: log_σ²=0.162847, weight=0.849721 |
|
|
2025-07-10 21:06:29,651 - INFO - log_σ² gradient: -0.458423 |
|
|
2025-07-10 21:06:29,722 - INFO - Optimizer step 17: log_σ²=0.163311, weight=0.849327 |
|
|
2025-07-10 21:06:53,984 - INFO - log_σ² gradient: -0.454605 |
|
|
2025-07-10 21:06:54,055 - INFO - Optimizer step 18: log_σ²=0.163776, weight=0.848932 |
|
|
2025-07-10 21:07:15,980 - INFO - log_σ² gradient: -0.442779 |
|
|
2025-07-10 21:07:16,048 - INFO - Optimizer step 19: log_σ²=0.164241, weight=0.848538 |
|
|
2025-07-10 21:07:40,192 - INFO - log_σ² gradient: -0.460882 |
|
|
2025-07-10 21:07:40,268 - INFO - Optimizer step 20: log_σ²=0.164708, weight=0.848142 |
|
|
2025-07-10 21:08:02,700 - INFO - log_σ² gradient: -0.437358 |
|
|
2025-07-10 21:08:02,772 - INFO - Optimizer step 21: log_σ²=0.165174, weight=0.847747 |
|
|
2025-07-10 21:08:27,267 - INFO - log_σ² gradient: -0.459259 |
|
|
2025-07-10 21:08:27,341 - INFO - Optimizer step 22: log_σ²=0.165641, weight=0.847350 |
|
|
2025-07-10 21:08:51,391 - INFO - log_σ² gradient: -0.448040 |
|
|
2025-07-10 21:08:51,467 - INFO - Optimizer step 23: log_σ²=0.166110, weight=0.846953 |
|
|
2025-07-10 21:09:15,350 - INFO - log_σ² gradient: -0.453189 |
|
|
2025-07-10 21:09:15,421 - INFO - Optimizer step 24: log_σ²=0.166579, weight=0.846556 |
|
|
2025-07-10 21:09:38,489 - INFO - log_σ² gradient: -0.449655 |
|
|
2025-07-10 21:09:38,557 - INFO - Optimizer step 25: log_σ²=0.167049, weight=0.846158 |
|
|
2025-07-10 21:10:03,632 - INFO - log_σ² gradient: -0.446735 |
|
|
2025-07-10 21:10:03,703 - INFO - Optimizer step 26: log_σ²=0.167519, weight=0.845760 |
|
|
2025-07-10 21:10:27,709 - INFO - log_σ² gradient: -0.454008 |
|
|
2025-07-10 21:10:27,781 - INFO - Optimizer step 27: log_σ²=0.167991, weight=0.845362 |
|
|
2025-07-10 21:10:52,527 - INFO - log_σ² gradient: -0.447455 |
|
|
2025-07-10 21:10:52,598 - INFO - Optimizer step 28: log_σ²=0.168463, weight=0.844963 |
|
|
2025-07-10 21:11:15,587 - INFO - log_σ² gradient: -0.449462 |
|
|
2025-07-10 21:11:15,663 - INFO - Optimizer step 29: log_σ²=0.168935, weight=0.844563 |
|
|
2025-07-10 21:11:39,172 - INFO - log_σ² gradient: -0.453156 |
|
|
2025-07-10 21:11:39,248 - INFO - Optimizer step 30: log_σ²=0.169409, weight=0.844163 |
|
|
2025-07-10 21:11:50,334 - INFO - log_σ² gradient: -0.205562 |
|
|
2025-07-10 21:11:50,408 - INFO - Optimizer step 31: log_σ²=0.169858, weight=0.843785 |
|
|
2025-07-10 21:11:50,582 - INFO - Epoch 22: Total optimizer steps: 31 |
|
|
2025-07-10 21:15:07,316 - INFO - Validation metrics: |
|
|
2025-07-10 21:15:07,316 - INFO - Loss: 0.5654 |
|
|
2025-07-10 21:15:07,316 - INFO - Average similarity: 0.9276 |
|
|
2025-07-10 21:15:07,316 - INFO - Median similarity: 0.9970 |
|
|
2025-07-10 21:15:07,316 - INFO - Clean sample similarity: 0.9276 |
|
|
2025-07-10 21:15:07,316 - INFO - Corrupted sample similarity: 0.3863 |
|
|
2025-07-10 21:15:07,316 - INFO - Similarity gap (clean - corrupt): 0.5414 |
|
|
2025-07-10 21:15:07,427 - INFO - Epoch 22/30 - Train Loss: 0.5808, Val Loss: 0.5654, Clean Sim: 0.9276, Corrupt Sim: 0.3863, Gap: 0.5414, Time: 972.21s |
|
|
2025-07-10 21:15:07,428 - INFO - New best validation loss: 0.5654 |
|
|
2025-07-10 21:15:13,401 - INFO - New best similarity gap: 0.5414 |
|
|
2025-07-10 21:17:58,417 - INFO - Epoch 22 Validation Alignment: Pos=0.160, Neg=0.085, Gap=0.075 |
|
|
2025-07-10 21:19:08,163 - INFO - log_σ² gradient: -0.456737 |
|
|
2025-07-10 21:19:08,238 - INFO - Optimizer step 1: log_σ²=0.170311, weight=0.843402 |
|
|
2025-07-10 21:19:31,384 - INFO - log_σ² gradient: -0.446817 |
|
|
2025-07-10 21:19:31,456 - INFO - Optimizer step 2: log_σ²=0.170767, weight=0.843018 |
|
|
2025-07-10 21:19:54,332 - INFO - log_σ² gradient: -0.450807 |
|
|
2025-07-10 21:19:54,404 - INFO - Optimizer step 3: log_σ²=0.171225, weight=0.842632 |
|
|
2025-07-10 21:20:16,470 - INFO - log_σ² gradient: -0.446411 |
|
|
2025-07-10 21:20:16,538 - INFO - Optimizer step 4: log_σ²=0.171686, weight=0.842244 |
|
|
2025-07-10 21:20:41,151 - INFO - log_σ² gradient: -0.443222 |
|
|
2025-07-10 21:20:41,223 - INFO - Optimizer step 5: log_σ²=0.172148, weight=0.841854 |
|
|
2025-07-10 21:21:04,886 - INFO - log_σ² gradient: -0.443374 |
|
|
2025-07-10 21:21:04,961 - INFO - Optimizer step 6: log_σ²=0.172612, weight=0.841464 |
|
|
2025-07-10 21:21:30,023 - INFO - log_σ² gradient: -0.442178 |
|
|
2025-07-10 21:21:30,099 - INFO - Optimizer step 7: log_σ²=0.173078, weight=0.841072 |
|
|
2025-07-10 21:21:54,801 - INFO - log_σ² gradient: -0.441047 |
|
|
2025-07-10 21:21:54,867 - INFO - Optimizer step 8: log_σ²=0.173545, weight=0.840679 |
|
|
2025-07-10 21:22:19,228 - INFO - log_σ² gradient: -0.456160 |
|
|
2025-07-10 21:22:19,302 - INFO - Optimizer step 9: log_σ²=0.174015, weight=0.840285 |
|
|
2025-07-10 21:22:42,970 - INFO - log_σ² gradient: -0.443725 |
|
|
2025-07-10 21:22:43,044 - INFO - Optimizer step 10: log_σ²=0.174485, weight=0.839889 |
|
|
2025-07-10 21:23:05,484 - INFO - log_σ² gradient: -0.435196 |
|
|
2025-07-10 21:23:05,555 - INFO - Optimizer step 11: log_σ²=0.174957, weight=0.839493 |
|
|
2025-07-10 21:23:29,870 - INFO - log_σ² gradient: -0.443148 |
|
|
2025-07-10 21:23:29,945 - INFO - Optimizer step 12: log_σ²=0.175429, weight=0.839097 |
|
|
2025-07-10 21:23:54,617 - INFO - log_σ² gradient: -0.445349 |
|
|
2025-07-10 21:23:54,688 - INFO - Optimizer step 13: log_σ²=0.175903, weight=0.838699 |
|
|
2025-07-10 21:24:18,519 - INFO - log_σ² gradient: -0.443459 |
|
|
2025-07-10 21:24:18,586 - INFO - Optimizer step 14: log_σ²=0.176379, weight=0.838301 |
|
|
2025-07-10 21:24:42,880 - INFO - log_σ² gradient: -0.451787 |
|
|
2025-07-10 21:24:42,952 - INFO - Optimizer step 15: log_σ²=0.176856, weight=0.837901 |
|
|
2025-07-10 21:25:08,014 - INFO - log_σ² gradient: -0.440473 |
|
|
2025-07-10 21:25:08,090 - INFO - Optimizer step 16: log_σ²=0.177334, weight=0.837500 |
|
|
2025-07-10 21:25:31,916 - INFO - log_σ² gradient: -0.448305 |
|
|
2025-07-10 21:25:31,990 - INFO - Optimizer step 17: log_σ²=0.177813, weight=0.837099 |
|
|
2025-07-10 21:25:54,883 - INFO - log_σ² gradient: -0.445740 |
|
|
2025-07-10 21:25:54,946 - INFO - Optimizer step 18: log_σ²=0.178294, weight=0.836697 |
|
|
2025-07-10 21:26:18,975 - INFO - log_σ² gradient: -0.443972 |
|
|
2025-07-10 21:26:19,051 - INFO - Optimizer step 19: log_σ²=0.178775, weight=0.836294 |
|
|
2025-07-10 21:26:44,332 - INFO - log_σ² gradient: -0.451246 |
|
|
2025-07-10 21:26:44,406 - INFO - Optimizer step 20: log_σ²=0.179259, weight=0.835890 |
|
|
2025-07-10 21:27:08,626 - INFO - log_σ² gradient: -0.449943 |
|
|
2025-07-10 21:27:08,704 - INFO - Optimizer step 21: log_σ²=0.179744, weight=0.835484 |
|
|
2025-07-10 21:27:31,307 - INFO - log_σ² gradient: -0.446790 |
|
|
2025-07-10 21:27:31,375 - INFO - Optimizer step 22: log_σ²=0.180229, weight=0.835079 |
|
|
2025-07-10 21:27:55,331 - INFO - log_σ² gradient: -0.437465 |
|
|
2025-07-10 21:27:55,405 - INFO - Optimizer step 23: log_σ²=0.180715, weight=0.834673 |
|
|
2025-07-10 21:28:19,496 - INFO - log_σ² gradient: -0.446622 |
|
|
2025-07-10 21:28:19,573 - INFO - Optimizer step 24: log_σ²=0.181203, weight=0.834266 |
|
|
2025-07-10 21:28:43,786 - INFO - log_σ² gradient: -0.449566 |
|
|
2025-07-10 21:28:43,858 - INFO - Optimizer step 25: log_σ²=0.181691, weight=0.833859 |
|
|
2025-07-10 21:29:08,134 - INFO - log_σ² gradient: -0.444855 |
|
|
2025-07-10 21:29:08,200 - INFO - Optimizer step 26: log_σ²=0.182181, weight=0.833451 |
|
|
2025-07-10 21:29:31,844 - INFO - log_σ² gradient: -0.441570 |
|
|
2025-07-10 21:29:31,916 - INFO - Optimizer step 27: log_σ²=0.182670, weight=0.833043 |
|
|
2025-07-10 21:29:56,538 - INFO - log_σ² gradient: -0.444178 |
|
|
2025-07-10 21:29:56,604 - INFO - Optimizer step 28: log_σ²=0.183161, weight=0.832634 |
|
|
2025-07-10 21:30:20,383 - INFO - log_σ² gradient: -0.449660 |
|
|
2025-07-10 21:30:20,462 - INFO - Optimizer step 29: log_σ²=0.183653, weight=0.832224 |
|
|
2025-07-10 21:30:45,068 - INFO - log_σ² gradient: -0.440187 |
|
|
2025-07-10 21:30:45,141 - INFO - Optimizer step 30: log_σ²=0.184146, weight=0.831815 |
|
|
2025-07-10 21:30:55,860 - INFO - log_σ² gradient: -0.204320 |
|
|
2025-07-10 21:30:55,935 - INFO - Optimizer step 31: log_σ²=0.184612, weight=0.831427 |
|
|
2025-07-10 21:30:56,112 - INFO - Epoch 23: Total optimizer steps: 31 |
|
|
2025-07-10 21:34:12,776 - INFO - Validation metrics: |
|
|
2025-07-10 21:34:12,776 - INFO - Loss: 0.5627 |
|
|
2025-07-10 21:34:12,776 - INFO - Average similarity: 0.9523 |
|
|
2025-07-10 21:34:12,776 - INFO - Median similarity: 0.9992 |
|
|
2025-07-10 21:34:12,777 - INFO - Clean sample similarity: 0.9523 |
|
|
2025-07-10 21:34:12,777 - INFO - Corrupted sample similarity: 0.4336 |
|
|
2025-07-10 21:34:12,777 - INFO - Similarity gap (clean - corrupt): 0.5187 |
|
|
2025-07-10 21:34:12,874 - INFO - Epoch 23/30 - Train Loss: 0.5731, Val Loss: 0.5627, Clean Sim: 0.9523, Corrupt Sim: 0.4336, Gap: 0.5187, Time: 974.46s |
|
|
2025-07-10 21:34:12,874 - INFO - New best validation loss: 0.5627 |
|
|
2025-07-10 21:35:27,495 - INFO - log_σ² gradient: -0.447303 |
|
|
2025-07-10 21:35:27,567 - INFO - Optimizer step 1: log_σ²=0.185083, weight=0.831035 |
|
|
2025-07-10 21:35:51,965 - INFO - log_σ² gradient: -0.437356 |
|
|
2025-07-10 21:35:52,039 - INFO - Optimizer step 2: log_σ²=0.185556, weight=0.830643 |
|
|
2025-07-10 21:36:17,272 - INFO - log_σ² gradient: -0.440588 |
|
|
2025-07-10 21:36:17,352 - INFO - Optimizer step 3: log_σ²=0.186031, weight=0.830248 |
|
|
2025-07-10 21:36:41,785 - INFO - log_σ² gradient: -0.443758 |
|
|
2025-07-10 21:36:41,861 - INFO - Optimizer step 4: log_σ²=0.186509, weight=0.829851 |
|
|
2025-07-10 21:37:07,466 - INFO - log_σ² gradient: -0.440368 |
|
|
2025-07-10 21:37:07,546 - INFO - Optimizer step 5: log_σ²=0.186990, weight=0.829452 |
|
|
2025-07-10 21:37:31,580 - INFO - log_σ² gradient: -0.431564 |
|
|
2025-07-10 21:37:31,656 - INFO - Optimizer step 6: log_σ²=0.187471, weight=0.829053 |
|
|
2025-07-10 21:37:55,623 - INFO - log_σ² gradient: -0.435089 |
|
|
2025-07-10 21:37:55,691 - INFO - Optimizer step 7: log_σ²=0.187954, weight=0.828653 |
|
|
2025-07-10 21:38:19,020 - INFO - log_σ² gradient: -0.436996 |
|
|
2025-07-10 21:38:19,095 - INFO - Optimizer step 8: log_σ²=0.188439, weight=0.828251 |
|
|
2025-07-10 21:38:42,366 - INFO - log_σ² gradient: -0.440764 |
|
|
2025-07-10 21:38:42,432 - INFO - Optimizer step 9: log_σ²=0.188925, weight=0.827848 |
|
|
2025-07-10 21:39:06,236 - INFO - log_σ² gradient: -0.444360 |
|
|
2025-07-10 21:39:06,316 - INFO - Optimizer step 10: log_σ²=0.189414, weight=0.827444 |
|
|
2025-07-10 21:39:31,380 - INFO - log_σ² gradient: -0.446680 |
|
|
2025-07-10 21:39:31,458 - INFO - Optimizer step 11: log_σ²=0.189905, weight=0.827038 |
|
|
2025-07-10 21:39:54,556 - INFO - log_σ² gradient: -0.439355 |
|
|
2025-07-10 21:39:54,634 - INFO - Optimizer step 12: log_σ²=0.190398, weight=0.826630 |
|
|
2025-07-10 21:40:19,325 - INFO - log_σ² gradient: -0.440676 |
|
|
2025-07-10 21:40:19,401 - INFO - Optimizer step 13: log_σ²=0.190891, weight=0.826222 |
|
|
2025-07-10 21:40:42,959 - INFO - log_σ² gradient: -0.428243 |
|
|
2025-07-10 21:40:43,032 - INFO - Optimizer step 14: log_σ²=0.191385, weight=0.825814 |
|
|
2025-07-10 21:41:06,507 - INFO - log_σ² gradient: -0.434136 |
|
|
2025-07-10 21:41:06,578 - INFO - Optimizer step 15: log_σ²=0.191880, weight=0.825406 |
|
|
2025-07-10 21:41:30,399 - INFO - log_σ² gradient: -0.439460 |
|
|
2025-07-10 21:41:30,475 - INFO - Optimizer step 16: log_σ²=0.192376, weight=0.824997 |
|
|
2025-07-10 21:41:55,246 - INFO - log_σ² gradient: -0.438002 |
|
|
2025-07-10 21:41:55,317 - INFO - Optimizer step 17: log_σ²=0.192873, weight=0.824587 |
|
|
2025-07-10 21:42:17,861 - INFO - log_σ² gradient: -0.435170 |
|
|
2025-07-10 21:42:17,935 - INFO - Optimizer step 18: log_σ²=0.193371, weight=0.824176 |
|
|
2025-07-10 21:42:41,758 - INFO - log_σ² gradient: -0.434703 |
|
|
2025-07-10 21:42:41,838 - INFO - Optimizer step 19: log_σ²=0.193870, weight=0.823765 |
|
|
2025-07-10 21:43:03,896 - INFO - log_σ² gradient: -0.441669 |
|
|
2025-07-10 21:43:03,969 - INFO - Optimizer step 20: log_σ²=0.194370, weight=0.823353 |
|
|
2025-07-10 21:43:27,731 - INFO - log_σ² gradient: -0.437104 |
|
|
2025-07-10 21:43:27,799 - INFO - Optimizer step 21: log_σ²=0.194871, weight=0.822941 |
|
|
2025-07-10 21:43:51,402 - INFO - log_σ² gradient: -0.433859 |
|
|
2025-07-10 21:43:51,473 - INFO - Optimizer step 22: log_σ²=0.195373, weight=0.822528 |
|
|
2025-07-10 21:44:16,023 - INFO - log_σ² gradient: -0.425901 |
|
|
2025-07-10 21:44:16,099 - INFO - Optimizer step 23: log_σ²=0.195874, weight=0.822116 |
|
|
2025-07-10 21:44:40,705 - INFO - log_σ² gradient: -0.429111 |
|
|
2025-07-10 21:44:40,777 - INFO - Optimizer step 24: log_σ²=0.196376, weight=0.821704 |
|
|
2025-07-10 21:45:04,236 - INFO - log_σ² gradient: -0.433855 |
|
|
2025-07-10 21:45:04,315 - INFO - Optimizer step 25: log_σ²=0.196878, weight=0.821291 |
|
|
2025-07-10 21:45:27,403 - INFO - log_σ² gradient: -0.432852 |
|
|
2025-07-10 21:45:27,469 - INFO - Optimizer step 26: log_σ²=0.197381, weight=0.820878 |
|
|
2025-07-10 21:45:51,401 - INFO - log_σ² gradient: -0.434980 |
|
|
2025-07-10 21:45:51,467 - INFO - Optimizer step 27: log_σ²=0.197885, weight=0.820464 |
|
|
2025-07-10 21:46:15,726 - INFO - log_σ² gradient: -0.435032 |
|
|
2025-07-10 21:46:15,798 - INFO - Optimizer step 28: log_σ²=0.198389, weight=0.820050 |
|
|
2025-07-10 21:46:38,916 - INFO - log_σ² gradient: -0.434975 |
|
|
2025-07-10 21:46:38,988 - INFO - Optimizer step 29: log_σ²=0.198895, weight=0.819636 |
|
|
2025-07-10 21:47:03,319 - INFO - log_σ² gradient: -0.429731 |
|
|
2025-07-10 21:47:03,398 - INFO - Optimizer step 30: log_σ²=0.199401, weight=0.819221 |
|
|
2025-07-10 21:47:13,652 - INFO - log_σ² gradient: -0.207213 |
|
|
2025-07-10 21:47:13,716 - INFO - Optimizer step 31: log_σ²=0.199881, weight=0.818828 |
|
|
2025-07-10 21:47:13,871 - INFO - Epoch 24: Total optimizer steps: 31 |
|
|
2025-07-10 21:50:31,009 - INFO - Validation metrics: |
|
|
2025-07-10 21:50:31,009 - INFO - Loss: 0.5518 |
|
|
2025-07-10 21:50:31,009 - INFO - Average similarity: 0.9326 |
|
|
2025-07-10 21:50:31,009 - INFO - Median similarity: 0.9972 |
|
|
2025-07-10 21:50:31,009 - INFO - Clean sample similarity: 0.9326 |
|
|
2025-07-10 21:50:31,009 - INFO - Corrupted sample similarity: 0.4124 |
|
|
2025-07-10 21:50:31,009 - INFO - Similarity gap (clean - corrupt): 0.5202 |
|
|
2025-07-10 21:50:31,118 - INFO - Epoch 24/30 - Train Loss: 0.5619, Val Loss: 0.5518, Clean Sim: 0.9326, Corrupt Sim: 0.4124, Gap: 0.5202, Time: 972.07s |
|
|
2025-07-10 21:50:31,118 - INFO - New best validation loss: 0.5518 |
|
|
2025-07-10 21:53:15,616 - INFO - Epoch 24 Validation Alignment: Pos=0.174, Neg=0.089, Gap=0.085 |
|
|
2025-07-10 21:54:24,420 - INFO - log_σ² gradient: -0.442240 |
|
|
2025-07-10 21:54:24,500 - INFO - Optimizer step 1: log_σ²=0.200366, weight=0.818431 |
|
|
2025-07-10 21:54:48,527 - INFO - log_σ² gradient: -0.441682 |
|
|
2025-07-10 21:54:48,599 - INFO - Optimizer step 2: log_σ²=0.200855, weight=0.818031 |
|
|
2025-07-10 21:55:12,837 - INFO - log_σ² gradient: -0.435871 |
|
|
2025-07-10 21:55:12,918 - INFO - Optimizer step 3: log_σ²=0.201347, weight=0.817629 |
|
|
2025-07-10 21:55:38,553 - INFO - log_σ² gradient: -0.434178 |
|
|
2025-07-10 21:55:38,625 - INFO - Optimizer step 4: log_σ²=0.201841, weight=0.817225 |
|
|
2025-07-10 21:56:04,788 - INFO - log_σ² gradient: -0.437822 |
|
|
2025-07-10 21:56:04,859 - INFO - Optimizer step 5: log_σ²=0.202339, weight=0.816818 |
|
|
2025-07-10 21:56:28,385 - INFO - log_σ² gradient: -0.425139 |
|
|
2025-07-10 21:56:28,456 - INFO - Optimizer step 6: log_σ²=0.202837, weight=0.816411 |
|
|
2025-07-10 21:56:51,487 - INFO - log_σ² gradient: -0.438997 |
|
|
2025-07-10 21:56:51,558 - INFO - Optimizer step 7: log_σ²=0.203338, weight=0.816002 |
|
|
2025-07-10 21:57:15,783 - INFO - log_σ² gradient: -0.427790 |
|
|
2025-07-10 21:57:15,854 - INFO - Optimizer step 8: log_σ²=0.203840, weight=0.815593 |
|
|
2025-07-10 21:57:38,706 - INFO - log_σ² gradient: -0.436095 |
|
|
2025-07-10 21:57:38,779 - INFO - Optimizer step 9: log_σ²=0.204345, weight=0.815181 |
|
|
2025-07-10 21:58:02,518 - INFO - log_σ² gradient: -0.437719 |
|
|
2025-07-10 21:58:02,584 - INFO - Optimizer step 10: log_σ²=0.204852, weight=0.814768 |
|
|
2025-07-10 21:58:26,647 - INFO - log_σ² gradient: -0.434537 |
|
|
2025-07-10 21:58:26,721 - INFO - Optimizer step 11: log_σ²=0.205360, weight=0.814354 |
|
|
2025-07-10 21:58:49,611 - INFO - log_σ² gradient: -0.431969 |
|
|
2025-07-10 21:58:49,683 - INFO - Optimizer step 12: log_σ²=0.205870, weight=0.813939 |
|
|
2025-07-10 21:59:13,152 - INFO - log_σ² gradient: -0.428660 |
|
|
2025-07-10 21:59:13,227 - INFO - Optimizer step 13: log_σ²=0.206381, weight=0.813523 |
|
|
2025-07-10 21:59:37,670 - INFO - log_σ² gradient: -0.432785 |
|
|
2025-07-10 21:59:37,741 - INFO - Optimizer step 14: log_σ²=0.206893, weight=0.813107 |
|
|
2025-07-10 22:00:01,265 - INFO - log_σ² gradient: -0.430052 |
|
|
2025-07-10 22:00:01,339 - INFO - Optimizer step 15: log_σ²=0.207406, weight=0.812690 |
|
|
2025-07-10 22:00:27,152 - INFO - log_σ² gradient: -0.431740 |
|
|
2025-07-10 22:00:27,227 - INFO - Optimizer step 16: log_σ²=0.207920, weight=0.812272 |
|
|
2025-07-10 22:00:52,121 - INFO - log_σ² gradient: -0.431393 |
|
|
2025-07-10 22:00:52,192 - INFO - Optimizer step 17: log_σ²=0.208435, weight=0.811854 |
|
|
2025-07-10 22:01:15,568 - INFO - log_σ² gradient: -0.425337 |
|
|
2025-07-10 22:01:15,637 - INFO - Optimizer step 18: log_σ²=0.208951, weight=0.811435 |
|
|
2025-07-10 22:01:37,912 - INFO - log_σ² gradient: -0.436811 |
|
|
2025-07-10 22:01:37,980 - INFO - Optimizer step 19: log_σ²=0.209468, weight=0.811016 |
|
|
2025-07-10 22:02:03,015 - INFO - log_σ² gradient: -0.433785 |
|
|
2025-07-10 22:02:03,085 - INFO - Optimizer step 20: log_σ²=0.209987, weight=0.810595 |
|
|
2025-07-10 22:02:26,661 - INFO - log_σ² gradient: -0.429193 |
|
|
2025-07-10 22:02:26,737 - INFO - Optimizer step 21: log_σ²=0.210506, weight=0.810174 |
|
|
2025-07-10 22:02:51,548 - INFO - log_σ² gradient: -0.430464 |
|
|
2025-07-10 22:02:51,627 - INFO - Optimizer step 22: log_σ²=0.211026, weight=0.809753 |
|
|
2025-07-10 22:03:15,443 - INFO - log_σ² gradient: -0.433659 |
|
|
2025-07-10 22:03:15,518 - INFO - Optimizer step 23: log_σ²=0.211548, weight=0.809331 |
|
|
2025-07-10 22:03:40,182 - INFO - log_σ² gradient: -0.430474 |
|
|
2025-07-10 22:03:40,261 - INFO - Optimizer step 24: log_σ²=0.212070, weight=0.808908 |
|
|
2025-07-10 22:04:04,150 - INFO - log_σ² gradient: -0.441929 |
|
|
2025-07-10 22:04:04,221 - INFO - Optimizer step 25: log_σ²=0.212594, weight=0.808484 |
|
|
2025-07-10 22:04:28,314 - INFO - log_σ² gradient: -0.429217 |
|
|
2025-07-10 22:04:28,394 - INFO - Optimizer step 26: log_σ²=0.213119, weight=0.808060 |
|
|
2025-07-10 22:04:52,292 - INFO - log_σ² gradient: -0.430236 |
|
|
2025-07-10 22:04:52,366 - INFO - Optimizer step 27: log_σ²=0.213645, weight=0.807635 |
|
|
2025-07-10 22:05:16,665 - INFO - log_σ² gradient: -0.431700 |
|
|
2025-07-10 22:05:16,744 - INFO - Optimizer step 28: log_σ²=0.214171, weight=0.807210 |
|
|
2025-07-10 22:05:40,396 - INFO - log_σ² gradient: -0.431682 |
|
|
2025-07-10 22:05:40,469 - INFO - Optimizer step 29: log_σ²=0.214699, weight=0.806784 |
|
|
2025-07-10 22:06:03,558 - INFO - log_σ² gradient: -0.430751 |
|
|
2025-07-10 22:06:03,637 - INFO - Optimizer step 30: log_σ²=0.215227, weight=0.806358 |
|
|
2025-07-10 22:06:14,415 - INFO - log_σ² gradient: -0.197611 |
|
|
2025-07-10 22:06:14,489 - INFO - Optimizer step 31: log_σ²=0.215728, weight=0.805955 |
|
|
2025-07-10 22:06:14,656 - INFO - Epoch 25: Total optimizer steps: 31 |
|
|
2025-07-10 22:09:32,376 - INFO - Validation metrics: |
|
|
2025-07-10 22:09:32,376 - INFO - Loss: 0.5247 |
|
|
2025-07-10 22:09:32,376 - INFO - Average similarity: 0.8787 |
|
|
2025-07-10 22:09:32,376 - INFO - Median similarity: 0.9927 |
|
|
2025-07-10 22:09:32,376 - INFO - Clean sample similarity: 0.8787 |
|
|
2025-07-10 22:09:32,376 - INFO - Corrupted sample similarity: 0.3013 |
|
|
2025-07-10 22:09:32,376 - INFO - Similarity gap (clean - corrupt): 0.5775 |
|
|
2025-07-10 22:09:32,502 - INFO - Epoch 25/30 - Train Loss: 0.5577, Val Loss: 0.5247, Clean Sim: 0.8787, Corrupt Sim: 0.3013, Gap: 0.5775, Time: 976.89s |
|
|
2025-07-10 22:09:32,502 - INFO - New best validation loss: 0.5247 |
|
|
2025-07-10 22:09:38,535 - INFO - New best similarity gap: 0.5775 |
|
|
2025-07-10 22:10:52,802 - INFO - log_σ² gradient: -0.426334 |
|
|
2025-07-10 22:10:52,877 - INFO - Optimizer step 1: log_σ²=0.216231, weight=0.805549 |
|
|
2025-07-10 22:11:17,192 - INFO - log_σ² gradient: -0.422210 |
|
|
2025-07-10 22:11:17,271 - INFO - Optimizer step 2: log_σ²=0.216737, weight=0.805142 |
|
|
2025-07-10 22:11:41,349 - INFO - log_σ² gradient: -0.422365 |
|
|
2025-07-10 22:11:41,417 - INFO - Optimizer step 3: log_σ²=0.217245, weight=0.804733 |
|
|
2025-07-10 22:12:03,788 - INFO - log_σ² gradient: -0.421572 |
|
|
2025-07-10 22:12:03,855 - INFO - Optimizer step 4: log_σ²=0.217755, weight=0.804322 |
|
|
2025-07-10 22:12:27,156 - INFO - log_σ² gradient: -0.422913 |
|
|
2025-07-10 22:12:27,224 - INFO - Optimizer step 5: log_σ²=0.218267, weight=0.803911 |
|
|
2025-07-10 22:12:51,662 - INFO - log_σ² gradient: -0.425621 |
|
|
2025-07-10 22:12:51,728 - INFO - Optimizer step 6: log_σ²=0.218781, weight=0.803497 |
|
|
2025-07-10 22:13:15,734 - INFO - log_σ² gradient: -0.426530 |
|
|
2025-07-10 22:13:15,802 - INFO - Optimizer step 7: log_σ²=0.219298, weight=0.803083 |
|
|
2025-07-10 22:13:39,010 - INFO - log_σ² gradient: -0.425806 |
|
|
2025-07-10 22:13:39,085 - INFO - Optimizer step 8: log_σ²=0.219816, weight=0.802666 |
|
|
2025-07-10 22:14:01,667 - INFO - log_σ² gradient: -0.432663 |
|
|
2025-07-10 22:14:01,742 - INFO - Optimizer step 9: log_σ²=0.220337, weight=0.802248 |
|
|
2025-07-10 22:14:25,470 - INFO - log_σ² gradient: -0.413550 |
|
|
2025-07-10 22:14:25,546 - INFO - Optimizer step 10: log_σ²=0.220859, weight=0.801830 |
|
|
2025-07-10 22:14:48,886 - INFO - log_σ² gradient: -0.419853 |
|
|
2025-07-10 22:14:48,958 - INFO - Optimizer step 11: log_σ²=0.221381, weight=0.801411 |
|
|
2025-07-10 22:15:13,301 - INFO - log_σ² gradient: -0.425393 |
|
|
2025-07-10 22:15:13,373 - INFO - Optimizer step 12: log_σ²=0.221905, weight=0.800991 |
|
|
2025-07-10 22:15:35,322 - INFO - log_σ² gradient: -0.426195 |
|
|
2025-07-10 22:15:35,396 - INFO - Optimizer step 13: log_σ²=0.222431, weight=0.800571 |
|
|
2025-07-10 22:15:58,620 - INFO - log_σ² gradient: -0.421831 |
|
|
2025-07-10 22:15:58,692 - INFO - Optimizer step 14: log_σ²=0.222957, weight=0.800149 |
|
|
2025-07-10 22:16:22,931 - INFO - log_σ² gradient: -0.414767 |
|
|
2025-07-10 22:16:23,005 - INFO - Optimizer step 15: log_σ²=0.223484, weight=0.799728 |
|
|
2025-07-10 22:16:46,045 - INFO - log_σ² gradient: -0.430506 |
|
|
2025-07-10 22:16:46,117 - INFO - Optimizer step 16: log_σ²=0.224013, weight=0.799305 |
|
|
2025-07-10 22:17:11,357 - INFO - log_σ² gradient: -0.423460 |
|
|
2025-07-10 22:17:11,431 - INFO - Optimizer step 17: log_σ²=0.224543, weight=0.798881 |
|
|
2025-07-10 22:17:34,600 - INFO - log_σ² gradient: -0.422109 |
|
|
2025-07-10 22:17:34,675 - INFO - Optimizer step 18: log_σ²=0.225075, weight=0.798457 |
|
|
2025-07-10 22:17:57,717 - INFO - log_σ² gradient: -0.429039 |
|
|
2025-07-10 22:17:57,783 - INFO - Optimizer step 19: log_σ²=0.225607, weight=0.798031 |
|
|
2025-07-10 22:18:22,795 - INFO - log_σ² gradient: -0.426637 |
|
|
2025-07-10 22:18:22,874 - INFO - Optimizer step 20: log_σ²=0.226142, weight=0.797605 |
|
|
2025-07-10 22:18:46,314 - INFO - log_σ² gradient: -0.422398 |
|
|
2025-07-10 22:18:46,385 - INFO - Optimizer step 21: log_σ²=0.226677, weight=0.797178 |
|
|
2025-07-10 22:19:10,465 - INFO - log_σ² gradient: -0.420785 |
|
|
2025-07-10 22:19:10,536 - INFO - Optimizer step 22: log_σ²=0.227213, weight=0.796751 |
|
|
2025-07-10 22:19:33,904 - INFO - log_σ² gradient: -0.421726 |
|
|
2025-07-10 22:19:33,969 - INFO - Optimizer step 23: log_σ²=0.227749, weight=0.796324 |
|
|
2025-07-10 22:19:57,720 - INFO - log_σ² gradient: -0.421712 |
|
|
2025-07-10 22:19:57,792 - INFO - Optimizer step 24: log_σ²=0.228287, weight=0.795896 |
|
|
2025-07-10 22:20:22,398 - INFO - log_σ² gradient: -0.422968 |
|
|
2025-07-10 22:20:22,461 - INFO - Optimizer step 25: log_σ²=0.228825, weight=0.795468 |
|
|
2025-07-10 22:20:46,070 - INFO - log_σ² gradient: -0.420888 |
|
|
2025-07-10 22:20:46,138 - INFO - Optimizer step 26: log_σ²=0.229364, weight=0.795039 |
|
|
2025-07-10 22:21:10,197 - INFO - log_σ² gradient: -0.416712 |
|
|
2025-07-10 22:21:10,275 - INFO - Optimizer step 27: log_σ²=0.229903, weight=0.794611 |
|
|
2025-07-10 22:21:33,321 - INFO - log_σ² gradient: -0.416728 |
|
|
2025-07-10 22:21:33,389 - INFO - Optimizer step 28: log_σ²=0.230442, weight=0.794182 |
|
|
2025-07-10 22:21:59,186 - INFO - log_σ² gradient: -0.417007 |
|
|
2025-07-10 22:21:59,264 - INFO - Optimizer step 29: log_σ²=0.230982, weight=0.793754 |
|
|
2025-07-10 22:22:22,987 - INFO - log_σ² gradient: -0.413504 |
|
|
2025-07-10 22:22:23,060 - INFO - Optimizer step 30: log_σ²=0.231521, weight=0.793326 |
|
|
2025-07-10 22:22:34,772 - INFO - log_σ² gradient: -0.193922 |
|
|
2025-07-10 22:22:34,850 - INFO - Optimizer step 31: log_σ²=0.232033, weight=0.792920 |
|
|
2025-07-10 22:22:35,046 - INFO - Epoch 26: Total optimizer steps: 31 |
|
|
2025-07-10 22:25:53,226 - INFO - Validation metrics: |
|
|
2025-07-10 22:25:53,226 - INFO - Loss: 0.5252 |
|
|
2025-07-10 22:25:53,226 - INFO - Average similarity: 0.8957 |
|
|
2025-07-10 22:25:53,226 - INFO - Median similarity: 0.9922 |
|
|
2025-07-10 22:25:53,226 - INFO - Clean sample similarity: 0.8957 |
|
|
2025-07-10 22:25:53,226 - INFO - Corrupted sample similarity: 0.3398 |
|
|
2025-07-10 22:25:53,226 - INFO - Similarity gap (clean - corrupt): 0.5559 |
|
|
2025-07-10 22:25:53,333 - INFO - Epoch 26/30 - Train Loss: 0.5422, Val Loss: 0.5252, Clean Sim: 0.8957, Corrupt Sim: 0.3398, Gap: 0.5559, Time: 967.97s |
|
|
2025-07-10 22:28:30,838 - INFO - Epoch 26 Validation Alignment: Pos=0.147, Neg=0.071, Gap=0.076 |
|
|
2025-07-10 22:29:39,297 - INFO - log_σ² gradient: -0.419699 |
|
|
2025-07-10 22:29:39,369 - INFO - Optimizer step 1: log_σ²=0.232548, weight=0.792512 |
|
|
2025-07-10 22:30:03,206 - INFO - log_σ² gradient: -0.423079 |
|
|
2025-07-10 22:30:03,286 - INFO - Optimizer step 2: log_σ²=0.233066, weight=0.792101 |
|
|
2025-07-10 22:30:27,440 - INFO - log_σ² gradient: -0.423107 |
|
|
2025-07-10 22:30:27,506 - INFO - Optimizer step 3: log_σ²=0.233589, weight=0.791687 |
|
|
2025-07-10 22:30:51,530 - INFO - log_σ² gradient: -0.419849 |
|
|
2025-07-10 22:30:51,604 - INFO - Optimizer step 4: log_σ²=0.234114, weight=0.791272 |
|
|
2025-07-10 22:31:16,836 - INFO - log_σ² gradient: -0.424075 |
|
|
2025-07-10 22:31:16,909 - INFO - Optimizer step 5: log_σ²=0.234642, weight=0.790854 |
|
|
2025-07-10 22:31:39,758 - INFO - log_σ² gradient: -0.418975 |
|
|
2025-07-10 22:31:39,829 - INFO - Optimizer step 6: log_σ²=0.235173, weight=0.790434 |
|
|
2025-07-10 22:32:04,635 - INFO - log_σ² gradient: -0.412314 |
|
|
2025-07-10 22:32:04,703 - INFO - Optimizer step 7: log_σ²=0.235705, weight=0.790014 |
|
|
2025-07-10 22:32:30,238 - INFO - log_σ² gradient: -0.427261 |
|
|
2025-07-10 22:32:30,309 - INFO - Optimizer step 8: log_σ²=0.236240, weight=0.789591 |
|
|
2025-07-10 22:32:53,875 - INFO - log_σ² gradient: -0.415846 |
|
|
2025-07-10 22:32:53,950 - INFO - Optimizer step 9: log_σ²=0.236777, weight=0.789168 |
|
|
2025-07-10 22:33:19,473 - INFO - log_σ² gradient: -0.417668 |
|
|
2025-07-10 22:33:19,541 - INFO - Optimizer step 10: log_σ²=0.237315, weight=0.788743 |
|
|
2025-07-10 22:33:41,706 - INFO - log_σ² gradient: -0.414233 |
|
|
2025-07-10 22:33:41,780 - INFO - Optimizer step 11: log_σ²=0.237854, weight=0.788318 |
|
|
2025-07-10 22:34:04,729 - INFO - log_σ² gradient: -0.417848 |
|
|
2025-07-10 22:34:04,800 - INFO - Optimizer step 12: log_σ²=0.238395, weight=0.787891 |
|
|
2025-07-10 22:34:29,668 - INFO - log_σ² gradient: -0.415106 |
|
|
2025-07-10 22:34:29,732 - INFO - Optimizer step 13: log_σ²=0.238937, weight=0.787464 |
|
|
2025-07-10 22:34:53,844 - INFO - log_σ² gradient: -0.408954 |
|
|
2025-07-10 22:34:53,923 - INFO - Optimizer step 14: log_σ²=0.239479, weight=0.787038 |
|
|
2025-07-10 22:35:17,243 - INFO - log_σ² gradient: -0.409812 |
|
|
2025-07-10 22:35:17,315 - INFO - Optimizer step 15: log_σ²=0.240022, weight=0.786611 |
|
|
2025-07-10 22:35:39,607 - INFO - log_σ² gradient: -0.421921 |
|
|
2025-07-10 22:35:39,678 - INFO - Optimizer step 16: log_σ²=0.240567, weight=0.786182 |
|
|
2025-07-10 22:36:03,717 - INFO - log_σ² gradient: -0.411932 |
|
|
2025-07-10 22:36:03,783 - INFO - Optimizer step 17: log_σ²=0.241112, weight=0.785753 |
|
|
2025-07-10 22:36:26,794 - INFO - log_σ² gradient: -0.409461 |
|
|
2025-07-10 22:36:26,856 - INFO - Optimizer step 18: log_σ²=0.241658, weight=0.785325 |
|
|
2025-07-10 22:36:51,115 - INFO - log_σ² gradient: -0.414012 |
|
|
2025-07-10 22:36:51,193 - INFO - Optimizer step 19: log_σ²=0.242205, weight=0.784896 |
|
|
2025-07-10 22:37:14,801 - INFO - log_σ² gradient: -0.415098 |
|
|
2025-07-10 22:37:14,868 - INFO - Optimizer step 20: log_σ²=0.242752, weight=0.784466 |
|
|
2025-07-10 22:37:39,752 - INFO - log_σ² gradient: -0.415714 |
|
|
2025-07-10 22:37:39,822 - INFO - Optimizer step 21: log_σ²=0.243301, weight=0.784035 |
|
|
2025-07-10 22:38:03,563 - INFO - log_σ² gradient: -0.421395 |
|
|
2025-07-10 22:38:03,638 - INFO - Optimizer step 22: log_σ²=0.243852, weight=0.783603 |
|
|
2025-07-10 22:38:26,892 - INFO - log_σ² gradient: -0.415266 |
|
|
2025-07-10 22:38:26,960 - INFO - Optimizer step 23: log_σ²=0.244404, weight=0.783171 |
|
|
2025-07-10 22:38:50,280 - INFO - log_σ² gradient: -0.410478 |
|
|
2025-07-10 22:38:50,359 - INFO - Optimizer step 24: log_σ²=0.244956, weight=0.782739 |
|
|
2025-07-10 22:39:14,917 - INFO - log_σ² gradient: -0.412311 |
|
|
2025-07-10 22:39:14,993 - INFO - Optimizer step 25: log_σ²=0.245509, weight=0.782306 |
|
|
2025-07-10 22:39:40,568 - INFO - log_σ² gradient: -0.415902 |
|
|
2025-07-10 22:39:40,640 - INFO - Optimizer step 26: log_σ²=0.246063, weight=0.781873 |
|
|
2025-07-10 22:40:03,752 - INFO - log_σ² gradient: -0.399741 |
|
|
2025-07-10 22:40:03,826 - INFO - Optimizer step 27: log_σ²=0.246615, weight=0.781441 |
|
|
2025-07-10 22:40:27,799 - INFO - log_σ² gradient: -0.413292 |
|
|
2025-07-10 22:40:27,871 - INFO - Optimizer step 28: log_σ²=0.247169, weight=0.781009 |
|
|
2025-07-10 22:40:52,940 - INFO - log_σ² gradient: -0.417621 |
|
|
2025-07-10 22:40:53,007 - INFO - Optimizer step 29: log_σ²=0.247724, weight=0.780575 |
|
|
2025-07-10 22:41:16,478 - INFO - log_σ² gradient: -0.413058 |
|
|
2025-07-10 22:41:16,549 - INFO - Optimizer step 30: log_σ²=0.248280, weight=0.780141 |
|
|
2025-07-10 22:41:28,642 - INFO - log_σ² gradient: -0.188966 |
|
|
2025-07-10 22:41:28,718 - INFO - Optimizer step 31: log_σ²=0.248807, weight=0.779731 |
|
|
2025-07-10 22:41:28,873 - INFO - Epoch 27: Total optimizer steps: 31 |
|
|
2025-07-10 22:44:47,067 - INFO - Validation metrics: |
|
|
2025-07-10 22:44:47,067 - INFO - Loss: 0.5085 |
|
|
2025-07-10 22:44:47,067 - INFO - Average similarity: 0.8074 |
|
|
2025-07-10 22:44:47,067 - INFO - Median similarity: 0.9717 |
|
|
2025-07-10 22:44:47,067 - INFO - Clean sample similarity: 0.8074 |
|
|
2025-07-10 22:44:47,067 - INFO - Corrupted sample similarity: 0.2591 |
|
|
2025-07-10 22:44:47,067 - INFO - Similarity gap (clean - corrupt): 0.5483 |
|
|
2025-07-10 22:44:47,189 - INFO - Epoch 27/30 - Train Loss: 0.5337, Val Loss: 0.5085, Clean Sim: 0.8074, Corrupt Sim: 0.2591, Gap: 0.5483, Time: 976.35s |
|
|
2025-07-10 22:44:47,189 - INFO - New best validation loss: 0.5085 |
|
|
2025-07-10 22:46:01,561 - INFO - log_σ² gradient: -0.413667 |
|
|
2025-07-10 22:46:01,633 - INFO - Optimizer step 1: log_σ²=0.249337, weight=0.779317 |
|
|
2025-07-10 22:46:26,250 - INFO - log_σ² gradient: -0.414822 |
|
|
2025-07-10 22:46:26,318 - INFO - Optimizer step 2: log_σ²=0.249872, weight=0.778901 |
|
|
2025-07-10 22:46:50,314 - INFO - log_σ² gradient: -0.411625 |
|
|
2025-07-10 22:46:50,387 - INFO - Optimizer step 3: log_σ²=0.250409, weight=0.778482 |
|
|
2025-07-10 22:47:14,253 - INFO - log_σ² gradient: -0.412753 |
|
|
2025-07-10 22:47:14,317 - INFO - Optimizer step 4: log_σ²=0.250949, weight=0.778062 |
|
|
2025-07-10 22:47:38,995 - INFO - log_σ² gradient: -0.414952 |
|
|
2025-07-10 22:47:39,082 - INFO - Optimizer step 5: log_σ²=0.251492, weight=0.777639 |
|
|
2025-07-10 22:48:00,996 - INFO - log_σ² gradient: -0.405456 |
|
|
2025-07-10 22:48:01,066 - INFO - Optimizer step 6: log_σ²=0.252037, weight=0.777216 |
|
|
2025-07-10 22:48:24,953 - INFO - log_σ² gradient: -0.415193 |
|
|
2025-07-10 22:48:25,025 - INFO - Optimizer step 7: log_σ²=0.252585, weight=0.776790 |
|
|
2025-07-10 22:48:46,327 - INFO - log_σ² gradient: -0.403567 |
|
|
2025-07-10 22:48:46,399 - INFO - Optimizer step 8: log_σ²=0.253133, weight=0.776364 |
|
|
2025-07-10 22:49:11,613 - INFO - log_σ² gradient: -0.413557 |
|
|
2025-07-10 22:49:11,687 - INFO - Optimizer step 9: log_σ²=0.253684, weight=0.775937 |
|
|
2025-07-10 22:49:36,078 - INFO - log_σ² gradient: -0.420608 |
|
|
2025-07-10 22:49:36,149 - INFO - Optimizer step 10: log_σ²=0.254238, weight=0.775507 |
|
|
2025-07-10 22:49:59,794 - INFO - log_σ² gradient: -0.404697 |
|
|
2025-07-10 22:49:59,866 - INFO - Optimizer step 11: log_σ²=0.254793, weight=0.775077 |
|
|
2025-07-10 22:50:23,531 - INFO - log_σ² gradient: -0.414637 |
|
|
2025-07-10 22:50:23,613 - INFO - Optimizer step 12: log_σ²=0.255350, weight=0.774645 |
|
|
2025-07-10 22:50:47,454 - INFO - log_σ² gradient: -0.409827 |
|
|
2025-07-10 22:50:47,524 - INFO - Optimizer step 13: log_σ²=0.255909, weight=0.774213 |
|
|
2025-07-10 22:51:12,696 - INFO - log_σ² gradient: -0.409518 |
|
|
2025-07-10 22:51:12,770 - INFO - Optimizer step 14: log_σ²=0.256468, weight=0.773780 |
|
|
2025-07-10 22:51:36,679 - INFO - log_σ² gradient: -0.408953 |
|
|
2025-07-10 22:51:36,755 - INFO - Optimizer step 15: log_σ²=0.257029, weight=0.773346 |
|
|
2025-07-10 22:52:00,545 - INFO - log_σ² gradient: -0.411743 |
|
|
2025-07-10 22:52:00,619 - INFO - Optimizer step 16: log_σ²=0.257591, weight=0.772911 |
|
|
2025-07-10 22:52:24,659 - INFO - log_σ² gradient: -0.409812 |
|
|
2025-07-10 22:52:24,723 - INFO - Optimizer step 17: log_σ²=0.258154, weight=0.772476 |
|
|
2025-07-10 22:52:49,687 - INFO - log_σ² gradient: -0.406604 |
|
|
2025-07-10 22:52:49,763 - INFO - Optimizer step 18: log_σ²=0.258718, weight=0.772040 |
|
|
2025-07-10 22:53:12,344 - INFO - log_σ² gradient: -0.405698 |
|
|
2025-07-10 22:53:12,415 - INFO - Optimizer step 19: log_σ²=0.259283, weight=0.771605 |
|
|
2025-07-10 22:53:37,019 - INFO - log_σ² gradient: -0.410662 |
|
|
2025-07-10 22:53:37,093 - INFO - Optimizer step 20: log_σ²=0.259849, weight=0.771168 |
|
|
2025-07-10 22:54:01,524 - INFO - log_σ² gradient: -0.408365 |
|
|
2025-07-10 22:54:01,602 - INFO - Optimizer step 21: log_σ²=0.260415, weight=0.770732 |
|
|
2025-07-10 22:54:26,778 - INFO - log_σ² gradient: -0.404820 |
|
|
2025-07-10 22:54:26,853 - INFO - Optimizer step 22: log_σ²=0.260982, weight=0.770295 |
|
|
2025-07-10 22:54:49,769 - INFO - log_σ² gradient: -0.401714 |
|
|
2025-07-10 22:54:49,836 - INFO - Optimizer step 23: log_σ²=0.261549, weight=0.769858 |
|
|
2025-07-10 22:55:14,530 - INFO - log_σ² gradient: -0.409264 |
|
|
2025-07-10 22:55:14,604 - INFO - Optimizer step 24: log_σ²=0.262118, weight=0.769421 |
|
|
2025-07-10 22:55:37,537 - INFO - log_σ² gradient: -0.408126 |
|
|
2025-07-10 22:55:37,612 - INFO - Optimizer step 25: log_σ²=0.262687, weight=0.768983 |
|
|
2025-07-10 22:56:02,423 - INFO - log_σ² gradient: -0.399906 |
|
|
2025-07-10 22:56:02,494 - INFO - Optimizer step 26: log_σ²=0.263256, weight=0.768545 |
|
|
2025-07-10 22:56:26,536 - INFO - log_σ² gradient: -0.402494 |
|
|
2025-07-10 22:56:26,600 - INFO - Optimizer step 27: log_σ²=0.263825, weight=0.768108 |
|
|
2025-07-10 22:56:50,711 - INFO - log_σ² gradient: -0.400197 |
|
|
2025-07-10 22:56:50,785 - INFO - Optimizer step 28: log_σ²=0.264394, weight=0.767671 |
|
|
2025-07-10 22:57:14,679 - INFO - log_σ² gradient: -0.405515 |
|
|
2025-07-10 22:57:14,758 - INFO - Optimizer step 29: log_σ²=0.264964, weight=0.767233 |
|
|
2025-07-10 22:57:37,002 - INFO - log_σ² gradient: -0.410242 |
|
|
2025-07-10 22:57:37,080 - INFO - Optimizer step 30: log_σ²=0.265536, weight=0.766795 |
|
|
2025-07-10 22:57:48,581 - INFO - log_σ² gradient: -0.183440 |
|
|
2025-07-10 22:57:48,660 - INFO - Optimizer step 31: log_σ²=0.266077, weight=0.766380 |
|
|
2025-07-10 22:57:48,834 - INFO - Epoch 28: Total optimizer steps: 31 |
|
|
2025-07-10 23:01:06,110 - INFO - Validation metrics: |
|
|
2025-07-10 23:01:06,110 - INFO - Loss: 0.5027 |
|
|
2025-07-10 23:01:06,110 - INFO - Average similarity: 0.7844 |
|
|
2025-07-10 23:01:06,110 - INFO - Median similarity: 0.9843 |
|
|
2025-07-10 23:01:06,110 - INFO - Clean sample similarity: 0.7844 |
|
|
2025-07-10 23:01:06,110 - INFO - Corrupted sample similarity: 0.2361 |
|
|
2025-07-10 23:01:06,110 - INFO - Similarity gap (clean - corrupt): 0.5483 |
|
|
2025-07-10 23:01:06,230 - INFO - Epoch 28/30 - Train Loss: 0.5293, Val Loss: 0.5027, Clean Sim: 0.7844, Corrupt Sim: 0.2361, Gap: 0.5483, Time: 973.04s |
|
|
2025-07-10 23:01:06,231 - INFO - New best validation loss: 0.5027 |
|
|
2025-07-10 23:03:51,526 - INFO - Epoch 28 Validation Alignment: Pos=0.135, Neg=0.064, Gap=0.071 |
|
|
2025-07-10 23:04:57,093 - INFO - log_σ² gradient: -0.408042 |
|
|
2025-07-10 23:04:57,172 - INFO - Optimizer step 1: log_σ²=0.266622, weight=0.765962 |
|
|
2025-07-10 23:05:21,290 - INFO - log_σ² gradient: -0.405703 |
|
|
2025-07-10 23:05:21,361 - INFO - Optimizer step 2: log_σ²=0.267171, weight=0.765542 |
|
|
2025-07-10 23:05:45,924 - INFO - log_σ² gradient: -0.411940 |
|
|
2025-07-10 23:05:45,995 - INFO - Optimizer step 3: log_σ²=0.267724, weight=0.765119 |
|
|
2025-07-10 23:06:10,965 - INFO - log_σ² gradient: -0.397752 |
|
|
2025-07-10 23:06:11,041 - INFO - Optimizer step 4: log_σ²=0.268279, weight=0.764694 |
|
|
2025-07-10 23:06:34,876 - INFO - log_σ² gradient: -0.398251 |
|
|
2025-07-10 23:06:34,950 - INFO - Optimizer step 5: log_σ²=0.268836, weight=0.764269 |
|
|
2025-07-10 23:06:58,233 - INFO - log_σ² gradient: -0.401476 |
|
|
2025-07-10 23:06:58,305 - INFO - Optimizer step 6: log_σ²=0.269395, weight=0.763842 |
|
|
2025-07-10 23:07:21,635 - INFO - log_σ² gradient: -0.399305 |
|
|
2025-07-10 23:07:21,709 - INFO - Optimizer step 7: log_σ²=0.269955, weight=0.763414 |
|
|
2025-07-10 23:07:46,001 - INFO - log_σ² gradient: -0.402837 |
|
|
2025-07-10 23:07:46,079 - INFO - Optimizer step 8: log_σ²=0.270518, weight=0.762984 |
|
|
2025-07-10 23:08:09,267 - INFO - log_σ² gradient: -0.394636 |
|
|
2025-07-10 23:08:09,345 - INFO - Optimizer step 9: log_σ²=0.271081, weight=0.762555 |
|
|
2025-07-10 23:08:33,227 - INFO - log_σ² gradient: -0.405319 |
|
|
2025-07-10 23:08:33,295 - INFO - Optimizer step 10: log_σ²=0.271647, weight=0.762123 |
|
|
2025-07-10 23:08:58,537 - INFO - log_σ² gradient: -0.397017 |
|
|
2025-07-10 23:08:58,608 - INFO - Optimizer step 11: log_σ²=0.272214, weight=0.761691 |
|
|
2025-07-10 23:09:22,874 - INFO - log_σ² gradient: -0.400395 |
|
|
2025-07-10 23:09:22,948 - INFO - Optimizer step 12: log_σ²=0.272782, weight=0.761259 |
|
|
2025-07-10 23:09:47,101 - INFO - log_σ² gradient: -0.398416 |
|
|
2025-07-10 23:09:47,177 - INFO - Optimizer step 13: log_σ²=0.273352, weight=0.760825 |
|
|
2025-07-10 23:10:11,896 - INFO - log_σ² gradient: -0.401500 |
|
|
2025-07-10 23:10:11,975 - INFO - Optimizer step 14: log_σ²=0.273923, weight=0.760391 |
|
|
2025-07-10 23:10:36,727 - INFO - log_σ² gradient: -0.404077 |
|
|
2025-07-10 23:10:36,801 - INFO - Optimizer step 15: log_σ²=0.274495, weight=0.759955 |
|
|
2025-07-10 23:11:00,123 - INFO - log_σ² gradient: -0.402971 |
|
|
2025-07-10 23:11:00,194 - INFO - Optimizer step 16: log_σ²=0.275070, weight=0.759519 |
|
|
2025-07-10 23:11:27,722 - INFO - log_σ² gradient: -0.399028 |
|
|
2025-07-10 23:11:27,801 - INFO - Optimizer step 17: log_σ²=0.275645, weight=0.759082 |
|
|
2025-07-10 23:11:49,967 - INFO - log_σ² gradient: -0.402856 |
|
|
2025-07-10 23:11:50,041 - INFO - Optimizer step 18: log_σ²=0.276222, weight=0.758645 |
|
|
2025-07-10 23:12:14,614 - INFO - log_σ² gradient: -0.392941 |
|
|
2025-07-10 23:12:14,685 - INFO - Optimizer step 19: log_σ²=0.276799, weight=0.758207 |
|
|
2025-07-10 23:12:39,173 - INFO - log_σ² gradient: -0.398856 |
|
|
2025-07-10 23:12:39,256 - INFO - Optimizer step 20: log_σ²=0.277377, weight=0.757769 |
|
|
2025-07-10 23:13:00,447 - INFO - log_σ² gradient: -0.401482 |
|
|
2025-07-10 23:13:00,517 - INFO - Optimizer step 21: log_σ²=0.277956, weight=0.757331 |
|
|
2025-07-10 23:13:24,180 - INFO - log_σ² gradient: -0.397162 |
|
|
2025-07-10 23:13:24,251 - INFO - Optimizer step 22: log_σ²=0.278535, weight=0.756892 |
|
|
2025-07-10 23:13:47,745 - INFO - log_σ² gradient: -0.408184 |
|
|
2025-07-10 23:13:47,819 - INFO - Optimizer step 23: log_σ²=0.279117, weight=0.756451 |
|
|
2025-07-10 23:14:12,334 - INFO - log_σ² gradient: -0.400418 |
|
|
2025-07-10 23:14:12,408 - INFO - Optimizer step 24: log_σ²=0.279700, weight=0.756011 |
|
|
2025-07-10 23:14:36,117 - INFO - log_σ² gradient: -0.394485 |
|
|
2025-07-10 23:14:36,192 - INFO - Optimizer step 25: log_σ²=0.280283, weight=0.755570 |
|
|
2025-07-10 23:15:01,822 - INFO - log_σ² gradient: -0.397468 |
|
|
2025-07-10 23:15:01,901 - INFO - Optimizer step 26: log_σ²=0.280866, weight=0.755129 |
|
|
2025-07-10 23:15:25,148 - INFO - log_σ² gradient: -0.398808 |
|
|
2025-07-10 23:15:25,227 - INFO - Optimizer step 27: log_σ²=0.281450, weight=0.754688 |
|
|
2025-07-10 23:15:48,647 - INFO - log_σ² gradient: -0.405459 |
|
|
2025-07-10 23:15:48,721 - INFO - Optimizer step 28: log_σ²=0.282036, weight=0.754246 |
|
|
2025-07-10 23:16:13,667 - INFO - log_σ² gradient: -0.396640 |
|
|
2025-07-10 23:16:13,743 - INFO - Optimizer step 29: log_σ²=0.282623, weight=0.753804 |
|
|
2025-07-10 23:16:36,361 - INFO - log_σ² gradient: -0.400180 |
|
|
2025-07-10 23:16:36,433 - INFO - Optimizer step 30: log_σ²=0.283210, weight=0.753361 |
|
|
2025-07-10 23:16:47,095 - INFO - log_σ² gradient: -0.178048 |
|
|
2025-07-10 23:16:47,163 - INFO - Optimizer step 31: log_σ²=0.283766, weight=0.752943 |
|
|
2025-07-10 23:16:47,332 - INFO - Epoch 29: Total optimizer steps: 31 |
|
|
2025-07-10 23:20:04,739 - INFO - Validation metrics: |
|
|
2025-07-10 23:20:04,739 - INFO - Loss: 0.4967 |
|
|
2025-07-10 23:20:04,739 - INFO - Average similarity: 0.8746 |
|
|
2025-07-10 23:20:04,739 - INFO - Median similarity: 0.9826 |
|
|
2025-07-10 23:20:04,739 - INFO - Clean sample similarity: 0.8746 |
|
|
2025-07-10 23:20:04,739 - INFO - Corrupted sample similarity: 0.3010 |
|
|
2025-07-10 23:20:04,739 - INFO - Similarity gap (clean - corrupt): 0.5735 |
|
|
2025-07-10 23:20:04,847 - INFO - Epoch 29/30 - Train Loss: 0.5201, Val Loss: 0.4967, Clean Sim: 0.8746, Corrupt Sim: 0.3010, Gap: 0.5735, Time: 973.32s |
|
|
2025-07-10 23:20:04,848 - INFO - New best validation loss: 0.4967 |
|
|
2025-07-10 23:21:17,901 - INFO - log_σ² gradient: -0.389742 |
|
|
2025-07-10 23:21:17,972 - INFO - Optimizer step 1: log_σ²=0.284324, weight=0.752522 |
|
|
2025-07-10 23:21:42,724 - INFO - log_σ² gradient: -0.399307 |
|
|
2025-07-10 23:21:42,792 - INFO - Optimizer step 2: log_σ²=0.284887, weight=0.752099 |
|
|
2025-07-10 23:22:06,044 - INFO - log_σ² gradient: -0.396182 |
|
|
2025-07-10 23:22:06,120 - INFO - Optimizer step 3: log_σ²=0.285452, weight=0.751674 |
|
|
2025-07-10 23:22:29,873 - INFO - log_σ² gradient: -0.387181 |
|
|
2025-07-10 23:22:29,938 - INFO - Optimizer step 4: log_σ²=0.286019, weight=0.751248 |
|
|
2025-07-10 23:22:54,343 - INFO - log_σ² gradient: -0.397746 |
|
|
2025-07-10 23:22:54,417 - INFO - Optimizer step 5: log_σ²=0.286589, weight=0.750820 |
|
|
2025-07-10 23:23:19,766 - INFO - log_σ² gradient: -0.393937 |
|
|
2025-07-10 23:23:19,841 - INFO - Optimizer step 6: log_σ²=0.287161, weight=0.750391 |
|
|
2025-07-10 23:23:43,357 - INFO - log_σ² gradient: -0.396711 |
|
|
2025-07-10 23:23:43,434 - INFO - Optimizer step 7: log_σ²=0.287736, weight=0.749959 |
|
|
2025-07-10 23:24:07,571 - INFO - log_σ² gradient: -0.393242 |
|
|
2025-07-10 23:24:07,637 - INFO - Optimizer step 8: log_σ²=0.288313, weight=0.749527 |
|
|
2025-07-10 23:24:33,730 - INFO - log_σ² gradient: -0.392161 |
|
|
2025-07-10 23:24:33,806 - INFO - Optimizer step 9: log_σ²=0.288891, weight=0.749094 |
|
|
2025-07-10 23:24:56,691 - INFO - log_σ² gradient: -0.400291 |
|
|
2025-07-10 23:24:56,767 - INFO - Optimizer step 10: log_σ²=0.289472, weight=0.748659 |
|
|
2025-07-10 23:25:21,262 - INFO - log_σ² gradient: -0.399264 |
|
|
2025-07-10 23:25:21,338 - INFO - Optimizer step 11: log_σ²=0.290055, weight=0.748222 |
|
|
2025-07-10 23:25:45,648 - INFO - log_σ² gradient: -0.390364 |
|
|
2025-07-10 23:25:45,719 - INFO - Optimizer step 12: log_σ²=0.290639, weight=0.747785 |
|
|
2025-07-10 23:26:07,150 - INFO - log_σ² gradient: -0.394739 |
|
|
2025-07-10 23:26:07,226 - INFO - Optimizer step 13: log_σ²=0.291225, weight=0.747347 |
|
|
2025-07-10 23:26:30,172 - INFO - log_σ² gradient: -0.379849 |
|
|
2025-07-10 23:26:30,244 - INFO - Optimizer step 14: log_σ²=0.291810, weight=0.746910 |
|
|
2025-07-10 23:26:53,634 - INFO - log_σ² gradient: -0.394951 |
|
|
2025-07-10 23:26:53,710 - INFO - Optimizer step 15: log_σ²=0.292397, weight=0.746472 |
|
|
2025-07-10 23:27:17,927 - INFO - log_σ² gradient: -0.396081 |
|
|
2025-07-10 23:27:18,006 - INFO - Optimizer step 16: log_σ²=0.292985, weight=0.746033 |
|
|
2025-07-10 23:27:42,022 - INFO - log_σ² gradient: -0.389301 |
|
|
2025-07-10 23:27:42,094 - INFO - Optimizer step 17: log_σ²=0.293574, weight=0.745594 |
|
|
2025-07-10 23:28:05,756 - INFO - log_σ² gradient: -0.393791 |
|
|
2025-07-10 23:28:05,823 - INFO - Optimizer step 18: log_σ²=0.294164, weight=0.745154 |
|
|
2025-07-10 23:28:29,491 - INFO - log_σ² gradient: -0.394413 |
|
|
2025-07-10 23:28:29,564 - INFO - Optimizer step 19: log_σ²=0.294756, weight=0.744713 |
|
|
2025-07-10 23:28:54,059 - INFO - log_σ² gradient: -0.397074 |
|
|
2025-07-10 23:28:54,131 - INFO - Optimizer step 20: log_σ²=0.295349, weight=0.744272 |
|
|
2025-07-10 23:29:17,555 - INFO - log_σ² gradient: -0.389693 |
|
|
2025-07-10 23:29:17,627 - INFO - Optimizer step 21: log_σ²=0.295943, weight=0.743830 |
|
|
2025-07-10 23:29:42,099 - INFO - log_σ² gradient: -0.386676 |
|
|
2025-07-10 23:29:42,178 - INFO - Optimizer step 22: log_σ²=0.296537, weight=0.743388 |
|
|
2025-07-10 23:30:06,197 - INFO - log_σ² gradient: -0.387596 |
|
|
2025-07-10 23:30:06,276 - INFO - Optimizer step 23: log_σ²=0.297131, weight=0.742947 |
|
|
2025-07-10 23:30:29,953 - INFO - log_σ² gradient: -0.390319 |
|
|
2025-07-10 23:30:30,026 - INFO - Optimizer step 24: log_σ²=0.297726, weight=0.742505 |
|
|
2025-07-10 23:30:54,176 - INFO - log_σ² gradient: -0.382272 |
|
|
2025-07-10 23:30:54,251 - INFO - Optimizer step 25: log_σ²=0.298321, weight=0.742063 |
|
|
2025-07-10 23:31:18,783 - INFO - log_σ² gradient: -0.386564 |
|
|
2025-07-10 23:31:18,855 - INFO - Optimizer step 26: log_σ²=0.298916, weight=0.741622 |
|
|
2025-07-10 23:31:42,945 - INFO - log_σ² gradient: -0.382704 |
|
|
2025-07-10 23:31:43,010 - INFO - Optimizer step 27: log_σ²=0.299510, weight=0.741181 |
|
|
2025-07-10 23:32:06,687 - INFO - log_σ² gradient: -0.391649 |
|
|
2025-07-10 23:32:06,763 - INFO - Optimizer step 28: log_σ²=0.300106, weight=0.740739 |
|
|
2025-07-10 23:32:31,419 - INFO - log_σ² gradient: -0.389508 |
|
|
2025-07-10 23:32:31,494 - INFO - Optimizer step 29: log_σ²=0.300703, weight=0.740298 |
|
|
2025-07-10 23:32:53,051 - INFO - log_σ² gradient: -0.387492 |
|
|
2025-07-10 23:32:53,123 - INFO - Optimizer step 30: log_σ²=0.301300, weight=0.739855 |
|
|
2025-07-10 23:33:04,061 - INFO - log_σ² gradient: -0.182172 |
|
|
2025-07-10 23:33:04,129 - INFO - Optimizer step 31: log_σ²=0.301867, weight=0.739436 |
|
|
2025-07-10 23:33:04,294 - INFO - Epoch 30: Total optimizer steps: 31 |
|
|
2025-07-10 23:36:22,351 - INFO - Validation metrics: |
|
|
2025-07-10 23:36:22,351 - INFO - Loss: 0.4777 |
|
|
2025-07-10 23:36:22,351 - INFO - Average similarity: 0.7371 |
|
|
2025-07-10 23:36:22,352 - INFO - Median similarity: 0.9539 |
|
|
2025-07-10 23:36:22,352 - INFO - Clean sample similarity: 0.7371 |
|
|
2025-07-10 23:36:22,352 - INFO - Corrupted sample similarity: 0.2115 |
|
|
2025-07-10 23:36:22,352 - INFO - Similarity gap (clean - corrupt): 0.5256 |
|
|
2025-07-10 23:36:22,477 - INFO - Epoch 30/30 - Train Loss: 0.5073, Val Loss: 0.4777, Clean Sim: 0.7371, Corrupt Sim: 0.2115, Gap: 0.5256, Time: 971.64s |
|
|
2025-07-10 23:36:22,477 - INFO - New best validation loss: 0.4777 |
|
|
2025-07-10 23:39:11,943 - INFO - Epoch 30 Validation Alignment: Pos=0.145, Neg=0.068, Gap=0.077 |
|
|
2025-07-10 23:39:11,943 - INFO - Training completed! |
|
|
2025-07-10 23:39:17,631 - INFO - Evaluating best models on test set... |
|
|
2025-07-10 23:39:21,161 - INFO - Loaded best loss model from epoch 30 |
|
|
2025-07-10 23:42:56,776 - INFO - Test (Best Loss) metrics: |
|
|
2025-07-10 23:42:56,776 - INFO - Loss: 0.4783 |
|
|
2025-07-10 23:42:56,776 - INFO - Average similarity: 0.7342 |
|
|
2025-07-10 23:42:56,776 - INFO - Median similarity: 0.9554 |
|
|
2025-07-10 23:42:56,776 - INFO - Clean sample similarity: 0.7342 |
|
|
2025-07-10 23:42:56,776 - INFO - Corrupted sample similarity: 0.2117 |
|
|
2025-07-10 23:42:56,776 - INFO - Similarity gap (clean - corrupt): 0.5226 |
|
|
2025-07-10 23:46:02,047 - INFO - Loaded best gap model from epoch 25 |
|
|
2025-07-10 23:49:31,744 - INFO - Test (Best Gap) metrics: |
|
|
2025-07-10 23:49:31,744 - INFO - Loss: 0.4949 |
|
|
2025-07-10 23:49:31,744 - INFO - Average similarity: 0.8834 |
|
|
2025-07-10 23:49:31,744 - INFO - Median similarity: 0.9939 |
|
|
2025-07-10 23:49:31,744 - INFO - Clean sample similarity: 0.8834 |
|
|
2025-07-10 23:49:31,744 - INFO - Corrupted sample similarity: 0.3090 |
|
|
2025-07-10 23:49:31,744 - INFO - Similarity gap (clean - corrupt): 0.5744 |
|
|
2025-07-10 23:52:25,093 - INFO - Evaluation completed! |
|
|
2025-07-10 23:52:25,093 - INFO - Test results for best_loss_model: |
|
|
2025-07-10 23:52:25,093 - INFO - Loss: 0.4783 |
|
|
2025-07-10 23:52:25,093 - INFO - Clean Sample Similarity: 0.7342 |
|
|
2025-07-10 23:52:25,093 - INFO - Corrupted Sample Similarity: 0.2117 |
|
|
2025-07-10 23:52:25,093 - INFO - Similarity Gap: 0.5226 |
|
|
2025-07-10 23:52:25,093 - INFO - Test results for best_gap_model: |
|
|
2025-07-10 23:52:25,093 - INFO - Loss: 0.4949 |
|
|
2025-07-10 23:52:25,093 - INFO - Clean Sample Similarity: 0.8834 |
|
|
2025-07-10 23:52:25,093 - INFO - Corrupted Sample Similarity: 0.3090 |
|
|
2025-07-10 23:52:25,093 - INFO - Similarity Gap: 0.5744 |
|
|
2025-07-10 23:52:25,373 - INFO - All tasks completed! |
|
|
|
