CUDA Available:  True
cuDNN Version:  92000
Skipping file: 097_SVDB_id_14_Medical_tr_1031_1st_1131.csv due to filter criteria.
Skipping file: 085_SVDB_id_2_Medical_tr_50000_1st_54982.csv due to filter criteria.
Skipping file: 050_GHL_id_19_Sensor_tr_43750_1st_55001.csv due to filter criteria.
Skipping file: 098_SVDB_id_15_Medical_tr_500_1st_303.csv due to filter criteria.
Processing file: 067_SMD_id_11_Facility_tr_1980_1st_2080.csv, Data shape: (22703, 38)
Label shape: (22703,)
Train data shape: (1980, 38), Test data shape: (20723, 38), Label test shape: (20723,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 067_SMD_id_11_Facility_tr_1980_1st_2080.csv...
Using CUDA device: 2
Data shape: (20723, 38)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=38), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 20723, window_size 5000, adding 4277 samples


  0%|          | 0/5 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


 40%|████      | 2/5 [00:01<00:01,  1.50it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


 80%|████████  | 4/5 [00:02<00:00,  1.72it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])

100%|██████████| 5/5 [00:02<00:00,  1.73it/s]
Original shapes - Output: (25000,), Label: (20723,)
Aligned shapes - Output: (20723,), Label: (20723,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]

Calculating Affiliation Metrics:  40%|███▉      | 79/200 [00:01<00:01, 78.08it/s]

Calculating Affiliation Metrics:  79%|███████▉  | 158/200 [00:02<00:00, 59.70it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:03<00:00, 65.77it/s]
Table1-only metrics completed in 3.16s
Evaluation Result:  {'AUC-PR': 0.09253496501966642, 'AUC-ROC': 0.6159651191436625, 'VUS-PR': 0.08523057947132615, 'VUS-ROC': 0.6025652469807934, 'Standard-F1': 0.14142259414176372, 'Standard-Precision': 0.12930374904361122, 'Standard-Recall': 0.15604801477377642, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.7775718849413973, 'Affiliation-P': 0.6407555067524632, 'Affiliation-R': 0.9886778758226232, 'F1_T': 0.1409034397011214, 'Precision_T': 0.12781954887218036, 'Recall_T': 0.15697137580794077}
Results for 067_SMD_id_11_Facility_tr_1980_1st_2080.csv: {'filename': '067_SMD_id_11_Facility_tr_1980_1st_2080.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '1980', 'data_shape': '22703x38', 'output_length': 25000, 'label_length': 20723, 'aligned_length': 20723, 'AUC-PR': 0.09253496501966642, 'AUC-ROC': 0.6159651191436625, 'VUS-PR': 0.08523057947132615, 'VUS-ROC': 0.6025652469807934, 'Standard-F1': 0.14142259414176372, 'Standard-Precision': 0.12930374904361122, 'Standard-Recall': 0.15604801477377642, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.7775718849413973, 'Affiliation-P': 0.6407555067524632, 'Affiliation-R': 0.9886778758226232, 'F1_T': 0.1409034397011214, 'Precision_T': 0.12781954887218036, 'Recall_T': 0.15697137580794077}
Logits metrics skipped for 067_SMD_id_11_Facility_tr_1980_1st_2080.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/067_SMD_id_11_Facility_tr_1980_1st_2080_results.pkl
Skipping file: 019_MITDB_id_1_Medical_tr_37500_1st_103211.csv due to filter criteria.
Skipping file: 186_Exathlon_id_13_Facility_tr_10766_1st_12590.csv due to filter criteria.
Skipping file: 080_LTDB_id_2_Medical_tr_500_1st_266.csv due to filter criteria.
Skipping file: 046_GHL_id_15_Sensor_tr_50000_1st_156462.csv due to filter criteria.
Skipping file: 141_CATSv2_id_4_Sensor_tr_41727_1st_41827.csv due to filter criteria.
Skipping file: 189_Exathlon_id_16_Facility_tr_11663_1st_13482.csv due to filter criteria.
Skipping file: 084_SVDB_id_1_Medical_tr_12973_1st_13073.csv due to filter criteria.
Processing file: 005_MSL_id_4_Sensor_tr_855_1st_2700.csv, Data shape: (3422, 55)
Label shape: (3422,)
Train data shape: (855, 55), Test data shape: (2567, 55), Label test shape: (2567,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 005_MSL_id_4_Sensor_tr_855_1st_2700.csv...
Using CUDA device: 2
Data shape: (2567, 55)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=55), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda


  0%|          | 0/1 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 2567, 55])
batched_mask shape: torch.Size([1, 2567])

100%|██████████| 1/1 [00:00<00:00,  3.49it/s]
Original shapes - Output: (2567,), Label: (2567,)
Aligned shapes - Output: (2567,), Label: (2567,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:00<00:00, 319.32it/s]
Table1-only metrics completed in 0.65s
Evaluation Result:  {'AUC-PR': 0.08242435774660002, 'AUC-ROC': 0.7565343986998916, 'VUS-PR': 0.08275951667390499, 'VUS-ROC': 0.7567093264716505, 'Standard-F1': 0.1477832512310708, 'Standard-Precision': 0.11363636363636279, 'Standard-Recall': 0.21126760563379984, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.7327710767301381, 'Affiliation-P': 0.5803768853871182, 'Affiliation-R': 0.993692972012049, 'F1_T': 0.142180094786282, 'Precision_T': 0.10714285714285639, 'Recall_T': 0.21126760563379984}
Results for 005_MSL_id_4_Sensor_tr_855_1st_2700.csv: {'filename': '005_MSL_id_4_Sensor_tr_855_1st_2700.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '855', 'data_shape': '3422x55', 'output_length': 2567, 'label_length': 2567, 'aligned_length': 2567, 'AUC-PR': 0.08242435774660002, 'AUC-ROC': 0.7565343986998916, 'VUS-PR': 0.08275951667390499, 'VUS-ROC': 0.7567093264716505, 'Standard-F1': 0.1477832512310708, 'Standard-Precision': 0.11363636363636279, 'Standard-Recall': 0.21126760563379984, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.7327710767301381, 'Affiliation-P': 0.5803768853871182, 'Affiliation-R': 0.993692972012049, 'F1_T': 0.142180094786282, 'Precision_T': 0.10714285714285639, 'Recall_T': 0.21126760563379984}
Logits metrics skipped for 005_MSL_id_4_Sensor_tr_855_1st_2700.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/005_MSL_id_4_Sensor_tr_855_1st_2700_results.pkl
Processing file: 169_SMAP_id_26_Sensor_tr_1811_1st_4510.csv, Data shape: (7244, 25)
Label shape: (7244,)
Train data shape: (1811, 25), Test data shape: (5433, 25), Label test shape: (5433,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 169_SMAP_id_26_Sensor_tr_1811_1st_4510.csv...
Using CUDA device: 2
Data shape: (5433, 25)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=25), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 5433, window_size 5000, adding 4567 samples


  0%|          | 0/2 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])

100%|██████████| 2/2 [00:00<00:00,  4.47it/s]
Original shapes - Output: (10000,), Label: (5433,)
Aligned shapes - Output: (5433,), Label: (5433,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]

Calculating Affiliation Metrics:  52%|█████▏    | 103/200 [00:01<00:00, 102.78it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:01<00:00, 115.17it/s]
Table1-only metrics completed in 1.77s
Evaluation Result:  {'AUC-PR': 0.2351533965945649, 'AUC-ROC': 0.700165162671391, 'VUS-PR': 0.1856607361786852, 'VUS-ROC': 0.7000486722397534, 'Standard-F1': 0.24999999999951464, 'Standard-Precision': 0.3090909090909035, 'Standard-Recall': 0.20987654320987398, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.7668551974078311, 'Affiliation-P': 0.6237280113591543, 'Affiliation-R': 0.995231472960168, 'F1_T': 0.24999999999951464, 'Precision_T': 0.3090909090909035, 'Recall_T': 0.20987654320987398}
Results for 169_SMAP_id_26_Sensor_tr_1811_1st_4510.csv: {'filename': '169_SMAP_id_26_Sensor_tr_1811_1st_4510.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '1811', 'data_shape': '7244x25', 'output_length': 10000, 'label_length': 5433, 'aligned_length': 5433, 'AUC-PR': 0.2351533965945649, 'AUC-ROC': 0.700165162671391, 'VUS-PR': 0.1856607361786852, 'VUS-ROC': 0.7000486722397534, 'Standard-F1': 0.24999999999951464, 'Standard-Precision': 0.3090909090909035, 'Standard-Recall': 0.20987654320987398, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.7668551974078311, 'Affiliation-P': 0.6237280113591543, 'Affiliation-R': 0.995231472960168, 'F1_T': 0.24999999999951464, 'Precision_T': 0.3090909090909035, 'Recall_T': 0.20987654320987398}
Logits metrics skipped for 169_SMAP_id_26_Sensor_tr_1811_1st_4510.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/169_SMAP_id_26_Sensor_tr_1811_1st_4510_results.pkl
Processing file: 071_SMD_id_15_Facility_tr_1109_1st_1209.csv, Data shape: (28700, 38)
Label shape: (28700,)
Train data shape: (1109, 38), Test data shape: (27591, 38), Label test shape: (27591,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 071_SMD_id_15_Facility_tr_1109_1st_1209.csv...
Using CUDA device: 2
Data shape: (27591, 38)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=38), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 27591, window_size 5000, adding 2409 samples


  0%|          | 0/6 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


 33%|███▎      | 2/6 [00:01<00:02,  1.92it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


 67%|██████▋   | 4/6 [00:02<00:01,  1.92it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


100%|██████████| 6/6 [00:03<00:00,  1.92it/s]
100%|██████████| 6/6 [00:03<00:00,  1.92it/s]
Original shapes - Output: (30000,), Label: (27591,)
Aligned shapes - Output: (27591,), Label: (27591,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]

Calculating Affiliation Metrics:  32%|███▏      | 64/200 [00:01<00:02, 63.23it/s]

Calculating Affiliation Metrics:  64%|██████▍   | 128/200 [00:02<00:01, 58.99it/s]

Calculating Affiliation Metrics:  98%|█████████▊| 195/200 [00:03<00:00, 62.33it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:03<00:00, 63.02it/s]
Table1-only metrics completed in 3.29s
Evaluation Result:  {'AUC-PR': 0.6675192153615741, 'AUC-ROC': 0.9608580172896781, 'VUS-PR': 0.6086754794874176, 'VUS-ROC': 0.9606154449681097, 'Standard-F1': 0.6917808219173073, 'Standard-Precision': 0.7318840579710117, 'Standard-Recall': 0.6558441558441537, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9741599857020744, 'Affiliation-P': 0.9530106880670364, 'Affiliation-R': 0.9962692818842109, 'F1_T': 0.6917808219173073, 'Precision_T': 0.7318840579710117, 'Recall_T': 0.6558441558441537}
Results for 071_SMD_id_15_Facility_tr_1109_1st_1209.csv: {'filename': '071_SMD_id_15_Facility_tr_1109_1st_1209.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '1109', 'data_shape': '28700x38', 'output_length': 30000, 'label_length': 27591, 'aligned_length': 27591, 'AUC-PR': 0.6675192153615741, 'AUC-ROC': 0.9608580172896781, 'VUS-PR': 0.6086754794874176, 'VUS-ROC': 0.9606154449681097, 'Standard-F1': 0.6917808219173073, 'Standard-Precision': 0.7318840579710117, 'Standard-Recall': 0.6558441558441537, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9741599857020744, 'Affiliation-P': 0.9530106880670364, 'Affiliation-R': 0.9962692818842109, 'F1_T': 0.6917808219173073, 'Precision_T': 0.7318840579710117, 'Recall_T': 0.6558441558441537}
Logits metrics skipped for 071_SMD_id_15_Facility_tr_1109_1st_1209.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/071_SMD_id_15_Facility_tr_1109_1st_1209_results.pkl
Processing file: 065_SMD_id_9_Facility_tr_737_1st_837.csv, Data shape: (23697, 38)
Label shape: (23697,)
Train data shape: (737, 38), Test data shape: (22960, 38), Label test shape: (22960,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 065_SMD_id_9_Facility_tr_737_1st_837.csv...
Using CUDA device: 2
Data shape: (22960, 38)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=38), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 22960, window_size 5000, adding 2040 samples


  0%|          | 0/5 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


 40%|████      | 2/5 [00:01<00:01,  1.92it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


 80%|████████  | 4/5 [00:02<00:00,  1.92it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])

100%|██████████| 5/5 [00:02<00:00,  1.92it/s]
Original shapes - Output: (25000,), Label: (22960,)
Aligned shapes - Output: (22960,), Label: (22960,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]

Calculating Affiliation Metrics:  31%|███       | 62/200 [00:01<00:02, 61.99it/s]

Calculating Affiliation Metrics:  62%|██████▏   | 124/200 [00:02<00:01, 48.37it/s]

Calculating Affiliation Metrics:  90%|████████▉ | 179/200 [00:03<00:00, 50.86it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:03<00:00, 54.84it/s]
Table1-only metrics completed in 3.77s
Evaluation Result:  {'AUC-PR': 0.5789875847138798, 'AUC-ROC': 0.8680627839770829, 'VUS-PR': 0.5741954585017064, 'VUS-ROC': 0.8679228276125421, 'Standard-F1': 0.6678344646371186, 'Standard-Precision': 0.7030889810972796, 'Standard-Recall': 0.6359466221851541, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9065343629300623, 'Affiliation-P': 0.8556446249609395, 'Affiliation-R': 0.9638602604613767, 'F1_T': 0.6676923076918088, 'Precision_T': 0.7058550185873603, 'Recall_T': 0.6334445371142616}
Results for 065_SMD_id_9_Facility_tr_737_1st_837.csv: {'filename': '065_SMD_id_9_Facility_tr_737_1st_837.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '737', 'data_shape': '23697x38', 'output_length': 25000, 'label_length': 22960, 'aligned_length': 22960, 'AUC-PR': 0.5789875847138798, 'AUC-ROC': 0.8680627839770829, 'VUS-PR': 0.5741954585017064, 'VUS-ROC': 0.8679228276125421, 'Standard-F1': 0.6678344646371186, 'Standard-Precision': 0.7030889810972796, 'Standard-Recall': 0.6359466221851541, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9065343629300623, 'Affiliation-P': 0.8556446249609395, 'Affiliation-R': 0.9638602604613767, 'F1_T': 0.6676923076918088, 'Precision_T': 0.7058550185873603, 'Recall_T': 0.6334445371142616}
Logits metrics skipped for 065_SMD_id_9_Facility_tr_737_1st_837.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/065_SMD_id_9_Facility_tr_737_1st_837_results.pkl
Skipping file: 099_SVDB_id_16_Medical_tr_1115_1st_1215.csv due to filter criteria.
Skipping file: 180_Exathlon_id_7_Facility_tr_10766_1st_12590.csv due to filter criteria.
Skipping file: 104_SVDB_id_21_Medical_tr_50000_1st_106634.csv due to filter criteria.
Skipping file: 035_GHL_id_4_Sensor_tr_50000_1st_90001.csv due to filter criteria.
Skipping file: 192_Exathlon_id_19_Facility_tr_11665_1st_13484.csv due to filter criteria.
Processing file: 070_SMD_id_14_Facility_tr_1070_1st_1170.csv, Data shape: (28743, 38)
Label shape: (28743,)
Train data shape: (1070, 38), Test data shape: (27673, 38), Label test shape: (27673,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 070_SMD_id_14_Facility_tr_1070_1st_1170.csv...
Using CUDA device: 2
Data shape: (27673, 38)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=38), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 27673, window_size 5000, adding 2327 samples


  0%|          | 0/6 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


 33%|███▎      | 2/6 [00:01<00:02,  1.92it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


 67%|██████▋   | 4/6 [00:02<00:01,  1.92it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


100%|██████████| 6/6 [00:03<00:00,  1.92it/s]
100%|██████████| 6/6 [00:03<00:00,  1.92it/s]
Original shapes - Output: (30000,), Label: (27673,)
Aligned shapes - Output: (27673,), Label: (27673,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]

Calculating Affiliation Metrics:  30%|███       | 61/200 [00:01<00:02, 60.58it/s]

Calculating Affiliation Metrics:  61%|██████    | 122/200 [00:02<00:01, 42.36it/s]

Calculating Affiliation Metrics:  93%|█████████▎| 186/200 [00:03<00:00, 50.34it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:03<00:00, 52.06it/s]
Table1-only metrics completed in 3.97s
Evaluation Result:  {'AUC-PR': 0.3959139442340299, 'AUC-ROC': 0.7808777992199126, 'VUS-PR': 0.30391161450515425, 'VUS-ROC': 0.7803250374602635, 'Standard-F1': 0.4422253922962397, 'Standard-Precision': 0.5595667870036081, 'Standard-Recall': 0.36556603773584817, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8542167391330773, 'Affiliation-P': 0.7486579389086055, 'Affiliation-R': 0.9944285489312509, 'F1_T': 0.4422253922962397, 'Precision_T': 0.5595667870036081, 'Recall_T': 0.36556603773584817}
Results for 070_SMD_id_14_Facility_tr_1070_1st_1170.csv: {'filename': '070_SMD_id_14_Facility_tr_1070_1st_1170.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '1070', 'data_shape': '28743x38', 'output_length': 30000, 'label_length': 27673, 'aligned_length': 27673, 'AUC-PR': 0.3959139442340299, 'AUC-ROC': 0.7808777992199126, 'VUS-PR': 0.30391161450515425, 'VUS-ROC': 0.7803250374602635, 'Standard-F1': 0.4422253922962397, 'Standard-Precision': 0.5595667870036081, 'Standard-Recall': 0.36556603773584817, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8542167391330773, 'Affiliation-P': 0.7486579389086055, 'Affiliation-R': 0.9944285489312509, 'F1_T': 0.4422253922962397, 'Precision_T': 0.5595667870036081, 'Recall_T': 0.36556603773584817}
Logits metrics skipped for 070_SMD_id_14_Facility_tr_1070_1st_1170.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/070_SMD_id_14_Facility_tr_1070_1st_1170_results.pkl
Skipping file: 139_CATSv2_id_2_Sensor_tr_5592_1st_5692.csv due to filter criteria.
Skipping file: 090_SVDB_id_7_Medical_tr_12157_1st_12257.csv due to filter criteria.
Processing file: 147_SMAP_id_4_Sensor_tr_2160_1st_6449.csv, Data shape: (8640, 25)
Label shape: (8640,)
Train data shape: (2160, 25), Test data shape: (6480, 25), Label test shape: (6480,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 147_SMAP_id_4_Sensor_tr_2160_1st_6449.csv...
Using CUDA device: 2
Data shape: (6480, 25)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=25), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 6480, window_size 5000, adding 3520 samples


  0%|          | 0/2 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])

100%|██████████| 2/2 [00:00<00:00,  4.46it/s]
Original shapes - Output: (10000,), Label: (6480,)
Aligned shapes - Output: (6480,), Label: (6480,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:00<00:00, 248.61it/s]
Table1-only metrics completed in 0.84s
Evaluation Result:  {'AUC-PR': 0.015498436478482348, 'AUC-ROC': 0.3648073466512615, 'VUS-PR': 0.01550917700918254, 'VUS-ROC': 0.3647774547430011, 'Standard-F1': 0.04036697247686677, 'Standard-Precision': 0.02270381836945302, 'Standard-Recall': 0.18181818181818032, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.7313761419948337, 'Affiliation-P': 0.6204557145529368, 'Affiliation-R': 0.890588970513213, 'F1_T': 0.039568345323546986, 'Precision_T': 0.022199798183652853, 'Recall_T': 0.18181818181818032}
Results for 147_SMAP_id_4_Sensor_tr_2160_1st_6449.csv: {'filename': '147_SMAP_id_4_Sensor_tr_2160_1st_6449.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '2160', 'data_shape': '8640x25', 'output_length': 10000, 'label_length': 6480, 'aligned_length': 6480, 'AUC-PR': 0.015498436478482348, 'AUC-ROC': 0.3648073466512615, 'VUS-PR': 0.01550917700918254, 'VUS-ROC': 0.3647774547430011, 'Standard-F1': 0.04036697247686677, 'Standard-Precision': 0.02270381836945302, 'Standard-Recall': 0.18181818181818032, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.7313761419948337, 'Affiliation-P': 0.6204557145529368, 'Affiliation-R': 0.890588970513213, 'F1_T': 0.039568345323546986, 'Precision_T': 0.022199798183652853, 'Recall_T': 0.18181818181818032}
Logits metrics skipped for 147_SMAP_id_4_Sensor_tr_2160_1st_6449.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/147_SMAP_id_4_Sensor_tr_2160_1st_6449_results.pkl
Skipping file: 052_GHL_id_21_Sensor_tr_50000_1st_98001.csv due to filter criteria.
Skipping file: 179_Exathlon_id_6_Facility_tr_11665_1st_13484.csv due to filter criteria.
Skipping file: 091_SVDB_id_8_Medical_tr_500_1st_126.csv due to filter criteria.
Processing file: 060_SMD_id_4_Facility_tr_7176_1st_10609.csv, Data shape: (28705, 38)
Label shape: (28705,)
Train data shape: (7176, 38), Test data shape: (21529, 38), Label test shape: (21529,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 060_SMD_id_4_Facility_tr_7176_1st_10609.csv...
Using CUDA device: 2
Data shape: (21529, 38)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=38), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 21529, window_size 5000, adding 3471 samples


  0%|          | 0/5 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


 40%|████      | 2/5 [00:01<00:01,  1.92it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


 80%|████████  | 4/5 [00:02<00:00,  1.92it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])

100%|██████████| 5/5 [00:02<00:00,  1.92it/s]
Original shapes - Output: (25000,), Label: (21529,)
Aligned shapes - Output: (21529,), Label: (21529,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]

Calculating Affiliation Metrics:  40%|████      | 80/200 [00:01<00:01, 78.63it/s]

Calculating Affiliation Metrics:  80%|███████▉  | 159/200 [00:02<00:00, 61.51it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:02<00:00, 69.94it/s]
Table1-only metrics completed in 2.96s
Evaluation Result:  {'AUC-PR': 0.3026073269949071, 'AUC-ROC': 0.9047873182869246, 'VUS-PR': 0.25789295225406456, 'VUS-ROC': 0.9046649947625565, 'Standard-F1': 0.3236994219648493, 'Standard-Precision': 0.4341085271317812, 'Standard-Recall': 0.25806451612903164, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.7799160068125879, 'Affiliation-P': 0.6407904981292629, 'Affiliation-R': 0.9962081807948883, 'F1_T': 0.32307692307647834, 'Precision_T': 0.4861111111111089, 'Recall_T': 0.24193548387096717}
Results for 060_SMD_id_4_Facility_tr_7176_1st_10609.csv: {'filename': '060_SMD_id_4_Facility_tr_7176_1st_10609.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '7176', 'data_shape': '28705x38', 'output_length': 25000, 'label_length': 21529, 'aligned_length': 21529, 'AUC-PR': 0.3026073269949071, 'AUC-ROC': 0.9047873182869246, 'VUS-PR': 0.25789295225406456, 'VUS-ROC': 0.9046649947625565, 'Standard-F1': 0.3236994219648493, 'Standard-Precision': 0.4341085271317812, 'Standard-Recall': 0.25806451612903164, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.7799160068125879, 'Affiliation-P': 0.6407904981292629, 'Affiliation-R': 0.9962081807948883, 'F1_T': 0.32307692307647834, 'Precision_T': 0.4861111111111089, 'Recall_T': 0.24193548387096717}
Logits metrics skipped for 060_SMD_id_4_Facility_tr_7176_1st_10609.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/060_SMD_id_4_Facility_tr_7176_1st_10609_results.pkl
Skipping file: 119_TAO_id_4_Environment_tr_500_1st_1.csv due to filter criteria.
Processing file: 161_SMAP_id_18_Sensor_tr_2075_1st_5550.csv, Data shape: (8302, 25)
Label shape: (8302,)
Train data shape: (2075, 25), Test data shape: (6227, 25), Label test shape: (6227,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 161_SMAP_id_18_Sensor_tr_2075_1st_5550.csv...
Using CUDA device: 2
Data shape: (6227, 25)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=25), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 6227, window_size 5000, adding 3773 samples


  0%|          | 0/2 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])

100%|██████████| 2/2 [00:00<00:00,  4.46it/s]
Original shapes - Output: (10000,), Label: (6227,)
Aligned shapes - Output: (6227,), Label: (6227,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:00<00:00, 349.64it/s]
Table1-only metrics completed in 0.61s
Evaluation Result:  {'AUC-PR': 0.13723259154934164, 'AUC-ROC': 0.5331509312108359, 'VUS-PR': 0.1310403664211906, 'VUS-ROC': 0.5332163865179522, 'Standard-F1': 0.20072992700683803, 'Standard-Precision': 0.27918781725888187, 'Standard-Recall': 0.15669515669515624, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9336719022268684, 'Affiliation-P': 0.9024610566886141, 'Affiliation-R': 0.9671188835312812, 'F1_T': 0.1975308641970585, 'Precision_T': 0.2592592592592581, 'Recall_T': 0.15954415954415907}
Results for 161_SMAP_id_18_Sensor_tr_2075_1st_5550.csv: {'filename': '161_SMAP_id_18_Sensor_tr_2075_1st_5550.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '2075', 'data_shape': '8302x25', 'output_length': 10000, 'label_length': 6227, 'aligned_length': 6227, 'AUC-PR': 0.13723259154934164, 'AUC-ROC': 0.5331509312108359, 'VUS-PR': 0.1310403664211906, 'VUS-ROC': 0.5332163865179522, 'Standard-F1': 0.20072992700683803, 'Standard-Precision': 0.27918781725888187, 'Standard-Recall': 0.15669515669515624, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9336719022268684, 'Affiliation-P': 0.9024610566886141, 'Affiliation-R': 0.9671188835312812, 'F1_T': 0.1975308641970585, 'Precision_T': 0.2592592592592581, 'Recall_T': 0.15954415954415907}
Logits metrics skipped for 161_SMAP_id_18_Sensor_tr_2075_1st_5550.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/161_SMAP_id_18_Sensor_tr_2075_1st_5550_results.pkl
Skipping file: 187_Exathlon_id_14_Facility_tr_6193_1st_6293.csv due to filter criteria.
Skipping file: 109_SVDB_id_26_Medical_tr_1038_1st_1138.csv due to filter criteria.
Skipping file: 102_SVDB_id_19_Medical_tr_50000_1st_107354.csv due to filter criteria.
Skipping file: 022_MITDB_id_4_Medical_tr_50000_1st_54253.csv due to filter criteria.
Skipping file: 177_Exathlon_id_4_Facility_tr_11665_1st_13484.csv due to filter criteria.
Skipping file: 086_SVDB_id_3_Medical_tr_533_1st_633.csv due to filter criteria.
Skipping file: 056_GHL_id_25_Sensor_tr_50000_1st_105568.csv due to filter criteria.
Skipping file: 083_LTDB_id_5_Medical_tr_15502_1st_15602.csv due to filter criteria.
Skipping file: 029_MITDB_id_11_Medical_tr_50000_1st_518037.csv due to filter criteria.
Processing file: 061_SMD_id_5_Facility_tr_7176_1st_15144.csv, Data shape: (28704, 38)
Label shape: (28704,)
Train data shape: (7176, 38), Test data shape: (21528, 38), Label test shape: (21528,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 061_SMD_id_5_Facility_tr_7176_1st_15144.csv...
Using CUDA device: 2
Data shape: (21528, 38)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=38), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 21528, window_size 5000, adding 3472 samples


  0%|          | 0/5 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


 40%|████      | 2/5 [00:01<00:01,  1.92it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


 80%|████████  | 4/5 [00:02<00:00,  1.92it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])

100%|██████████| 5/5 [00:02<00:00,  1.92it/s]
Original shapes - Output: (25000,), Label: (21528,)
Aligned shapes - Output: (21528,), Label: (21528,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]

Calculating Affiliation Metrics:  41%|████      | 82/200 [00:01<00:01, 81.94it/s]

Calculating Affiliation Metrics:  82%|████████▏ | 164/200 [00:02<00:00, 68.31it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:02<00:00, 75.21it/s]
Table1-only metrics completed in 2.77s
Evaluation Result:  {'AUC-PR': 0.22376001002372983, 'AUC-ROC': 0.6434983555602017, 'VUS-PR': 0.19342774281279607, 'VUS-ROC': 0.6432502756353146, 'Standard-F1': 0.23685608072185554, 'Standard-Precision': 0.4355468749999991, 'Standard-Recall': 0.16265499635302688, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8619303641212652, 'Affiliation-P': 0.7667833224955407, 'Affiliation-R': 0.984035328622757, 'F1_T': 0.23622047244054112, 'Precision_T': 0.42134831460674077, 'Recall_T': 0.16411378555798675}
Results for 061_SMD_id_5_Facility_tr_7176_1st_15144.csv: {'filename': '061_SMD_id_5_Facility_tr_7176_1st_15144.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '7176', 'data_shape': '28704x38', 'output_length': 25000, 'label_length': 21528, 'aligned_length': 21528, 'AUC-PR': 0.22376001002372983, 'AUC-ROC': 0.6434983555602017, 'VUS-PR': 0.19342774281279607, 'VUS-ROC': 0.6432502756353146, 'Standard-F1': 0.23685608072185554, 'Standard-Precision': 0.4355468749999991, 'Standard-Recall': 0.16265499635302688, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8619303641212652, 'Affiliation-P': 0.7667833224955407, 'Affiliation-R': 0.984035328622757, 'F1_T': 0.23622047244054112, 'Precision_T': 0.42134831460674077, 'Recall_T': 0.16411378555798675}
Logits metrics skipped for 061_SMD_id_5_Facility_tr_7176_1st_15144.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/061_SMD_id_5_Facility_tr_7176_1st_15144_results.pkl
Skipping file: 190_Exathlon_id_17_Facility_tr_12538_1st_12638.csv due to filter criteria.
Skipping file: 125_TAO_id_10_Environment_tr_500_1st_9.csv due to filter criteria.
Processing file: 073_SMD_id_17_Facility_tr_5926_1st_10620.csv, Data shape: (23706, 38)
Label shape: (23706,)
Train data shape: (5926, 38), Test data shape: (17780, 38), Label test shape: (17780,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 073_SMD_id_17_Facility_tr_5926_1st_10620.csv...
Using CUDA device: 2
Data shape: (17780, 38)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=38), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 17780, window_size 5000, adding 2220 samples


  0%|          | 0/4 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


 50%|█████     | 2/4 [00:01<00:01,  1.92it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


100%|██████████| 4/4 [00:02<00:00,  1.92it/s]
100%|██████████| 4/4 [00:02<00:00,  1.92it/s]
Original shapes - Output: (20000,), Label: (17780,)
Aligned shapes - Output: (17780,), Label: (17780,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]

Calculating Affiliation Metrics:  36%|███▌      | 71/200 [00:01<00:01, 70.57it/s]

Calculating Affiliation Metrics:  71%|███████   | 142/200 [00:02<00:00, 68.76it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:02<00:00, 77.42it/s]
Table1-only metrics completed in 2.68s
Evaluation Result:  {'AUC-PR': 0.5680911184003024, 'AUC-ROC': 0.966356334841629, 'VUS-PR': 0.44905805590054826, 'VUS-ROC': 0.9660646816418875, 'Standard-F1': 0.503597122301694, 'Standard-Precision': 0.39325842696628993, 'Standard-Recall': 0.6999999999999931, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9355612497642728, 'Affiliation-P': 0.8805179238408393, 'Affiliation-R': 0.9979452442833427, 'F1_T': 0.503597122301694, 'Precision_T': 0.39325842696628993, 'Recall_T': 0.6999999999999931}
Results for 073_SMD_id_17_Facility_tr_5926_1st_10620.csv: {'filename': '073_SMD_id_17_Facility_tr_5926_1st_10620.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '5926', 'data_shape': '23706x38', 'output_length': 20000, 'label_length': 17780, 'aligned_length': 17780, 'AUC-PR': 0.5680911184003024, 'AUC-ROC': 0.966356334841629, 'VUS-PR': 0.44905805590054826, 'VUS-ROC': 0.9660646816418875, 'Standard-F1': 0.503597122301694, 'Standard-Precision': 0.39325842696628993, 'Standard-Recall': 0.6999999999999931, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9355612497642728, 'Affiliation-P': 0.8805179238408393, 'Affiliation-R': 0.9979452442833427, 'F1_T': 0.503597122301694, 'Precision_T': 0.39325842696628993, 'Recall_T': 0.6999999999999931}
Logits metrics skipped for 073_SMD_id_17_Facility_tr_5926_1st_10620.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/073_SMD_id_17_Facility_tr_5926_1st_10620_results.pkl
Skipping file: 174_Exathlon_id_1_Facility_tr_10766_1st_12590.csv due to filter criteria.
Skipping file: 089_SVDB_id_6_Medical_tr_50000_1st_83083.csv due to filter criteria.
Skipping file: 123_TAO_id_8_Environment_tr_500_1st_62.csv due to filter criteria.
Processing file: 003_MSL_id_2_Sensor_tr_883_1st_1238.csv, Data shape: (3535, 55)
Label shape: (3535,)
Train data shape: (883, 55), Test data shape: (2652, 55), Label test shape: (2652,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 003_MSL_id_2_Sensor_tr_883_1st_1238.csv...
Using CUDA device: 2
Data shape: (2652, 55)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=55), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda


  0%|          | 0/1 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 2652, 55])
batched_mask shape: torch.Size([1, 2652])

100%|██████████| 1/1 [00:00<00:00,  3.24it/s]
Original shapes - Output: (2652,), Label: (2652,)
Aligned shapes - Output: (2652,), Label: (2652,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:00<00:00, 333.30it/s]
Table1-only metrics completed in 0.63s
Evaluation Result:  {'AUC-PR': 0.16557670466438104, 'AUC-ROC': 0.7537011102426032, 'VUS-PR': 0.1632407804171706, 'VUS-ROC': 0.7535645011869607, 'Standard-F1': 0.291072768191764, 'Standard-Precision': 0.17556561085972835, 'Standard-Recall': 0.8508771929824525, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.7879428044388743, 'Affiliation-P': 0.663009857129171, 'Affiliation-R': 0.9708906601702247, 'F1_T': 0.29110105580665296, 'Precision_T': 0.17577413479052809, 'Recall_T': 0.8464912280701717}
Results for 003_MSL_id_2_Sensor_tr_883_1st_1238.csv: {'filename': '003_MSL_id_2_Sensor_tr_883_1st_1238.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '883', 'data_shape': '3535x55', 'output_length': 2652, 'label_length': 2652, 'aligned_length': 2652, 'AUC-PR': 0.16557670466438104, 'AUC-ROC': 0.7537011102426032, 'VUS-PR': 0.1632407804171706, 'VUS-ROC': 0.7535645011869607, 'Standard-F1': 0.291072768191764, 'Standard-Precision': 0.17556561085972835, 'Standard-Recall': 0.8508771929824525, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.7879428044388743, 'Affiliation-P': 0.663009857129171, 'Affiliation-R': 0.9708906601702247, 'F1_T': 0.29110105580665296, 'Precision_T': 0.17577413479052809, 'Recall_T': 0.8464912280701717}
Logits metrics skipped for 003_MSL_id_2_Sensor_tr_883_1st_1238.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/003_MSL_id_2_Sensor_tr_883_1st_1238_results.pkl
Skipping file: 129_OPPORTUNITY_id_1_HumanActivity_tr_1801_1st_1901.csv due to filter criteria.
Skipping file: 184_Exathlon_id_11_Facility_tr_11665_1st_13484.csv due to filter criteria.
Skipping file: 032_GHL_id_1_Sensor_tr_50000_1st_65001.csv due to filter criteria.
Skipping file: 095_SVDB_id_12_Medical_tr_6264_1st_6364.csv due to filter criteria.
Processing file: 064_SMD_id_8_Facility_tr_2272_1st_2372.csv, Data shape: (21203, 38)
Label shape: (21203,)
Train data shape: (2272, 38), Test data shape: (18931, 38), Label test shape: (18931,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 064_SMD_id_8_Facility_tr_2272_1st_2372.csv...
Using CUDA device: 2
Data shape: (18931, 38)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=38), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 18931, window_size 5000, adding 1069 samples


  0%|          | 0/4 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


 50%|█████     | 2/4 [00:01<00:01,  1.92it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


100%|██████████| 4/4 [00:02<00:00,  1.92it/s]
100%|██████████| 4/4 [00:02<00:00,  1.92it/s]
Original shapes - Output: (20000,), Label: (18931,)
Aligned shapes - Output: (18931,), Label: (18931,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]

Calculating Affiliation Metrics:  38%|███▊      | 76/200 [00:01<00:01, 75.08it/s]

Calculating Affiliation Metrics:  76%|███████▌  | 152/200 [00:02<00:00, 53.94it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:03<00:00, 63.82it/s]
Table1-only metrics completed in 3.24s
Evaluation Result:  {'AUC-PR': 0.5844032723676748, 'AUC-ROC': 0.9270769288437752, 'VUS-PR': 0.5196712757557564, 'VUS-ROC': 0.9190343144183435, 'Standard-F1': 0.5858328015310883, 'Standard-Precision': 0.5847133757961775, 'Standard-Recall': 0.5869565217391297, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8805441889225616, 'Affiliation-P': 0.814535838999265, 'Affiliation-R': 0.9581943351009979, 'F1_T': 0.5839320705416287, 'Precision_T': 0.5967957276368483, 'Recall_T': 0.5716112531969302}
Results for 064_SMD_id_8_Facility_tr_2272_1st_2372.csv: {'filename': '064_SMD_id_8_Facility_tr_2272_1st_2372.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '2272', 'data_shape': '21203x38', 'output_length': 20000, 'label_length': 18931, 'aligned_length': 18931, 'AUC-PR': 0.5844032723676748, 'AUC-ROC': 0.9270769288437752, 'VUS-PR': 0.5196712757557564, 'VUS-ROC': 0.9190343144183435, 'Standard-F1': 0.5858328015310883, 'Standard-Precision': 0.5847133757961775, 'Standard-Recall': 0.5869565217391297, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8805441889225616, 'Affiliation-P': 0.814535838999265, 'Affiliation-R': 0.9581943351009979, 'F1_T': 0.5839320705416287, 'Precision_T': 0.5967957276368483, 'Recall_T': 0.5716112531969302}
Logits metrics skipped for 064_SMD_id_8_Facility_tr_2272_1st_2372.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/064_SMD_id_8_Facility_tr_2272_1st_2372_results.pkl
Processing file: 172_SWaT_id_2_Sensor_tr_23700_1st_23800.csv, Data shape: (399919, 51)
Label shape: (399919,)
Train data shape: (23700, 51), Test data shape: (376219, 51), Label test shape: (376219,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 172_SWaT_id_2_Sensor_tr_23700_1st_23800.csv...
Using CUDA device: 2
Data shape: (376219, 51)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=51), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 376219, window_size 5000, adding 3781 samples


  0%|          | 0/76 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


  3%|▎         | 2/76 [00:01<01:08,  1.08it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


  5%|▌         | 4/76 [00:03<01:06,  1.08it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


  8%|▊         | 6/76 [00:05<01:04,  1.08it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


 11%|█         | 8/76 [00:07<01:02,  1.09it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


 13%|█▎        | 10/76 [00:09<01:00,  1.09it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


 16%|█▌        | 12/76 [00:11<00:58,  1.09it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


 18%|█▊        | 14/76 [00:12<00:56,  1.09it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


 21%|██        | 16/76 [00:14<00:55,  1.09it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


 24%|██▎       | 18/76 [00:16<00:53,  1.09it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


 26%|██▋       | 20/76 [00:18<00:51,  1.09it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


 29%|██▉       | 22/76 [00:20<00:49,  1.09it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


 32%|███▏      | 24/76 [00:22<00:47,  1.09it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


 34%|███▍      | 26/76 [00:23<00:45,  1.09it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


 37%|███▋      | 28/76 [00:25<00:44,  1.09it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


 39%|███▉      | 30/76 [00:27<00:42,  1.09it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


 42%|████▏     | 32/76 [00:29<00:40,  1.09it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


 45%|████▍     | 34/76 [00:31<00:38,  1.09it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


 47%|████▋     | 36/76 [00:33<00:36,  1.09it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


 50%|█████     | 38/76 [00:34<00:34,  1.09it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


 53%|█████▎    | 40/76 [00:36<00:33,  1.09it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


 55%|█████▌    | 42/76 [00:38<00:31,  1.09it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


 58%|█████▊    | 44/76 [00:40<00:29,  1.09it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


 61%|██████    | 46/76 [00:42<00:27,  1.09it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


 63%|██████▎   | 48/76 [00:44<00:25,  1.09it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


 66%|██████▌   | 50/76 [00:45<00:23,  1.09it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


 68%|██████▊   | 52/76 [00:47<00:22,  1.09it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


 71%|███████   | 54/76 [00:49<00:20,  1.09it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


 74%|███████▎  | 56/76 [00:51<00:18,  1.09it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


 76%|███████▋  | 58/76 [00:53<00:16,  1.09it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


 79%|███████▉  | 60/76 [00:55<00:14,  1.09it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


 82%|████████▏ | 62/76 [00:56<00:12,  1.09it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


 84%|████████▍ | 64/76 [00:58<00:11,  1.09it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


 87%|████████▋ | 66/76 [01:00<00:09,  1.09it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


 89%|████████▉ | 68/76 [01:02<00:07,  1.09it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


 92%|█████████▏| 70/76 [01:04<00:05,  1.09it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


 95%|█████████▍| 72/76 [01:06<00:03,  1.09it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


 97%|█████████▋| 74/76 [01:07<00:01,  1.09it/s]batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 51])
batched_mask shape: torch.Size([1, 5000])


100%|██████████| 76/76 [01:09<00:00,  1.09it/s]
100%|██████████| 76/76 [01:09<00:00,  1.09it/s]
Original shapes - Output: (380000,), Label: (376219,)
Aligned shapes - Output: (376219,), Label: (376219,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]

Calculating Affiliation Metrics:   4%|▍         | 8/200 [00:01<00:24,  7.99it/s]

Calculating Affiliation Metrics:   8%|▊         | 16/200 [00:02<00:32,  5.64it/s]

Calculating Affiliation Metrics:  12%|█▏        | 23/200 [00:04<00:38,  4.65it/s]

Calculating Affiliation Metrics:  14%|█▍        | 28/200 [00:06<00:41,  4.11it/s]

Calculating Affiliation Metrics:  16%|█▋        | 33/200 [00:07<00:45,  3.65it/s]

Calculating Affiliation Metrics:  18%|█▊        | 37/200 [00:09<00:48,  3.34it/s]

Calculating Affiliation Metrics:  20%|██        | 41/200 [00:10<00:51,  3.07it/s]

Calculating Affiliation Metrics:  22%|██▎       | 45/200 [00:12<00:53,  2.87it/s]

Calculating Affiliation Metrics:  24%|██▍       | 48/200 [00:13<00:55,  2.76it/s]

Calculating Affiliation Metrics:  26%|██▌       | 51/200 [00:15<00:55,  2.67it/s]

Calculating Affiliation Metrics:  27%|██▋       | 54/200 [00:16<00:56,  2.60it/s]

Calculating Affiliation Metrics:  28%|██▊       | 57/200 [00:17<00:55,  2.56it/s]

Calculating Affiliation Metrics:  30%|███       | 60/200 [00:18<00:55,  2.53it/s]

Calculating Affiliation Metrics:  32%|███▏      | 63/200 [00:19<00:54,  2.51it/s]

Calculating Affiliation Metrics:  33%|███▎      | 66/200 [00:21<00:53,  2.48it/s]

Calculating Affiliation Metrics:  34%|███▍      | 69/200 [00:22<00:53,  2.45it/s]

Calculating Affiliation Metrics:  36%|███▌      | 72/200 [00:23<00:52,  2.43it/s]

Calculating Affiliation Metrics:  38%|███▊      | 75/200 [00:24<00:51,  2.42it/s]

Calculating Affiliation Metrics:  39%|███▉      | 78/200 [00:26<00:50,  2.43it/s]

Calculating Affiliation Metrics:  40%|████      | 81/200 [00:27<00:48,  2.43it/s]

Calculating Affiliation Metrics:  42%|████▏     | 84/200 [00:28<00:47,  2.43it/s]

Calculating Affiliation Metrics:  44%|████▎     | 87/200 [00:29<00:46,  2.41it/s]

Calculating Affiliation Metrics:  45%|████▌     | 90/200 [00:31<00:46,  2.38it/s]

Calculating Affiliation Metrics:  46%|████▋     | 93/200 [00:32<00:45,  2.36it/s]

Calculating Affiliation Metrics:  48%|████▊     | 96/200 [00:33<00:44,  2.35it/s]

Calculating Affiliation Metrics:  50%|████▉     | 99/200 [00:35<00:43,  2.34it/s]

Calculating Affiliation Metrics:  51%|█████     | 102/200 [00:36<00:41,  2.34it/s]

Calculating Affiliation Metrics:  52%|█████▎    | 105/200 [00:37<00:40,  2.35it/s]

Calculating Affiliation Metrics:  54%|█████▍    | 108/200 [00:38<00:38,  2.37it/s]

Calculating Affiliation Metrics:  56%|█████▌    | 111/200 [00:40<00:36,  2.41it/s]

Calculating Affiliation Metrics:  57%|█████▋    | 114/200 [00:41<00:34,  2.47it/s]

Calculating Affiliation Metrics:  58%|█████▊    | 117/200 [00:42<00:32,  2.54it/s]

Calculating Affiliation Metrics:  60%|██████    | 120/200 [00:43<00:30,  2.62it/s]

Calculating Affiliation Metrics:  62%|██████▏   | 123/200 [00:44<00:28,  2.71it/s]

Calculating Affiliation Metrics:  64%|██████▎   | 127/200 [00:45<00:25,  2.89it/s]

Calculating Affiliation Metrics:  66%|██████▌   | 131/200 [00:46<00:22,  3.07it/s]

Calculating Affiliation Metrics:  68%|██████▊   | 135/200 [00:47<00:20,  3.20it/s]

Calculating Affiliation Metrics:  70%|██████▉   | 139/200 [00:49<00:18,  3.23it/s]

Calculating Affiliation Metrics:  72%|███████▏  | 143/200 [00:50<00:17,  3.17it/s]

Calculating Affiliation Metrics:  74%|███████▎  | 147/200 [00:51<00:17,  3.06it/s]

Calculating Affiliation Metrics:  76%|███████▌  | 151/200 [00:53<00:16,  2.98it/s]

Calculating Affiliation Metrics:  78%|███████▊  | 155/200 [00:54<00:14,  3.00it/s]

Calculating Affiliation Metrics:  80%|███████▉  | 159/200 [00:55<00:13,  3.14it/s]

Calculating Affiliation Metrics:  82%|████████▏ | 163/200 [00:56<00:11,  3.36it/s]

Calculating Affiliation Metrics:  84%|████████▎ | 167/200 [00:57<00:09,  3.50it/s]

Calculating Affiliation Metrics:  86%|████████▌ | 171/200 [00:58<00:08,  3.62it/s]

Calculating Affiliation Metrics:  88%|████████▊ | 176/200 [00:59<00:06,  3.91it/s]

Calculating Affiliation Metrics:  91%|█████████ | 182/200 [01:00<00:04,  4.37it/s]

Calculating Affiliation Metrics:  94%|█████████▍| 188/200 [01:01<00:02,  4.81it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [01:02<00:00,  3.19it/s]
Table1-only metrics completed in 64.57s
Evaluation Result:  {'AUC-PR': 0.22584910846122275, 'AUC-ROC': 0.7076003990890116, 'VUS-PR': 0.22863093774965615, 'VUS-ROC': 0.6730392702202372, 'Standard-F1': 0.41276956420294514, 'Standard-Precision': 0.28868907897354285, 'Standard-Recall': 0.723911568697873, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.7207655564437833, 'Affiliation-P': 0.5750334858672074, 'Affiliation-R': 0.9654390031169592, 'F1_T': 0.4127133701553027, 'Precision_T': 0.2884095110398832, 'Recall_T': 0.7253282253880842}
Results for 172_SWaT_id_2_Sensor_tr_23700_1st_23800.csv: {'filename': '172_SWaT_id_2_Sensor_tr_23700_1st_23800.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '23700', 'data_shape': '399919x51', 'output_length': 380000, 'label_length': 376219, 'aligned_length': 376219, 'AUC-PR': 0.22584910846122275, 'AUC-ROC': 0.7076003990890116, 'VUS-PR': 0.22863093774965615, 'VUS-ROC': 0.6730392702202372, 'Standard-F1': 0.41276956420294514, 'Standard-Precision': 0.28868907897354285, 'Standard-Recall': 0.723911568697873, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.7207655564437833, 'Affiliation-P': 0.5750334858672074, 'Affiliation-R': 0.9654390031169592, 'F1_T': 0.4127133701553027, 'Precision_T': 0.2884095110398832, 'Recall_T': 0.7253282253880842}
Logits metrics skipped for 172_SWaT_id_2_Sensor_tr_23700_1st_23800.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/172_SWaT_id_2_Sensor_tr_23700_1st_23800_results.pkl
Skipping file: 122_TAO_id_7_Environment_tr_500_1st_19.csv due to filter criteria.
Skipping file: 027_MITDB_id_9_Medical_tr_25000_1st_52255.csv due to filter criteria.
Skipping file: 136_OPPORTUNITY_id_8_HumanActivity_tr_1495_1st_1595.csv due to filter criteria.
Skipping file: 030_MITDB_id_12_Medical_tr_50000_1st_84572.csv due to filter criteria.
Skipping file: 044_GHL_id_13_Sensor_tr_50000_1st_145001.csv due to filter criteria.
Skipping file: 088_SVDB_id_5_Medical_tr_500_1st_168.csv due to filter criteria.
Skipping file: 108_SVDB_id_25_Medical_tr_1238_1st_1338.csv due to filter criteria.
Skipping file: 198_Exathlon_id_25_Facility_tr_12538_1st_12638.csv due to filter criteria.
Skipping file: 175_Exathlon_id_2_Facility_tr_10684_1st_10784.csv due to filter criteria.
Processing file: 063_SMD_id_7_Facility_tr_5923_1st_6506.csv, Data shape: (23694, 38)
Label shape: (23694,)
Train data shape: (5923, 38), Test data shape: (17771, 38), Label test shape: (17771,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 063_SMD_id_7_Facility_tr_5923_1st_6506.csv...
Using CUDA device: 2
Data shape: (17771, 38)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=38), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 17771, window_size 5000, adding 2229 samples


  0%|          | 0/4 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


 50%|█████     | 2/4 [00:01<00:01,  1.90it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


100%|██████████| 4/4 [00:02<00:00,  1.91it/s]
100%|██████████| 4/4 [00:02<00:00,  1.91it/s]
Original shapes - Output: (20000,), Label: (17771,)
Aligned shapes - Output: (17771,), Label: (17771,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]

Calculating Affiliation Metrics:  40%|███▉      | 79/200 [00:01<00:01, 78.61it/s]

Calculating Affiliation Metrics:  79%|███████▉  | 158/200 [00:02<00:00, 51.76it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:03<00:00, 58.28it/s]
Table1-only metrics completed in 3.54s
Evaluation Result:  {'AUC-PR': 0.26064928529856946, 'AUC-ROC': 0.6733286585042504, 'VUS-PR': 0.2535116593587974, 'VUS-ROC': 0.673322535761314, 'Standard-F1': 0.30596285434946113, 'Standard-Precision': 0.35730593607305894, 'Standard-Recall': 0.2675213675213673, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8899953014448907, 'Affiliation-P': 0.818617653456996, 'Affiliation-R': 0.9750091803492119, 'F1_T': 0.3056640624995099, 'Precision_T': 0.3564920273348515, 'Recall_T': 0.2675213675213673}
Results for 063_SMD_id_7_Facility_tr_5923_1st_6506.csv: {'filename': '063_SMD_id_7_Facility_tr_5923_1st_6506.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '5923', 'data_shape': '23694x38', 'output_length': 20000, 'label_length': 17771, 'aligned_length': 17771, 'AUC-PR': 0.26064928529856946, 'AUC-ROC': 0.6733286585042504, 'VUS-PR': 0.2535116593587974, 'VUS-ROC': 0.673322535761314, 'Standard-F1': 0.30596285434946113, 'Standard-Precision': 0.35730593607305894, 'Standard-Recall': 0.2675213675213673, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8899953014448907, 'Affiliation-P': 0.818617653456996, 'Affiliation-R': 0.9750091803492119, 'F1_T': 0.3056640624995099, 'Precision_T': 0.3564920273348515, 'Recall_T': 0.2675213675213673}
Logits metrics skipped for 063_SMD_id_7_Facility_tr_5923_1st_6506.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/063_SMD_id_7_Facility_tr_5923_1st_6506_results.pkl
Skipping file: 110_SVDB_id_27_Medical_tr_30824_1st_30924.csv due to filter criteria.
Skipping file: 199_Exathlon_id_26_Facility_tr_12538_1st_12638.csv due to filter criteria.
Skipping file: 120_TAO_id_5_Environment_tr_500_1st_3.csv due to filter criteria.
Processing file: 075_SMD_id_19_Facility_tr_564_1st_664.csv, Data shape: (23696, 38)
Label shape: (23696,)
Train data shape: (564, 38), Test data shape: (23132, 38), Label test shape: (23132,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 075_SMD_id_19_Facility_tr_564_1st_664.csv...
Using CUDA device: 2
Data shape: (23132, 38)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=38), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 23132, window_size 5000, adding 1868 samples


  0%|          | 0/5 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


 40%|████      | 2/5 [00:01<00:01,  1.92it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


 80%|████████  | 4/5 [00:02<00:00,  1.92it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])

100%|██████████| 5/5 [00:02<00:00,  1.92it/s]
Original shapes - Output: (25000,), Label: (23132,)
Aligned shapes - Output: (23132,), Label: (23132,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]

Calculating Affiliation Metrics:  34%|███▎      | 67/200 [00:01<00:02, 66.42it/s]

Calculating Affiliation Metrics:  67%|██████▋   | 134/200 [00:03<00:01, 37.31it/s]

Calculating Affiliation Metrics:  90%|████████▉ | 179/200 [00:04<00:00, 37.10it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:04<00:00, 42.06it/s]
Table1-only metrics completed in 4.89s
Evaluation Result:  {'AUC-PR': 0.8128526412654049, 'AUC-ROC': 0.960460212063675, 'VUS-PR': 0.7088570592012189, 'VUS-ROC': 0.9562290893677312, 'Standard-F1': 0.8182912154026268, 'Standard-Precision': 0.821256038647341, 'Standard-Recall': 0.81534772182254, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8471839953544353, 'Affiliation-P': 0.7462544094270842, 'Affiliation-R': 0.9796846008505733, 'F1_T': 0.789053591789693, 'Precision_T': 0.7521739130434766, 'Recall_T': 0.829736211031173}
Results for 075_SMD_id_19_Facility_tr_564_1st_664.csv: {'filename': '075_SMD_id_19_Facility_tr_564_1st_664.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '564', 'data_shape': '23696x38', 'output_length': 25000, 'label_length': 23132, 'aligned_length': 23132, 'AUC-PR': 0.8128526412654049, 'AUC-ROC': 0.960460212063675, 'VUS-PR': 0.7088570592012189, 'VUS-ROC': 0.9562290893677312, 'Standard-F1': 0.8182912154026268, 'Standard-Precision': 0.821256038647341, 'Standard-Recall': 0.81534772182254, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8471839953544353, 'Affiliation-P': 0.7462544094270842, 'Affiliation-R': 0.9796846008505733, 'F1_T': 0.789053591789693, 'Precision_T': 0.7521739130434766, 'Recall_T': 0.829736211031173}
Logits metrics skipped for 075_SMD_id_19_Facility_tr_564_1st_664.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/075_SMD_id_19_Facility_tr_564_1st_664_results.pkl
Processing file: 150_SMAP_id_7_Sensor_tr_2077_1st_5394.csv, Data shape: (8310, 25)
Label shape: (8310,)
Train data shape: (2077, 25), Test data shape: (6233, 25), Label test shape: (6233,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 150_SMAP_id_7_Sensor_tr_2077_1st_5394.csv...
Using CUDA device: 2
Data shape: (6233, 25)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=25), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 6233, window_size 5000, adding 3767 samples


  0%|          | 0/2 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])

100%|██████████| 2/2 [00:00<00:00,  4.46it/s]
Original shapes - Output: (10000,), Label: (6233,)
Aligned shapes - Output: (6233,), Label: (6233,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:00<00:00, 326.09it/s]
Table1-only metrics completed in 0.65s
Evaluation Result:  {'AUC-PR': 0.09585401653729847, 'AUC-ROC': 0.5509072580645161, 'VUS-PR': 0.08973199220214434, 'VUS-ROC': 0.5507410410209314, 'Standard-F1': 0.1948310139160071, 'Standard-Precision': 0.22072072072071974, 'Standard-Recall': 0.1743772241992876, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8582921346378317, 'Affiliation-P': 0.754798157371028, 'Affiliation-R': 0.994677051830088, 'F1_T': 0.19318181818131955, 'Precision_T': 0.20647773279352144, 'Recall_T': 0.18149466192170752}
Results for 150_SMAP_id_7_Sensor_tr_2077_1st_5394.csv: {'filename': '150_SMAP_id_7_Sensor_tr_2077_1st_5394.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '2077', 'data_shape': '8310x25', 'output_length': 10000, 'label_length': 6233, 'aligned_length': 6233, 'AUC-PR': 0.09585401653729847, 'AUC-ROC': 0.5509072580645161, 'VUS-PR': 0.08973199220214434, 'VUS-ROC': 0.5507410410209314, 'Standard-F1': 0.1948310139160071, 'Standard-Precision': 0.22072072072071974, 'Standard-Recall': 0.1743772241992876, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8582921346378317, 'Affiliation-P': 0.754798157371028, 'Affiliation-R': 0.994677051830088, 'F1_T': 0.19318181818131955, 'Precision_T': 0.20647773279352144, 'Recall_T': 0.18149466192170752}
Logits metrics skipped for 150_SMAP_id_7_Sensor_tr_2077_1st_5394.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/150_SMAP_id_7_Sensor_tr_2077_1st_5394_results.pkl
Skipping file: 045_GHL_id_14_Sensor_tr_50000_1st_85076.csv due to filter criteria.
Processing file: 017_MSL_id_16_Sensor_tr_512_1st_1850.csv, Data shape: (2049, 55)
Label shape: (2049,)
Train data shape: (512, 55), Test data shape: (1537, 55), Label test shape: (1537,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 017_MSL_id_16_Sensor_tr_512_1st_1850.csv...
Using CUDA device: 2
Data shape: (1537, 55)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=55), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda


  0%|          | 0/1 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 1537, 55])
batched_mask shape: torch.Size([1, 1537])

100%|██████████| 1/1 [00:00<00:00,  8.45it/s]
Original shapes - Output: (1537,), Label: (1537,)
Aligned shapes - Output: (1537,), Label: (1537,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:00<00:00, 618.92it/s]
Table1-only metrics completed in 0.35s
Evaluation Result:  {'AUC-PR': 0.7178117758984145, 'AUC-ROC': 0.9674090190518099, 'VUS-PR': 0.7179736430025825, 'VUS-ROC': 0.9673906843331866, 'Standard-F1': 0.8188976377947724, 'Standard-Precision': 0.7799999999999961, 'Standard-Recall': 0.8618784530386693, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9569984281850995, 'Affiliation-P': 0.918444801472964, 'Affiliation-R': 0.9989306139174756, 'F1_T': 0.8148148148143113, 'Precision_T': 0.7817258883248691, 'Recall_T': 0.8508287292817632}
Results for 017_MSL_id_16_Sensor_tr_512_1st_1850.csv: {'filename': '017_MSL_id_16_Sensor_tr_512_1st_1850.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '512', 'data_shape': '2049x55', 'output_length': 1537, 'label_length': 1537, 'aligned_length': 1537, 'AUC-PR': 0.7178117758984145, 'AUC-ROC': 0.9674090190518099, 'VUS-PR': 0.7179736430025825, 'VUS-ROC': 0.9673906843331866, 'Standard-F1': 0.8188976377947724, 'Standard-Precision': 0.7799999999999961, 'Standard-Recall': 0.8618784530386693, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9569984281850995, 'Affiliation-P': 0.918444801472964, 'Affiliation-R': 0.9989306139174756, 'F1_T': 0.8148148148143113, 'Precision_T': 0.7817258883248691, 'Recall_T': 0.8508287292817632}
Logits metrics skipped for 017_MSL_id_16_Sensor_tr_512_1st_1850.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/017_MSL_id_16_Sensor_tr_512_1st_1850_results.pkl
Processing file: 009_MSL_id_8_Sensor_tr_714_1st_1390.csv, Data shape: (2856, 55)
Label shape: (2856,)
Train data shape: (714, 55), Test data shape: (2142, 55), Label test shape: (2142,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 009_MSL_id_8_Sensor_tr_714_1st_1390.csv...
Using CUDA device: 2
Data shape: (2142, 55)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=55), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda


  0%|          | 0/1 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 2142, 55])
batched_mask shape: torch.Size([1, 2142])

100%|██████████| 1/1 [00:00<00:00,  5.02it/s]
Original shapes - Output: (2142,), Label: (2142,)
Aligned shapes - Output: (2142,), Label: (2142,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:00<00:00, 542.14it/s]
Table1-only metrics completed in 0.39s
Evaluation Result:  {'AUC-PR': 0.01533880814067107, 'AUC-ROC': 0.6922161603915493, 'VUS-PR': 0.01539992184151629, 'VUS-ROC': 0.6918793920208347, 'Standard-F1': 0.04298642533932004, 'Standard-Precision': 0.022016222479721875, 'Standard-Recall': 0.9047619047618618, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.7574006377102643, 'Affiliation-P': 0.6095581458389807, 'Affiliation-R': 0.9999221910986617, 'F1_T': 0.04288939051914099, 'Precision_T': 0.02196531791907512, 'Recall_T': 0.9047619047618618}
Results for 009_MSL_id_8_Sensor_tr_714_1st_1390.csv: {'filename': '009_MSL_id_8_Sensor_tr_714_1st_1390.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '714', 'data_shape': '2856x55', 'output_length': 2142, 'label_length': 2142, 'aligned_length': 2142, 'AUC-PR': 0.01533880814067107, 'AUC-ROC': 0.6922161603915493, 'VUS-PR': 0.01539992184151629, 'VUS-ROC': 0.6918793920208347, 'Standard-F1': 0.04298642533932004, 'Standard-Precision': 0.022016222479721875, 'Standard-Recall': 0.9047619047618618, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.7574006377102643, 'Affiliation-P': 0.6095581458389807, 'Affiliation-R': 0.9999221910986617, 'F1_T': 0.04288939051914099, 'Precision_T': 0.02196531791907512, 'Recall_T': 0.9047619047618618}
Logits metrics skipped for 009_MSL_id_8_Sensor_tr_714_1st_1390.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/009_MSL_id_8_Sensor_tr_714_1st_1390_results.pkl
Processing file: 059_SMD_id_3_Facility_tr_757_1st_857.csv, Data shape: (23689, 38)
Label shape: (23689,)
Train data shape: (757, 38), Test data shape: (22932, 38), Label test shape: (22932,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 059_SMD_id_3_Facility_tr_757_1st_857.csv...
Using CUDA device: 2
Data shape: (22932, 38)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=38), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 22932, window_size 5000, adding 2068 samples


  0%|          | 0/5 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


 40%|████      | 2/5 [00:01<00:01,  1.92it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


 80%|████████  | 4/5 [00:02<00:00,  1.92it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])

100%|██████████| 5/5 [00:02<00:00,  1.92it/s]
Original shapes - Output: (25000,), Label: (22932,)
Aligned shapes - Output: (22932,), Label: (22932,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]

Calculating Affiliation Metrics:  40%|████      | 81/200 [00:01<00:01, 80.18it/s]

Calculating Affiliation Metrics:  81%|████████  | 162/200 [00:02<00:00, 55.81it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:03<00:00, 63.89it/s]
Table1-only metrics completed in 3.25s
Evaluation Result:  {'AUC-PR': 0.6843717932464811, 'AUC-ROC': 0.9566015517161343, 'VUS-PR': 0.6189570338729886, 'VUS-ROC': 0.9563700688297435, 'Standard-F1': 0.6764705882347917, 'Standard-Precision': 0.6690909090909066, 'Standard-Recall': 0.6840148698884733, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.89195228183576, 'Affiliation-P': 0.8134209335763751, 'Affiliation-R': 0.9872675736498142, 'F1_T': 0.6733466933862738, 'Precision_T': 0.7304347826086925, 'Recall_T': 0.6245353159851277}
Results for 059_SMD_id_3_Facility_tr_757_1st_857.csv: {'filename': '059_SMD_id_3_Facility_tr_757_1st_857.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '757', 'data_shape': '23689x38', 'output_length': 25000, 'label_length': 22932, 'aligned_length': 22932, 'AUC-PR': 0.6843717932464811, 'AUC-ROC': 0.9566015517161343, 'VUS-PR': 0.6189570338729886, 'VUS-ROC': 0.9563700688297435, 'Standard-F1': 0.6764705882347917, 'Standard-Precision': 0.6690909090909066, 'Standard-Recall': 0.6840148698884733, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.89195228183576, 'Affiliation-P': 0.8134209335763751, 'Affiliation-R': 0.9872675736498142, 'F1_T': 0.6733466933862738, 'Precision_T': 0.7304347826086925, 'Recall_T': 0.6245353159851277}
Logits metrics skipped for 059_SMD_id_3_Facility_tr_757_1st_857.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/059_SMD_id_3_Facility_tr_757_1st_857_results.pkl
Processing file: 160_SMAP_id_17_Sensor_tr_1832_1st_5300.csv, Data shape: (7331, 25)
Label shape: (7331,)
Train data shape: (1832, 25), Test data shape: (5499, 25), Label test shape: (5499,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 160_SMAP_id_17_Sensor_tr_1832_1st_5300.csv...
Using CUDA device: 2
Data shape: (5499, 25)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=25), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 5499, window_size 5000, adding 4501 samples


  0%|          | 0/2 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])

100%|██████████| 2/2 [00:00<00:00,  4.46it/s]
Original shapes - Output: (10000,), Label: (5499,)
Aligned shapes - Output: (5499,), Label: (5499,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:00<00:00, 319.27it/s]
Table1-only metrics completed in 0.66s
Evaluation Result:  {'AUC-PR': 0.06566478035687637, 'AUC-ROC': 0.36456933916904716, 'VUS-PR': 0.06502590924663071, 'VUS-ROC': 0.364581712956416, 'Standard-F1': 0.16029786400140747, 'Standard-Precision': 0.08786251342642318, 'Standard-Recall': 0.9129464285714265, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.7496908835630361, 'Affiliation-P': 0.6251503559245164, 'Affiliation-R': 0.9361972063102382, 'F1_T': 0.16020368194265222, 'Precision_T': 0.08780592528982394, 'Recall_T': 0.9129464285714265}
Results for 160_SMAP_id_17_Sensor_tr_1832_1st_5300.csv: {'filename': '160_SMAP_id_17_Sensor_tr_1832_1st_5300.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '1832', 'data_shape': '7331x25', 'output_length': 10000, 'label_length': 5499, 'aligned_length': 5499, 'AUC-PR': 0.06566478035687637, 'AUC-ROC': 0.36456933916904716, 'VUS-PR': 0.06502590924663071, 'VUS-ROC': 0.364581712956416, 'Standard-F1': 0.16029786400140747, 'Standard-Precision': 0.08786251342642318, 'Standard-Recall': 0.9129464285714265, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.7496908835630361, 'Affiliation-P': 0.6251503559245164, 'Affiliation-R': 0.9361972063102382, 'F1_T': 0.16020368194265222, 'Precision_T': 0.08780592528982394, 'Recall_T': 0.9129464285714265}
Logits metrics skipped for 160_SMAP_id_17_Sensor_tr_1832_1st_5300.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/160_SMAP_id_17_Sensor_tr_1832_1st_5300_results.pkl
Processing file: 155_SMAP_id_12_Sensor_tr_1907_1st_4800.csv, Data shape: (7628, 25)
Label shape: (7628,)
Train data shape: (1907, 25), Test data shape: (5721, 25), Label test shape: (5721,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 155_SMAP_id_12_Sensor_tr_1907_1st_4800.csv...
Using CUDA device: 2
Data shape: (5721, 25)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=25), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 5721, window_size 5000, adding 4279 samples


  0%|          | 0/2 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])

100%|██████████| 2/2 [00:00<00:00,  4.46it/s]
Original shapes - Output: (10000,), Label: (5721,)
Aligned shapes - Output: (5721,), Label: (5721,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:00<00:00, 282.07it/s]
Table1-only metrics completed in 0.75s
Evaluation Result:  {'AUC-PR': 0.0212496051701391, 'AUC-ROC': 0.7421378427914376, 'VUS-PR': 0.021411733539637885, 'VUS-ROC': 0.7422139226060794, 'Standard-F1': 0.05314009661824176, 'Standard-Precision': 0.028314028314028277, 'Standard-Recall': 0.43137254901959937, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.7246956541475158, 'Affiliation-P': 0.5925477219439572, 'Affiliation-R': 0.9327040727145605, 'F1_T': 0.05231866825196681, 'Precision_T': 0.02784810126582275, 'Recall_T': 0.43137254901959937}
Results for 155_SMAP_id_12_Sensor_tr_1907_1st_4800.csv: {'filename': '155_SMAP_id_12_Sensor_tr_1907_1st_4800.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '1907', 'data_shape': '7628x25', 'output_length': 10000, 'label_length': 5721, 'aligned_length': 5721, 'AUC-PR': 0.0212496051701391, 'AUC-ROC': 0.7421378427914376, 'VUS-PR': 0.021411733539637885, 'VUS-ROC': 0.7422139226060794, 'Standard-F1': 0.05314009661824176, 'Standard-Precision': 0.028314028314028277, 'Standard-Recall': 0.43137254901959937, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.7246956541475158, 'Affiliation-P': 0.5925477219439572, 'Affiliation-R': 0.9327040727145605, 'F1_T': 0.05231866825196681, 'Precision_T': 0.02784810126582275, 'Recall_T': 0.43137254901959937}
Logits metrics skipped for 155_SMAP_id_12_Sensor_tr_1907_1st_4800.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/155_SMAP_id_12_Sensor_tr_1907_1st_4800_results.pkl
Processing file: 165_SMAP_id_22_Sensor_tr_2129_1st_5000.csv, Data shape: (8516, 25)
Label shape: (8516,)
Train data shape: (2129, 25), Test data shape: (6387, 25), Label test shape: (6387,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 165_SMAP_id_22_Sensor_tr_2129_1st_5000.csv...
Using CUDA device: 2
Data shape: (6387, 25)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=25), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 6387, window_size 5000, adding 3613 samples


  0%|          | 0/2 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])

100%|██████████| 2/2 [00:00<00:00,  4.31it/s]
Original shapes - Output: (10000,), Label: (6387,)
Aligned shapes - Output: (6387,), Label: (6387,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:00<00:00, 284.82it/s]
Table1-only metrics completed in 0.74s
Evaluation Result:  {'AUC-PR': 0.05825854660536349, 'AUC-ROC': 0.29493439213107375, 'VUS-PR': 0.05597376942768952, 'VUS-ROC': 0.2947270446122794, 'Standard-F1': 0.15860480421175857, 'Standard-Precision': 0.08653500897666067, 'Standard-Recall': 0.9488188976377934, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.7902244834471503, 'Affiliation-P': 0.7244741522713418, 'Affiliation-R': 0.8691005158995246, 'F1_T': 0.1586570111914201, 'Precision_T': 0.08656609195402297, 'Recall_T': 0.9488188976377934}
Results for 165_SMAP_id_22_Sensor_tr_2129_1st_5000.csv: {'filename': '165_SMAP_id_22_Sensor_tr_2129_1st_5000.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '2129', 'data_shape': '8516x25', 'output_length': 10000, 'label_length': 6387, 'aligned_length': 6387, 'AUC-PR': 0.05825854660536349, 'AUC-ROC': 0.29493439213107375, 'VUS-PR': 0.05597376942768952, 'VUS-ROC': 0.2947270446122794, 'Standard-F1': 0.15860480421175857, 'Standard-Precision': 0.08653500897666067, 'Standard-Recall': 0.9488188976377934, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.7902244834471503, 'Affiliation-P': 0.7244741522713418, 'Affiliation-R': 0.8691005158995246, 'F1_T': 0.1586570111914201, 'Precision_T': 0.08656609195402297, 'Recall_T': 0.9488188976377934}
Logits metrics skipped for 165_SMAP_id_22_Sensor_tr_2129_1st_5000.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/165_SMAP_id_22_Sensor_tr_2129_1st_5000_results.pkl
Skipping file: 116_TAO_id_1_Environment_tr_500_1st_3.csv due to filter criteria.
Processing file: 167_SMAP_id_24_Sensor_tr_2094_1st_5600.csv, Data shape: (8376, 25)
Label shape: (8376,)
Train data shape: (2094, 25), Test data shape: (6282, 25), Label test shape: (6282,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 167_SMAP_id_24_Sensor_tr_2094_1st_5600.csv...
Using CUDA device: 2
Data shape: (6282, 25)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=25), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 6282, window_size 5000, adding 3718 samples


  0%|          | 0/2 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])

100%|██████████| 2/2 [00:00<00:00,  4.47it/s]
Original shapes - Output: (10000,), Label: (6282,)
Aligned shapes - Output: (6282,), Label: (6282,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:00<00:00, 294.30it/s]
Table1-only metrics completed in 0.72s
Evaluation Result:  {'AUC-PR': 0.5861825257865836, 'AUC-ROC': 0.9484076582473885, 'VUS-PR': 0.536640648266508, 'VUS-ROC': 0.9481946686154892, 'Standard-F1': 0.5384615384610504, 'Standard-Precision': 0.44444444444443737, 'Standard-Recall': 0.682926829268276, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9345970997235201, 'Affiliation-P': 0.8776937082139447, 'Affiliation-R': 0.9993904380304549, 'F1_T': 0.5384615384610504, 'Precision_T': 0.44444444444443737, 'Recall_T': 0.682926829268276}
Results for 167_SMAP_id_24_Sensor_tr_2094_1st_5600.csv: {'filename': '167_SMAP_id_24_Sensor_tr_2094_1st_5600.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '2094', 'data_shape': '8376x25', 'output_length': 10000, 'label_length': 6282, 'aligned_length': 6282, 'AUC-PR': 0.5861825257865836, 'AUC-ROC': 0.9484076582473885, 'VUS-PR': 0.536640648266508, 'VUS-ROC': 0.9481946686154892, 'Standard-F1': 0.5384615384610504, 'Standard-Precision': 0.44444444444443737, 'Standard-Recall': 0.682926829268276, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9345970997235201, 'Affiliation-P': 0.8776937082139447, 'Affiliation-R': 0.9993904380304549, 'F1_T': 0.5384615384610504, 'Precision_T': 0.44444444444443737, 'Recall_T': 0.682926829268276}
Logits metrics skipped for 167_SMAP_id_24_Sensor_tr_2094_1st_5600.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/167_SMAP_id_24_Sensor_tr_2094_1st_5600_results.pkl
Processing file: 149_SMAP_id_6_Sensor_tr_2128_1st_5000.csv, Data shape: (8512, 25)
Label shape: (8512,)
Train data shape: (2128, 25), Test data shape: (6384, 25), Label test shape: (6384,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 149_SMAP_id_6_Sensor_tr_2128_1st_5000.csv...
Using CUDA device: 2
Data shape: (6384, 25)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=25), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 6384, window_size 5000, adding 3616 samples


  0%|          | 0/2 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])

100%|██████████| 2/2 [00:00<00:00,  4.47it/s]
Original shapes - Output: (10000,), Label: (6384,)
Aligned shapes - Output: (6384,), Label: (6384,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:00<00:00, 297.11it/s]
Table1-only metrics completed in 0.71s
Evaluation Result:  {'AUC-PR': 0.06569455433456543, 'AUC-ROC': 0.281625602494158, 'VUS-PR': 0.06367795423959363, 'VUS-ROC': 0.2815793278677036, 'Standard-F1': 0.1723535946673451, 'Standard-Precision': 0.0949632738719832, 'Standard-Recall': 0.9313893653516279, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8164517117956961, 'Affiliation-P': 0.7512395308362931, 'Affiliation-R': 0.8940617324221196, 'F1_T': 0.17224425059459816, 'Precision_T': 0.09489688919958054, 'Recall_T': 0.9313893653516279}
Results for 149_SMAP_id_6_Sensor_tr_2128_1st_5000.csv: {'filename': '149_SMAP_id_6_Sensor_tr_2128_1st_5000.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '2128', 'data_shape': '8512x25', 'output_length': 10000, 'label_length': 6384, 'aligned_length': 6384, 'AUC-PR': 0.06569455433456543, 'AUC-ROC': 0.281625602494158, 'VUS-PR': 0.06367795423959363, 'VUS-ROC': 0.2815793278677036, 'Standard-F1': 0.1723535946673451, 'Standard-Precision': 0.0949632738719832, 'Standard-Recall': 0.9313893653516279, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8164517117956961, 'Affiliation-P': 0.7512395308362931, 'Affiliation-R': 0.8940617324221196, 'F1_T': 0.17224425059459816, 'Precision_T': 0.09489688919958054, 'Recall_T': 0.9313893653516279}
Logits metrics skipped for 149_SMAP_id_6_Sensor_tr_2128_1st_5000.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/149_SMAP_id_6_Sensor_tr_2128_1st_5000_results.pkl
Skipping file: 106_SVDB_id_23_Medical_tr_37500_1st_121963.csv due to filter criteria.
Processing file: 159_SMAP_id_16_Sensor_tr_1757_1st_2650.csv, Data shape: (7029, 25)
Label shape: (7029,)
Train data shape: (1757, 25), Test data shape: (5272, 25), Label test shape: (5272,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 159_SMAP_id_16_Sensor_tr_1757_1st_2650.csv...
Using CUDA device: 2
Data shape: (5272, 25)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=25), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 5272, window_size 5000, adding 4728 samples


  0%|          | 0/2 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])

100%|██████████| 2/2 [00:00<00:00,  4.46it/s]
Original shapes - Output: (10000,), Label: (5272,)
Aligned shapes - Output: (5272,), Label: (5272,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]

Calculating Affiliation Metrics:  49%|████▉     | 98/200 [00:01<00:01, 96.81it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:01<00:00, 106.25it/s]
Table1-only metrics completed in 1.92s
Evaluation Result:  {'AUC-PR': 0.5502233574438141, 'AUC-ROC': 0.8685596299080138, 'VUS-PR': 0.4603152014552946, 'VUS-ROC': 0.8527920201152511, 'Standard-F1': 0.6008583690982211, 'Standard-Precision': 0.7070707070707035, 'Standard-Recall': 0.5223880597014905, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8425350009065921, 'Affiliation-P': 0.7455691823330347, 'Affiliation-R': 0.9684934843258629, 'F1_T': 0.5964214711724619, 'Precision_T': 0.6382978723404228, 'Recall_T': 0.5597014925373113}
Results for 159_SMAP_id_16_Sensor_tr_1757_1st_2650.csv: {'filename': '159_SMAP_id_16_Sensor_tr_1757_1st_2650.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '1757', 'data_shape': '7029x25', 'output_length': 10000, 'label_length': 5272, 'aligned_length': 5272, 'AUC-PR': 0.5502233574438141, 'AUC-ROC': 0.8685596299080138, 'VUS-PR': 0.4603152014552946, 'VUS-ROC': 0.8527920201152511, 'Standard-F1': 0.6008583690982211, 'Standard-Precision': 0.7070707070707035, 'Standard-Recall': 0.5223880597014905, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8425350009065921, 'Affiliation-P': 0.7455691823330347, 'Affiliation-R': 0.9684934843258629, 'F1_T': 0.5964214711724619, 'Precision_T': 0.6382978723404228, 'Recall_T': 0.5597014925373113}
Logits metrics skipped for 159_SMAP_id_16_Sensor_tr_1757_1st_2650.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/159_SMAP_id_16_Sensor_tr_1757_1st_2650_results.pkl
Skipping file: 197_Exathlon_id_24_Facility_tr_10766_1st_12590.csv due to filter criteria.
Skipping file: 137_CreditCard_id_1_Finance_tr_500_1st_541.csv due to filter criteria.
Skipping file: 031_MITDB_id_13_Medical_tr_50000_1st_79433.csv due to filter criteria.
Skipping file: 193_Exathlon_id_20_Facility_tr_8898_1st_8998.csv due to filter criteria.
Processing file: 010_MSL_id_9_Sensor_tr_554_1st_1172.csv, Data shape: (2217, 55)
Label shape: (2217,)
Train data shape: (554, 55), Test data shape: (1663, 55), Label test shape: (1663,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 010_MSL_id_9_Sensor_tr_554_1st_1172.csv...
Using CUDA device: 2
Data shape: (1663, 55)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=55), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda


  0%|          | 0/1 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 1663, 55])
batched_mask shape: torch.Size([1, 1663])

100%|██████████| 1/1 [00:00<00:00,  8.14it/s]
Original shapes - Output: (1663,), Label: (1663,)
Aligned shapes - Output: (1663,), Label: (1663,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:00<00:00, 478.04it/s]
Table1-only metrics completed in 0.44s
Evaluation Result:  {'AUC-PR': 0.03408092280419505, 'AUC-ROC': 0.4369328823668467, 'VUS-PR': 0.0340366251453206, 'VUS-ROC': 0.4368828760024003, 'Standard-F1': 0.08263305322119645, 'Standard-Precision': 0.04341427520235464, 'Standard-Recall': 0.8550724637681036, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.6760630365895008, 'Affiliation-P': 0.5106663957297788, 'Affiliation-R': 0.9999215665769039, 'F1_T': 0.08263305322119645, 'Precision_T': 0.04341427520235464, 'Recall_T': 0.8550724637681036}
Results for 010_MSL_id_9_Sensor_tr_554_1st_1172.csv: {'filename': '010_MSL_id_9_Sensor_tr_554_1st_1172.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '554', 'data_shape': '2217x55', 'output_length': 1663, 'label_length': 1663, 'aligned_length': 1663, 'AUC-PR': 0.03408092280419505, 'AUC-ROC': 0.4369328823668467, 'VUS-PR': 0.0340366251453206, 'VUS-ROC': 0.4368828760024003, 'Standard-F1': 0.08263305322119645, 'Standard-Precision': 0.04341427520235464, 'Standard-Recall': 0.8550724637681036, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.6760630365895008, 'Affiliation-P': 0.5106663957297788, 'Affiliation-R': 0.9999215665769039, 'F1_T': 0.08263305322119645, 'Precision_T': 0.04341427520235464, 'Recall_T': 0.8550724637681036}
Logits metrics skipped for 010_MSL_id_9_Sensor_tr_554_1st_1172.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/010_MSL_id_9_Sensor_tr_554_1st_1172_results.pkl
Processing file: 077_SMD_id_21_Facility_tr_3026_1st_3126.csv, Data shape: (21207, 38)
Label shape: (21207,)
Train data shape: (3026, 38), Test data shape: (18181, 38), Label test shape: (18181,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 077_SMD_id_21_Facility_tr_3026_1st_3126.csv...
Using CUDA device: 2
Data shape: (18181, 38)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=38), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 18181, window_size 5000, adding 1819 samples


  0%|          | 0/4 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


 50%|█████     | 2/4 [00:01<00:01,  1.92it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


100%|██████████| 4/4 [00:02<00:00,  1.92it/s]
100%|██████████| 4/4 [00:02<00:00,  1.92it/s]
Original shapes - Output: (20000,), Label: (18181,)
Aligned shapes - Output: (18181,), Label: (18181,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]

Calculating Affiliation Metrics:  44%|████▎     | 87/200 [00:01<00:01, 86.49it/s]

Calculating Affiliation Metrics:  87%|████████▋ | 174/200 [00:02<00:00, 72.81it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:02<00:00, 81.16it/s]
Table1-only metrics completed in 2.57s
Evaluation Result:  {'AUC-PR': 0.6271999100747172, 'AUC-ROC': 0.9418782463151375, 'VUS-PR': 0.5978999371576388, 'VUS-ROC': 0.9410092298773932, 'Standard-F1': 0.6514935988615191, 'Standard-Precision': 0.6378830083565451, 'Standard-Recall': 0.6656976744186036, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.946867462440677, 'Affiliation-P': 0.9219033873394896, 'Affiliation-R': 0.9732211634332818, 'F1_T': 0.6524520255858531, 'Precision_T': 0.6383866481223913, 'Recall_T': 0.6671511627906966}
Results for 077_SMD_id_21_Facility_tr_3026_1st_3126.csv: {'filename': '077_SMD_id_21_Facility_tr_3026_1st_3126.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '3026', 'data_shape': '21207x38', 'output_length': 20000, 'label_length': 18181, 'aligned_length': 18181, 'AUC-PR': 0.6271999100747172, 'AUC-ROC': 0.9418782463151375, 'VUS-PR': 0.5978999371576388, 'VUS-ROC': 0.9410092298773932, 'Standard-F1': 0.6514935988615191, 'Standard-Precision': 0.6378830083565451, 'Standard-Recall': 0.6656976744186036, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.946867462440677, 'Affiliation-P': 0.9219033873394896, 'Affiliation-R': 0.9732211634332818, 'F1_T': 0.6524520255858531, 'Precision_T': 0.6383866481223913, 'Recall_T': 0.6671511627906966}
Logits metrics skipped for 077_SMD_id_21_Facility_tr_3026_1st_3126.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/077_SMD_id_21_Facility_tr_3026_1st_3126_results.pkl
Skipping file: 130_OPPORTUNITY_id_2_HumanActivity_tr_1045_1st_1145.csv due to filter criteria.
Skipping file: 181_Exathlon_id_8_Facility_tr_11663_1st_13482.csv due to filter criteria.
Skipping file: 101_SVDB_id_18_Medical_tr_1851_1st_1951.csv due to filter criteria.
Processing file: 162_SMAP_id_19_Sensor_tr_1908_1st_4690.csv, Data shape: (7632, 25)
Label shape: (7632,)
Train data shape: (1908, 25), Test data shape: (5724, 25), Label test shape: (5724,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 162_SMAP_id_19_Sensor_tr_1908_1st_4690.csv...
Using CUDA device: 2
Data shape: (5724, 25)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=25), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 5724, window_size 5000, adding 4276 samples


  0%|          | 0/2 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])

100%|██████████| 2/2 [00:00<00:00,  4.36it/s]
Original shapes - Output: (10000,), Label: (5724,)
Aligned shapes - Output: (5724,), Label: (5724,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]

Calculating Affiliation Metrics:  56%|█████▌    | 112/200 [00:01<00:00, 111.57it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:01<00:00, 137.19it/s]
Table1-only metrics completed in 1.49s
Evaluation Result:  {'AUC-PR': 0.3254407709476613, 'AUC-ROC': 0.9146433367519817, 'VUS-PR': 0.21713094426956767, 'VUS-ROC': 0.9144790149759466, 'Standard-F1': 0.2696629213478545, 'Standard-Precision': 0.20689655172413435, 'Standard-Recall': 0.38709677419353594, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.6732273399050926, 'Affiliation-P': 0.507418705046341, 'Affiliation-R': 0.9999943644191971, 'F1_T': 0.2696629213478545, 'Precision_T': 0.20689655172413435, 'Recall_T': 0.38709677419353594}
Results for 162_SMAP_id_19_Sensor_tr_1908_1st_4690.csv: {'filename': '162_SMAP_id_19_Sensor_tr_1908_1st_4690.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '1908', 'data_shape': '7632x25', 'output_length': 10000, 'label_length': 5724, 'aligned_length': 5724, 'AUC-PR': 0.3254407709476613, 'AUC-ROC': 0.9146433367519817, 'VUS-PR': 0.21713094426956767, 'VUS-ROC': 0.9144790149759466, 'Standard-F1': 0.2696629213478545, 'Standard-Precision': 0.20689655172413435, 'Standard-Recall': 0.38709677419353594, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.6732273399050926, 'Affiliation-P': 0.507418705046341, 'Affiliation-R': 0.9999943644191971, 'F1_T': 0.2696629213478545, 'Precision_T': 0.20689655172413435, 'Recall_T': 0.38709677419353594}
Logits metrics skipped for 162_SMAP_id_19_Sensor_tr_1908_1st_4690.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/162_SMAP_id_19_Sensor_tr_1908_1st_4690_results.pkl
Skipping file: 033_GHL_id_2_Sensor_tr_50000_1st_51001.csv due to filter criteria.
Skipping file: 049_GHL_id_18_Sensor_tr_50000_1st_109001.csv due to filter criteria.
Skipping file: 173_GECCO_id_1_Sensor_tr_16165_1st_16265.csv due to filter criteria.
Skipping file: 094_SVDB_id_11_Medical_tr_20100_1st_23611.csv due to filter criteria.
Skipping file: 143_CATSv2_id_6_Sensor_tr_50000_1st_60232.csv due to filter criteria.
Skipping file: 142_CATSv2_id_5_Sensor_tr_30704_1st_30804.csv due to filter criteria.
Skipping file: 105_SVDB_id_22_Medical_tr_500_1st_417.csv due to filter criteria.
Processing file: 057_SMD_id_1_Facility_tr_4529_1st_4629.csv, Data shape: (23694, 38)
Label shape: (23694,)
Train data shape: (4529, 38), Test data shape: (19165, 38), Label test shape: (19165,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 057_SMD_id_1_Facility_tr_4529_1st_4629.csv...
Using CUDA device: 2
Data shape: (19165, 38)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=38), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 19165, window_size 5000, adding 835 samples


  0%|          | 0/4 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


 50%|█████     | 2/4 [00:01<00:01,  1.92it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


100%|██████████| 4/4 [00:02<00:00,  1.92it/s]
100%|██████████| 4/4 [00:02<00:00,  1.92it/s]
Original shapes - Output: (20000,), Label: (19165,)
Aligned shapes - Output: (19165,), Label: (19165,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]

Calculating Affiliation Metrics:  43%|████▎     | 86/200 [00:01<00:01, 85.68it/s]

Calculating Affiliation Metrics:  86%|████████▌ | 172/200 [00:02<00:00, 69.35it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:02<00:00, 74.51it/s]
Table1-only metrics completed in 2.79s
Evaluation Result:  {'AUC-PR': 0.37183540510108476, 'AUC-ROC': 0.9375140806125345, 'VUS-PR': 0.35939313393020983, 'VUS-ROC': 0.9372531248804943, 'Standard-F1': 0.45958083832287033, 'Standard-Precision': 0.3866498740554151, 'Standard-Recall': 0.566420664206641, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9038236451775721, 'Affiliation-P': 0.8296545472533599, 'Affiliation-R': 0.9925557107718477, 'F1_T': 0.4599999999995131, 'Precision_T': 0.3944591029023741, 'Recall_T': 0.551660516605165}
Results for 057_SMD_id_1_Facility_tr_4529_1st_4629.csv: {'filename': '057_SMD_id_1_Facility_tr_4529_1st_4629.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '4529', 'data_shape': '23694x38', 'output_length': 20000, 'label_length': 19165, 'aligned_length': 19165, 'AUC-PR': 0.37183540510108476, 'AUC-ROC': 0.9375140806125345, 'VUS-PR': 0.35939313393020983, 'VUS-ROC': 0.9372531248804943, 'Standard-F1': 0.45958083832287033, 'Standard-Precision': 0.3866498740554151, 'Standard-Recall': 0.566420664206641, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9038236451775721, 'Affiliation-P': 0.8296545472533599, 'Affiliation-R': 0.9925557107718477, 'F1_T': 0.4599999999995131, 'Precision_T': 0.3944591029023741, 'Recall_T': 0.551660516605165}
Logits metrics skipped for 057_SMD_id_1_Facility_tr_4529_1st_4629.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/057_SMD_id_1_Facility_tr_4529_1st_4629_results.pkl
Skipping file: 133_OPPORTUNITY_id_5_HumanActivity_tr_1745_1st_6500.csv due to filter criteria.
Processing file: 152_SMAP_id_9_Sensor_tr_2073_1st_5600.csv, Data shape: (8294, 25)
Label shape: (8294,)
Train data shape: (2073, 25), Test data shape: (6221, 25), Label test shape: (6221,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 152_SMAP_id_9_Sensor_tr_2073_1st_5600.csv...
Using CUDA device: 2
Data shape: (6221, 25)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=25), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 6221, window_size 5000, adding 3779 samples


  0%|          | 0/2 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])

100%|██████████| 2/2 [00:00<00:00,  4.46it/s]
Original shapes - Output: (10000,), Label: (6221,)
Aligned shapes - Output: (6221,), Label: (6221,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:00<00:00, 324.02it/s]
Table1-only metrics completed in 0.66s
Evaluation Result:  {'AUC-PR': 0.09576083719821829, 'AUC-ROC': 0.35499392787369977, 'VUS-PR': 0.08710134913835313, 'VUS-ROC': 0.3550010560219653, 'Standard-F1': 0.13127413127365942, 'Standard-Precision': 0.17258883248730877, 'Standard-Recall': 0.10591900311526446, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.933848313535585, 'Affiliation-P': 0.9102711581739271, 'Affiliation-R': 0.9586792999893338, 'F1_T': 0.13043478260823124, 'Precision_T': 0.17837837837837742, 'Recall_T': 0.10280373831775669}
Results for 152_SMAP_id_9_Sensor_tr_2073_1st_5600.csv: {'filename': '152_SMAP_id_9_Sensor_tr_2073_1st_5600.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '2073', 'data_shape': '8294x25', 'output_length': 10000, 'label_length': 6221, 'aligned_length': 6221, 'AUC-PR': 0.09576083719821829, 'AUC-ROC': 0.35499392787369977, 'VUS-PR': 0.08710134913835313, 'VUS-ROC': 0.3550010560219653, 'Standard-F1': 0.13127413127365942, 'Standard-Precision': 0.17258883248730877, 'Standard-Recall': 0.10591900311526446, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.933848313535585, 'Affiliation-P': 0.9102711581739271, 'Affiliation-R': 0.9586792999893338, 'F1_T': 0.13043478260823124, 'Precision_T': 0.17837837837837742, 'Recall_T': 0.10280373831775669}
Logits metrics skipped for 152_SMAP_id_9_Sensor_tr_2073_1st_5600.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/152_SMAP_id_9_Sensor_tr_2073_1st_5600_results.pkl
Processing file: 062_SMD_id_6_Facility_tr_7180_1st_15131.csv, Data shape: (28722, 38)
Label shape: (28722,)
Train data shape: (7180, 38), Test data shape: (21542, 38), Label test shape: (21542,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 062_SMD_id_6_Facility_tr_7180_1st_15131.csv...
Using CUDA device: 2
Data shape: (21542, 38)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=38), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 21542, window_size 5000, adding 3458 samples


  0%|          | 0/5 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


 40%|████      | 2/5 [00:01<00:01,  1.92it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


 80%|████████  | 4/5 [00:02<00:00,  1.92it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])

100%|██████████| 5/5 [00:02<00:00,  1.92it/s]
Original shapes - Output: (25000,), Label: (21542,)
Aligned shapes - Output: (21542,), Label: (21542,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]

Calculating Affiliation Metrics:  36%|███▌      | 71/200 [00:01<00:01, 69.69it/s]

Calculating Affiliation Metrics:  70%|███████   | 141/200 [00:02<00:01, 56.81it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:03<00:00, 62.88it/s]
Table1-only metrics completed in 3.30s
Evaluation Result:  {'AUC-PR': 0.32281351502764416, 'AUC-ROC': 0.7092597128074967, 'VUS-PR': 0.3135116588612643, 'VUS-ROC': 0.7092360419090089, 'Standard-F1': 0.340740740740244, 'Standard-Precision': 0.3717171717171715, 'Standard-Recall': 0.31452991452991436, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9153126416216196, 'Affiliation-P': 0.8698398931498195, 'Affiliation-R': 0.9658020161608929, 'F1_T': 0.3403205918614021, 'Precision_T': 0.3707186030893215, 'Recall_T': 0.31452991452991436}
Results for 062_SMD_id_6_Facility_tr_7180_1st_15131.csv: {'filename': '062_SMD_id_6_Facility_tr_7180_1st_15131.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '7180', 'data_shape': '28722x38', 'output_length': 25000, 'label_length': 21542, 'aligned_length': 21542, 'AUC-PR': 0.32281351502764416, 'AUC-ROC': 0.7092597128074967, 'VUS-PR': 0.3135116588612643, 'VUS-ROC': 0.7092360419090089, 'Standard-F1': 0.340740740740244, 'Standard-Precision': 0.3717171717171715, 'Standard-Recall': 0.31452991452991436, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9153126416216196, 'Affiliation-P': 0.8698398931498195, 'Affiliation-R': 0.9658020161608929, 'F1_T': 0.3403205918614021, 'Precision_T': 0.3707186030893215, 'Recall_T': 0.31452991452991436}
Logits metrics skipped for 062_SMD_id_6_Facility_tr_7180_1st_15131.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/062_SMD_id_6_Facility_tr_7180_1st_15131_results.pkl
Skipping file: 127_TAO_id_12_Environment_tr_500_1st_24.csv due to filter criteria.
Skipping file: 194_Exathlon_id_21_Facility_tr_6985_1st_7085.csv due to filter criteria.
Processing file: 166_SMAP_id_23_Sensor_tr_1113_1st_1890.csv, Data shape: (4453, 25)
Label shape: (4453,)
Train data shape: (1113, 25), Test data shape: (3340, 25), Label test shape: (3340,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 166_SMAP_id_23_Sensor_tr_1113_1st_1890.csv...
Using CUDA device: 2
Data shape: (3340, 25)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=25), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda


  0%|          | 0/1 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 3340, 25])
batched_mask shape: torch.Size([1, 3340])

100%|██████████| 1/1 [00:00<00:00,  9.69it/s]
Original shapes - Output: (3340,), Label: (3340,)
Aligned shapes - Output: (3340,), Label: (3340,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:00<00:00, 253.84it/s]
Table1-only metrics completed in 0.82s
Evaluation Result:  {'AUC-PR': 0.8251667026870433, 'AUC-ROC': 0.9920005323120826, 'VUS-PR': 0.8030405744313674, 'VUS-ROC': 0.991860061068025, 'Standard-F1': 0.7733333333328171, 'Standard-Precision': 0.8529411764705631, 'Standard-Recall': 0.7073170731707145, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9990826995676622, 'Affiliation-P': 1.0, 'Affiliation-R': 0.9981670804732, 'F1_T': 0.7733333333328171, 'Precision_T': 0.8529411764705631, 'Recall_T': 0.7073170731707145}
Results for 166_SMAP_id_23_Sensor_tr_1113_1st_1890.csv: {'filename': '166_SMAP_id_23_Sensor_tr_1113_1st_1890.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '1113', 'data_shape': '4453x25', 'output_length': 3340, 'label_length': 3340, 'aligned_length': 3340, 'AUC-PR': 0.8251667026870433, 'AUC-ROC': 0.9920005323120826, 'VUS-PR': 0.8030405744313674, 'VUS-ROC': 0.991860061068025, 'Standard-F1': 0.7733333333328171, 'Standard-Precision': 0.8529411764705631, 'Standard-Recall': 0.7073170731707145, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9990826995676622, 'Affiliation-P': 1.0, 'Affiliation-R': 0.9981670804732, 'F1_T': 0.7733333333328171, 'Precision_T': 0.8529411764705631, 'Recall_T': 0.7073170731707145}
Logits metrics skipped for 166_SMAP_id_23_Sensor_tr_1113_1st_1890.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/166_SMAP_id_23_Sensor_tr_1113_1st_1890_results.pkl
Skipping file: 037_GHL_id_6_Sensor_tr_50000_1st_80001.csv due to filter criteria.
Skipping file: 182_Exathlon_id_9_Facility_tr_10766_1st_12590.csv due to filter criteria.
Skipping file: 113_SVDB_id_30_Medical_tr_4552_1st_4652.csv due to filter criteria.
Skipping file: 040_GHL_id_9_Sensor_tr_50000_1st_92001.csv due to filter criteria.
Processing file: 058_SMD_id_2_Facility_tr_1087_1st_1187.csv, Data shape: (28726, 38)
Label shape: (28726,)
Train data shape: (1087, 38), Test data shape: (27639, 38), Label test shape: (27639,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 058_SMD_id_2_Facility_tr_1087_1st_1187.csv...
Using CUDA device: 2
Data shape: (27639, 38)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=38), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 27639, window_size 5000, adding 2361 samples


  0%|          | 0/6 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


 33%|███▎      | 2/6 [00:01<00:02,  1.92it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


 67%|██████▋   | 4/6 [00:02<00:01,  1.92it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


100%|██████████| 6/6 [00:03<00:00,  1.92it/s]
100%|██████████| 6/6 [00:03<00:00,  1.92it/s]
Original shapes - Output: (30000,), Label: (27639,)
Aligned shapes - Output: (27639,), Label: (27639,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]

Calculating Affiliation Metrics:  29%|██▉       | 58/200 [00:01<00:02, 57.93it/s]

Calculating Affiliation Metrics:  58%|█████▊    | 116/200 [00:02<00:02, 41.97it/s]

Calculating Affiliation Metrics:  80%|████████  | 161/200 [00:03<00:00, 40.61it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:04<00:00, 45.94it/s]
Table1-only metrics completed in 4.49s
Evaluation Result:  {'AUC-PR': 0.5409334463266348, 'AUC-ROC': 0.9483946327718508, 'VUS-PR': 0.48128714098801667, 'VUS-ROC': 0.9145612211697036, 'Standard-F1': 0.5400144196101797, 'Standard-Precision': 0.47405063291139216, 'Standard-Recall': 0.627303182579564, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9207763158627553, 'Affiliation-P': 0.8802410445862808, 'Affiliation-R': 0.9652251203545166, 'F1_T': 0.540960451976913, 'Precision_T': 0.46764346764346737, 'Recall_T': 0.6415410385259627}
Results for 058_SMD_id_2_Facility_tr_1087_1st_1187.csv: {'filename': '058_SMD_id_2_Facility_tr_1087_1st_1187.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '1087', 'data_shape': '28726x38', 'output_length': 30000, 'label_length': 27639, 'aligned_length': 27639, 'AUC-PR': 0.5409334463266348, 'AUC-ROC': 0.9483946327718508, 'VUS-PR': 0.48128714098801667, 'VUS-ROC': 0.9145612211697036, 'Standard-F1': 0.5400144196101797, 'Standard-Precision': 0.47405063291139216, 'Standard-Recall': 0.627303182579564, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9207763158627553, 'Affiliation-P': 0.8802410445862808, 'Affiliation-R': 0.9652251203545166, 'F1_T': 0.540960451976913, 'Precision_T': 0.46764346764346737, 'Recall_T': 0.6415410385259627}
Logits metrics skipped for 058_SMD_id_2_Facility_tr_1087_1st_1187.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/058_SMD_id_2_Facility_tr_1087_1st_1187_results.pkl
Skipping file: 096_SVDB_id_13_Medical_tr_500_1st_444.csv due to filter criteria.
Skipping file: 134_OPPORTUNITY_id_6_HumanActivity_tr_1477_1st_1577.csv due to filter criteria.
Processing file: 158_SMAP_id_15_Sensor_tr_2075_1st_5610.csv, Data shape: (8300, 25)
Label shape: (8300,)
Train data shape: (2075, 25), Test data shape: (6225, 25), Label test shape: (6225,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 158_SMAP_id_15_Sensor_tr_2075_1st_5610.csv...
Using CUDA device: 2
Data shape: (6225, 25)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=25), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 6225, window_size 5000, adding 3775 samples


  0%|          | 0/2 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])

100%|██████████| 2/2 [00:00<00:00,  4.46it/s]
Original shapes - Output: (10000,), Label: (6225,)
Aligned shapes - Output: (6225,), Label: (6225,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:00<00:00, 334.10it/s]
Table1-only metrics completed in 0.64s
Evaluation Result:  {'AUC-PR': 0.38671957833233817, 'AUC-ROC': 0.8115827047877705, 'VUS-PR': 0.3560020957524067, 'VUS-ROC': 0.8115728645441262, 'Standard-F1': 0.43410852713127657, 'Standard-Precision': 0.44444444444443737, 'Standard-Recall': 0.42424242424241787, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9377250423676903, 'Affiliation-P': 0.8858107429718876, 'Affiliation-R': 0.9961032006815139, 'F1_T': 0.43410852713127657, 'Precision_T': 0.44444444444443737, 'Recall_T': 0.42424242424241787}
Results for 158_SMAP_id_15_Sensor_tr_2075_1st_5610.csv: {'filename': '158_SMAP_id_15_Sensor_tr_2075_1st_5610.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '2075', 'data_shape': '8300x25', 'output_length': 10000, 'label_length': 6225, 'aligned_length': 6225, 'AUC-PR': 0.38671957833233817, 'AUC-ROC': 0.8115827047877705, 'VUS-PR': 0.3560020957524067, 'VUS-ROC': 0.8115728645441262, 'Standard-F1': 0.43410852713127657, 'Standard-Precision': 0.44444444444443737, 'Standard-Recall': 0.42424242424241787, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9377250423676903, 'Affiliation-P': 0.8858107429718876, 'Affiliation-R': 0.9961032006815139, 'F1_T': 0.43410852713127657, 'Precision_T': 0.44444444444443737, 'Recall_T': 0.42424242424241787}
Logits metrics skipped for 158_SMAP_id_15_Sensor_tr_2075_1st_5610.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/158_SMAP_id_15_Sensor_tr_2075_1st_5610_results.pkl
Skipping file: 055_GHL_id_24_Sensor_tr_50000_1st_118124.csv due to filter criteria.
Processing file: 144_SMAP_id_1_Sensor_tr_2052_1st_5300.csv, Data shape: (8209, 25)
Label shape: (8209,)
Train data shape: (2052, 25), Test data shape: (6157, 25), Label test shape: (6157,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 144_SMAP_id_1_Sensor_tr_2052_1st_5300.csv...
Using CUDA device: 2
Data shape: (6157, 25)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=25), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 6157, window_size 5000, adding 3843 samples


  0%|          | 0/2 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])

100%|██████████| 2/2 [00:00<00:00,  4.47it/s]
Original shapes - Output: (10000,), Label: (6157,)
Aligned shapes - Output: (6157,), Label: (6157,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:00<00:00, 301.03it/s]
Table1-only metrics completed in 0.70s
Evaluation Result:  {'AUC-PR': 0.18517041501554635, 'AUC-ROC': 0.40753417499268424, 'VUS-PR': 0.1836800297350254, 'VUS-ROC': 0.4075287723218872, 'Standard-F1': 0.32354685646472864, 'Standard-Precision': 0.19402454205939887, 'Standard-Recall': 0.9732381801962525, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9017373042330669, 'Affiliation-P': 0.9275490262853191, 'Affiliation-R': 0.8773232587611962, 'F1_T': 0.3235774773433562, 'Precision_T': 0.1941176470588235, 'Recall_T': 0.9714540588760028}
Results for 144_SMAP_id_1_Sensor_tr_2052_1st_5300.csv: {'filename': '144_SMAP_id_1_Sensor_tr_2052_1st_5300.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '2052', 'data_shape': '8209x25', 'output_length': 10000, 'label_length': 6157, 'aligned_length': 6157, 'AUC-PR': 0.18517041501554635, 'AUC-ROC': 0.40753417499268424, 'VUS-PR': 0.1836800297350254, 'VUS-ROC': 0.4075287723218872, 'Standard-F1': 0.32354685646472864, 'Standard-Precision': 0.19402454205939887, 'Standard-Recall': 0.9732381801962525, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9017373042330669, 'Affiliation-P': 0.9275490262853191, 'Affiliation-R': 0.8773232587611962, 'F1_T': 0.3235774773433562, 'Precision_T': 0.1941176470588235, 'Recall_T': 0.9714540588760028}
Logits metrics skipped for 144_SMAP_id_1_Sensor_tr_2052_1st_5300.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/144_SMAP_id_1_Sensor_tr_2052_1st_5300_results.pkl
Skipping file: 048_GHL_id_17_Sensor_tr_50000_1st_154001.csv due to filter criteria.
Processing file: 008_MSL_id_7_Sensor_tr_656_1st_1630.csv, Data shape: (2625, 55)
Label shape: (2625,)
Train data shape: (656, 55), Test data shape: (1969, 55), Label test shape: (1969,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 008_MSL_id_7_Sensor_tr_656_1st_1630.csv...
Using CUDA device: 2
Data shape: (1969, 55)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=55), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda


  0%|          | 0/1 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 1969, 55])
batched_mask shape: torch.Size([1, 1969])

100%|██████████| 1/1 [00:00<00:00,  5.79it/s]
Original shapes - Output: (1969,), Label: (1969,)
Aligned shapes - Output: (1969,), Label: (1969,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:00<00:00, 635.85it/s]
Table1-only metrics completed in 0.34s
Evaluation Result:  {'AUC-PR': 0.2366805347791258, 'AUC-ROC': 0.8143896615562328, 'VUS-PR': 0.2376081570419249, 'VUS-ROC': 0.8143928845846418, 'Standard-F1': 0.4604316546758672, 'Standard-Precision': 0.3137254901960779, 'Standard-Recall': 0.864864864864861, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8746622066320175, 'Affiliation-P': 0.7778332320502848, 'Affiliation-R': 0.9990265459804922, 'F1_T': 0.4609843937571109, 'Precision_T': 0.3142389525368244, 'Recall_T': 0.864864864864861}
Results for 008_MSL_id_7_Sensor_tr_656_1st_1630.csv: {'filename': '008_MSL_id_7_Sensor_tr_656_1st_1630.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '656', 'data_shape': '2625x55', 'output_length': 1969, 'label_length': 1969, 'aligned_length': 1969, 'AUC-PR': 0.2366805347791258, 'AUC-ROC': 0.8143896615562328, 'VUS-PR': 0.2376081570419249, 'VUS-ROC': 0.8143928845846418, 'Standard-F1': 0.4604316546758672, 'Standard-Precision': 0.3137254901960779, 'Standard-Recall': 0.864864864864861, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8746622066320175, 'Affiliation-P': 0.7778332320502848, 'Affiliation-R': 0.9990265459804922, 'F1_T': 0.4609843937571109, 'Precision_T': 0.3142389525368244, 'Recall_T': 0.864864864864861}
Logits metrics skipped for 008_MSL_id_7_Sensor_tr_656_1st_1630.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/008_MSL_id_7_Sensor_tr_656_1st_1630_results.pkl
Processing file: 014_MSL_id_13_Sensor_tr_1525_1st_4575.csv, Data shape: (6100, 55)
Label shape: (6100,)
Train data shape: (1525, 55), Test data shape: (4575, 55), Label test shape: (4575,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 014_MSL_id_13_Sensor_tr_1525_1st_4575.csv...
Using CUDA device: 2
Data shape: (4575, 55)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=55), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda


  0%|          | 0/1 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 4575, 55])
batched_mask shape: torch.Size([1, 4575])

100%|██████████| 1/1 [00:00<00:00,  1.10it/s]
Original shapes - Output: (4575,), Label: (4575,)
Aligned shapes - Output: (4575,), Label: (4575,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:00<00:00, 227.49it/s]
Table1-only metrics completed in 0.91s
Evaluation Result:  {'AUC-PR': 0.2098938169669743, 'AUC-ROC': 0.8849687041847623, 'VUS-PR': 0.21108860243862615, 'VUS-ROC': 0.8850278003404943, 'Standard-F1': 0.3505535055346092, 'Standard-Precision': 0.26315789473684137, 'Standard-Recall': 0.5248618784530358, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8568775591176377, 'Affiliation-P': 0.7518460441910193, 'Affiliation-R': 0.9960196842073483, 'F1_T': 0.34990791896824675, 'Precision_T': 0.2624309392265186, 'Recall_T': 0.5248618784530358}
Results for 014_MSL_id_13_Sensor_tr_1525_1st_4575.csv: {'filename': '014_MSL_id_13_Sensor_tr_1525_1st_4575.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '1525', 'data_shape': '6100x55', 'output_length': 4575, 'label_length': 4575, 'aligned_length': 4575, 'AUC-PR': 0.2098938169669743, 'AUC-ROC': 0.8849687041847623, 'VUS-PR': 0.21108860243862615, 'VUS-ROC': 0.8850278003404943, 'Standard-F1': 0.3505535055346092, 'Standard-Precision': 0.26315789473684137, 'Standard-Recall': 0.5248618784530358, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8568775591176377, 'Affiliation-P': 0.7518460441910193, 'Affiliation-R': 0.9960196842073483, 'F1_T': 0.34990791896824675, 'Precision_T': 0.2624309392265186, 'Recall_T': 0.5248618784530358}
Logits metrics skipped for 014_MSL_id_13_Sensor_tr_1525_1st_4575.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/014_MSL_id_13_Sensor_tr_1525_1st_4575_results.pkl
Skipping file: 001_Genesis_id_1_Sensor_tr_4055_1st_15538.csv due to filter criteria.
Skipping file: 185_Exathlon_id_12_Facility_tr_11665_1st_13484.csv due to filter criteria.
Skipping file: 117_TAO_id_2_Environment_tr_500_1st_1.csv due to filter criteria.
Processing file: 076_SMD_id_20_Facility_tr_5925_1st_17580.csv, Data shape: (23703, 38)
Label shape: (23703,)
Train data shape: (5925, 38), Test data shape: (17778, 38), Label test shape: (17778,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 076_SMD_id_20_Facility_tr_5925_1st_17580.csv...
Using CUDA device: 2
Data shape: (17778, 38)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=38), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 17778, window_size 5000, adding 2222 samples


  0%|          | 0/4 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


 50%|█████     | 2/4 [00:01<00:01,  1.92it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


100%|██████████| 4/4 [00:02<00:00,  1.92it/s]
100%|██████████| 4/4 [00:02<00:00,  1.92it/s]
Original shapes - Output: (20000,), Label: (17778,)
Aligned shapes - Output: (17778,), Label: (17778,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]

Calculating Affiliation Metrics:  66%|██████▌   | 132/200 [00:01<00:00, 131.63it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:01<00:00, 145.69it/s]
Table1-only metrics completed in 1.45s
Evaluation Result:  {'AUC-PR': 0.9662844578492701, 'AUC-ROC': 0.9997294045101128, 'VUS-PR': 0.9316238215202775, 'VUS-ROC': 0.999586438423925, 'Standard-F1': 0.8967551622413839, 'Standard-Precision': 0.8539325842696581, 'Standard-Recall': 0.9440993788819817, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9956324732433577, 'Affiliation-P': 0.9919010387370158, 'Affiliation-R': 0.9993920883442373, 'F1_T': 0.8967551622413839, 'Precision_T': 0.8539325842696581, 'Recall_T': 0.9440993788819817}
Results for 076_SMD_id_20_Facility_tr_5925_1st_17580.csv: {'filename': '076_SMD_id_20_Facility_tr_5925_1st_17580.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '5925', 'data_shape': '23703x38', 'output_length': 20000, 'label_length': 17778, 'aligned_length': 17778, 'AUC-PR': 0.9662844578492701, 'AUC-ROC': 0.9997294045101128, 'VUS-PR': 0.9316238215202775, 'VUS-ROC': 0.999586438423925, 'Standard-F1': 0.8967551622413839, 'Standard-Precision': 0.8539325842696581, 'Standard-Recall': 0.9440993788819817, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9956324732433577, 'Affiliation-P': 0.9919010387370158, 'Affiliation-R': 0.9993920883442373, 'F1_T': 0.8967551622413839, 'Precision_T': 0.8539325842696581, 'Recall_T': 0.9440993788819817}
Logits metrics skipped for 076_SMD_id_20_Facility_tr_5925_1st_17580.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/076_SMD_id_20_Facility_tr_5925_1st_17580_results.pkl
Skipping file: 038_GHL_id_7_Sensor_tr_50000_1st_100001.csv due to filter criteria.
Skipping file: 103_SVDB_id_20_Medical_tr_12167_1st_12267.csv due to filter criteria.
Skipping file: 082_LTDB_id_4_Medical_tr_4456_1st_4556.csv due to filter criteria.
Skipping file: 188_Exathlon_id_15_Facility_tr_12538_1st_12638.csv due to filter criteria.
Skipping file: 112_SVDB_id_29_Medical_tr_8226_1st_8326.csv due to filter criteria.
Skipping file: 176_Exathlon_id_3_Facility_tr_10766_1st_12590.csv due to filter criteria.
Processing file: 171_SWaT_id_1_Sensor_tr_3749_1st_9522.csv, Data shape: (14996, 66)
Label shape: (14996,)
Train data shape: (3749, 66), Test data shape: (11247, 66), Label test shape: (11247,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 171_SWaT_id_1_Sensor_tr_3749_1st_9522.csv...
Using CUDA device: 2
Data shape: (11247, 66)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=66), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 11247, window_size 5000, adding 3753 samples


  0%|          | 0/3 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 66])
batched_mask shape: torch.Size([1, 5000])


 33%|███▎      | 1/3 [00:01<00:03,  1.57s/it]batched_time_series shape: torch.Size([1, 5000, 66])
batched_mask shape: torch.Size([1, 5000])


 67%|██████▋   | 2/3 [00:03<00:01,  1.57s/it]batched_time_series shape: torch.Size([1, 5000, 66])
batched_mask shape: torch.Size([1, 5000])


100%|██████████| 3/3 [00:04<00:00,  1.57s/it]
100%|██████████| 3/3 [00:04<00:00,  1.57s/it]
Original shapes - Output: (15000,), Label: (11247,)
Aligned shapes - Output: (11247,), Label: (11247,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]

Calculating Affiliation Metrics:  76%|███████▋  | 153/200 [00:01<00:00, 152.54it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:01<00:00, 173.22it/s]
Table1-only metrics completed in 1.22s
Evaluation Result:  {'AUC-PR': 0.16502869297053147, 'AUC-ROC': 0.34762563731674034, 'VUS-PR': 0.1408006212527017, 'VUS-ROC': 0.23810845464375963, 'Standard-F1': 0.29496951219486495, 'Standard-Precision': 0.17379198850368238, 'Standard-Recall': 0.9743202416918425, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.7163140283940767, 'Affiliation-P': 0.5580134427253659, 'Affiliation-R': 1.0, 'F1_T': 0.29496951219486495, 'Precision_T': 0.17379198850368238, 'Recall_T': 0.9743202416918425}
Results for 171_SWaT_id_1_Sensor_tr_3749_1st_9522.csv: {'filename': '171_SWaT_id_1_Sensor_tr_3749_1st_9522.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '3749', 'data_shape': '14996x66', 'output_length': 15000, 'label_length': 11247, 'aligned_length': 11247, 'AUC-PR': 0.16502869297053147, 'AUC-ROC': 0.34762563731674034, 'VUS-PR': 0.1408006212527017, 'VUS-ROC': 0.23810845464375963, 'Standard-F1': 0.29496951219486495, 'Standard-Precision': 0.17379198850368238, 'Standard-Recall': 0.9743202416918425, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.7163140283940767, 'Affiliation-P': 0.5580134427253659, 'Affiliation-R': 1.0, 'F1_T': 0.29496951219486495, 'Precision_T': 0.17379198850368238, 'Recall_T': 0.9743202416918425}
Logits metrics skipped for 171_SWaT_id_1_Sensor_tr_3749_1st_9522.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/171_SWaT_id_1_Sensor_tr_3749_1st_9522_results.pkl
Skipping file: 126_TAO_id_11_Environment_tr_500_1st_7.csv due to filter criteria.
Processing file: 074_SMD_id_18_Facility_tr_7174_1st_21230.csv, Data shape: (28696, 38)
Label shape: (28696,)
Train data shape: (7174, 38), Test data shape: (21522, 38), Label test shape: (21522,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 074_SMD_id_18_Facility_tr_7174_1st_21230.csv...
Using CUDA device: 2
Data shape: (21522, 38)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=38), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 21522, window_size 5000, adding 3478 samples


  0%|          | 0/5 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


 40%|████      | 2/5 [00:01<00:01,  1.92it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


 80%|████████  | 4/5 [00:02<00:00,  1.93it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])

100%|██████████| 5/5 [00:02<00:00,  1.93it/s]
Original shapes - Output: (25000,), Label: (21522,)
Aligned shapes - Output: (21522,), Label: (21522,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]

Calculating Affiliation Metrics:  42%|████▎     | 85/200 [00:01<00:01, 84.00it/s]

Calculating Affiliation Metrics:  85%|████████▌ | 170/200 [00:02<00:00, 76.96it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:02<00:00, 85.50it/s]
Table1-only metrics completed in 2.44s
Evaluation Result:  {'AUC-PR': 0.5263592233105022, 'AUC-ROC': 0.8026117960698713, 'VUS-PR': 0.4620981391435084, 'VUS-ROC': 0.8024797544001258, 'Standard-F1': 0.5458937198062616, 'Standard-Precision': 0.5231481481481457, 'Standard-Recall': 0.5707070707070678, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.931586060639615, 'Affiliation-P': 0.8742274766982064, 'Affiliation-R': 0.9969998103020759, 'F1_T': 0.5458937198062616, 'Precision_T': 0.5231481481481457, 'Recall_T': 0.5707070707070678}
Results for 074_SMD_id_18_Facility_tr_7174_1st_21230.csv: {'filename': '074_SMD_id_18_Facility_tr_7174_1st_21230.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '7174', 'data_shape': '28696x38', 'output_length': 25000, 'label_length': 21522, 'aligned_length': 21522, 'AUC-PR': 0.5263592233105022, 'AUC-ROC': 0.8026117960698713, 'VUS-PR': 0.4620981391435084, 'VUS-ROC': 0.8024797544001258, 'Standard-F1': 0.5458937198062616, 'Standard-Precision': 0.5231481481481457, 'Standard-Recall': 0.5707070707070678, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.931586060639615, 'Affiliation-P': 0.8742274766982064, 'Affiliation-R': 0.9969998103020759, 'F1_T': 0.5458937198062616, 'Precision_T': 0.5231481481481457, 'Recall_T': 0.5707070707070678}
Logits metrics skipped for 074_SMD_id_18_Facility_tr_7174_1st_21230.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/074_SMD_id_18_Facility_tr_7174_1st_21230_results.pkl
Processing file: 168_SMAP_id_25_Sensor_tr_1998_1st_2098.csv, Data shape: (8579, 25)
Label shape: (8579,)
Train data shape: (1998, 25), Test data shape: (6581, 25), Label test shape: (6581,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 168_SMAP_id_25_Sensor_tr_1998_1st_2098.csv...
Using CUDA device: 2
Data shape: (6581, 25)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=25), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 6581, window_size 5000, adding 3419 samples


  0%|          | 0/2 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])

100%|██████████| 2/2 [00:00<00:00,  4.48it/s]
Original shapes - Output: (10000,), Label: (6581,)
Aligned shapes - Output: (6581,), Label: (6581,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]

Calculating Affiliation Metrics:  54%|█████▎    | 107/200 [00:01<00:00, 106.67it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:01<00:00, 135.00it/s]
Table1-only metrics completed in 1.52s
Evaluation Result:  {'AUC-PR': 0.08187782575141103, 'AUC-ROC': 0.5376428998647463, 'VUS-PR': 0.05309890041272264, 'VUS-ROC': 0.5376375899708423, 'Standard-F1': 0.08997955010207896, 'Standard-Precision': 0.04966139954853271, 'Standard-Recall': 0.4782608695652148, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.794917692681444, 'Affiliation-P': 0.6676099235166433, 'Affiliation-R': 0.982218811712344, 'F1_T': 0.0894308943087735, 'Precision_T': 0.04932735426008966, 'Recall_T': 0.4782608695652148}
Results for 168_SMAP_id_25_Sensor_tr_1998_1st_2098.csv: {'filename': '168_SMAP_id_25_Sensor_tr_1998_1st_2098.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '1998', 'data_shape': '8579x25', 'output_length': 10000, 'label_length': 6581, 'aligned_length': 6581, 'AUC-PR': 0.08187782575141103, 'AUC-ROC': 0.5376428998647463, 'VUS-PR': 0.05309890041272264, 'VUS-ROC': 0.5376375899708423, 'Standard-F1': 0.08997955010207896, 'Standard-Precision': 0.04966139954853271, 'Standard-Recall': 0.4782608695652148, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.794917692681444, 'Affiliation-P': 0.6676099235166433, 'Affiliation-R': 0.982218811712344, 'F1_T': 0.0894308943087735, 'Precision_T': 0.04932735426008966, 'Recall_T': 0.4782608695652148}
Logits metrics skipped for 168_SMAP_id_25_Sensor_tr_1998_1st_2098.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/168_SMAP_id_25_Sensor_tr_1998_1st_2098_results.pkl
Processing file: 006_MSL_id_5_Sensor_tr_1150_1st_1250.csv, Data shape: (5054, 55)
Label shape: (5054,)
Train data shape: (1150, 55), Test data shape: (3904, 55), Label test shape: (3904,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 006_MSL_id_5_Sensor_tr_1150_1st_1250.csv...
Using CUDA device: 2
Data shape: (3904, 55)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=55), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda


  0%|          | 0/1 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 3904, 55])
batched_mask shape: torch.Size([1, 3904])

100%|██████████| 1/1 [00:00<00:00,  1.55it/s]
Original shapes - Output: (3904,), Label: (3904,)
Aligned shapes - Output: (3904,), Label: (3904,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]

Calculating Affiliation Metrics:  70%|██████▉   | 139/200 [00:01<00:00, 138.91it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:01<00:00, 168.13it/s]
Table1-only metrics completed in 1.22s
Evaluation Result:  {'AUC-PR': 0.08010601194127631, 'AUC-ROC': 0.38007202897458203, 'VUS-PR': 0.08053191339624383, 'VUS-ROC': 0.37741004923949417, 'Standard-F1': 0.19919028340060732, 'Standard-Precision': 0.11243144424131624, 'Standard-Recall': 0.8723404255319128, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.7102224777077385, 'Affiliation-P': 0.5878704479511405, 'Affiliation-R': 0.8968899828589271, 'F1_T': 0.19899177500643372, 'Precision_T': 0.11207411835026895, 'Recall_T': 0.8865248226950333}
Results for 006_MSL_id_5_Sensor_tr_1150_1st_1250.csv: {'filename': '006_MSL_id_5_Sensor_tr_1150_1st_1250.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '1150', 'data_shape': '5054x55', 'output_length': 3904, 'label_length': 3904, 'aligned_length': 3904, 'AUC-PR': 0.08010601194127631, 'AUC-ROC': 0.38007202897458203, 'VUS-PR': 0.08053191339624383, 'VUS-ROC': 0.37741004923949417, 'Standard-F1': 0.19919028340060732, 'Standard-Precision': 0.11243144424131624, 'Standard-Recall': 0.8723404255319128, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.7102224777077385, 'Affiliation-P': 0.5878704479511405, 'Affiliation-R': 0.8968899828589271, 'F1_T': 0.19899177500643372, 'Precision_T': 0.11207411835026895, 'Recall_T': 0.8865248226950333}
Logits metrics skipped for 006_MSL_id_5_Sensor_tr_1150_1st_1250.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/006_MSL_id_5_Sensor_tr_1150_1st_1250_results.pkl
Skipping file: 036_GHL_id_5_Sensor_tr_50000_1st_67147.csv due to filter criteria.
Skipping file: 021_MITDB_id_3_Medical_tr_50000_1st_57595.csv due to filter criteria.
Skipping file: 054_GHL_id_23_Sensor_tr_50000_1st_135001.csv due to filter criteria.
Processing file: 170_SMAP_id_27_Sensor_tr_2160_1st_4690.csv, Data shape: (8640, 25)
Label shape: (8640,)
Train data shape: (2160, 25), Test data shape: (6480, 25), Label test shape: (6480,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 170_SMAP_id_27_Sensor_tr_2160_1st_4690.csv...
Using CUDA device: 2
Data shape: (6480, 25)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=25), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 6480, window_size 5000, adding 3520 samples


  0%|          | 0/2 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])

100%|██████████| 2/2 [00:00<00:00,  4.48it/s]
Original shapes - Output: (10000,), Label: (6480,)
Aligned shapes - Output: (6480,), Label: (6480,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:00<00:00, 253.33it/s]
Table1-only metrics completed in 0.83s
Evaluation Result:  {'AUC-PR': 0.02456884627939646, 'AUC-ROC': 0.7264130984684726, 'VUS-PR': 0.02404617519754825, 'VUS-ROC': 0.7263220346778274, 'Standard-F1': 0.05452292441130041, 'Standard-Precision': 0.02877697841726617, 'Standard-Recall': 0.5176470588235234, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.7008474847380011, 'Affiliation-P': 0.5394661899676525, 'Affiliation-R': 0.9999963689179375, 'F1_T': 0.05438813349804625, 'Precision_T': 0.02870189171559033, 'Recall_T': 0.5176470588235234}
Results for 170_SMAP_id_27_Sensor_tr_2160_1st_4690.csv: {'filename': '170_SMAP_id_27_Sensor_tr_2160_1st_4690.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '2160', 'data_shape': '8640x25', 'output_length': 10000, 'label_length': 6480, 'aligned_length': 6480, 'AUC-PR': 0.02456884627939646, 'AUC-ROC': 0.7264130984684726, 'VUS-PR': 0.02404617519754825, 'VUS-ROC': 0.7263220346778274, 'Standard-F1': 0.05452292441130041, 'Standard-Precision': 0.02877697841726617, 'Standard-Recall': 0.5176470588235234, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.7008474847380011, 'Affiliation-P': 0.5394661899676525, 'Affiliation-R': 0.9999963689179375, 'F1_T': 0.05438813349804625, 'Precision_T': 0.02870189171559033, 'Recall_T': 0.5176470588235234}
Logits metrics skipped for 170_SMAP_id_27_Sensor_tr_2160_1st_4690.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/170_SMAP_id_27_Sensor_tr_2160_1st_4690_results.pkl
Skipping file: 051_GHL_id_20_Sensor_tr_50000_1st_75110.csv due to filter criteria.
Skipping file: 023_MITDB_id_5_Medical_tr_25000_1st_36913.csv due to filter criteria.
Skipping file: 018_Daphnet_id_1_HumanActivity_tr_9693_1st_20732.csv due to filter criteria.
Skipping file: 124_TAO_id_9_Environment_tr_500_1st_1.csv due to filter criteria.
Skipping file: 034_GHL_id_3_Sensor_tr_50000_1st_122001.csv due to filter criteria.
Processing file: 153_SMAP_id_10_Sensor_tr_1840_1st_4030.csv, Data shape: (7361, 25)
Label shape: (7361,)
Train data shape: (1840, 25), Test data shape: (5521, 25), Label test shape: (5521,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 153_SMAP_id_10_Sensor_tr_1840_1st_4030.csv...
Using CUDA device: 2
Data shape: (5521, 25)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=25), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 5521, window_size 5000, adding 4479 samples


  0%|          | 0/2 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])

100%|██████████| 2/2 [00:00<00:00,  4.48it/s]
Original shapes - Output: (10000,), Label: (5521,)
Aligned shapes - Output: (5521,), Label: (5521,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:00<00:00, 275.13it/s]
Table1-only metrics completed in 0.76s
Evaluation Result:  {'AUC-PR': 0.024672304889211048, 'AUC-ROC': 0.7492033113761795, 'VUS-PR': 0.022869814385021295, 'VUS-ROC': 0.7489384902973117, 'Standard-F1': 0.07499999999981564, 'Standard-Precision': 0.04178272980501381, 'Standard-Recall': 0.3658536585365764, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.68222477692638, 'Affiliation-P': 0.517709519029458, 'Affiliation-R': 1.0, 'F1_T': 0.07092198581542744, 'Precision_T': 0.03926701570680618, 'Recall_T': 0.3658536585365764}
Results for 153_SMAP_id_10_Sensor_tr_1840_1st_4030.csv: {'filename': '153_SMAP_id_10_Sensor_tr_1840_1st_4030.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '1840', 'data_shape': '7361x25', 'output_length': 10000, 'label_length': 5521, 'aligned_length': 5521, 'AUC-PR': 0.024672304889211048, 'AUC-ROC': 0.7492033113761795, 'VUS-PR': 0.022869814385021295, 'VUS-ROC': 0.7489384902973117, 'Standard-F1': 0.07499999999981564, 'Standard-Precision': 0.04178272980501381, 'Standard-Recall': 0.3658536585365764, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.68222477692638, 'Affiliation-P': 0.517709519029458, 'Affiliation-R': 1.0, 'F1_T': 0.07092198581542744, 'Precision_T': 0.03926701570680618, 'Recall_T': 0.3658536585365764}
Logits metrics skipped for 153_SMAP_id_10_Sensor_tr_1840_1st_4030.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/153_SMAP_id_10_Sensor_tr_1840_1st_4030_results.pkl
Skipping file: 025_MITDB_id_7_Medical_tr_37500_1st_88864.csv due to filter criteria.
Skipping file: 042_GHL_id_11_Sensor_tr_50000_1st_150001.csv due to filter criteria.
Skipping file: 132_OPPORTUNITY_id_4_HumanActivity_tr_895_1st_995.csv due to filter criteria.
Processing file: 148_SMAP_id_5_Sensor_tr_2011_1st_5060.csv, Data shape: (8044, 25)
Label shape: (8044,)
Train data shape: (2011, 25), Test data shape: (6033, 25), Label test shape: (6033,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 148_SMAP_id_5_Sensor_tr_2011_1st_5060.csv...
Using CUDA device: 2
Data shape: (6033, 25)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=25), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 6033, window_size 5000, adding 3967 samples


  0%|          | 0/2 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])

100%|██████████| 2/2 [00:00<00:00,  4.47it/s]
Original shapes - Output: (10000,), Label: (6033,)
Aligned shapes - Output: (6033,), Label: (6033,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]

Calculating Affiliation Metrics:  55%|█████▌    | 110/200 [00:01<00:00, 109.23it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:01<00:00, 145.25it/s]
Table1-only metrics completed in 1.41s
Evaluation Result:  {'AUC-PR': 0.8264384788893449, 'AUC-ROC': 0.994247605728298, 'VUS-PR': 0.7890047714037622, 'VUS-ROC': 0.9942830414219636, 'Standard-F1': 0.7499999999994897, 'Standard-Precision': 0.7397260273972502, 'Standard-Recall': 0.7605633802816795, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9990711434049424, 'Affiliation-P': 1.0, 'Affiliation-R': 0.9981440107577338, 'F1_T': 0.7272727272722191, 'Precision_T': 0.7868852459016265, 'Recall_T': 0.6760563380281596}
Results for 148_SMAP_id_5_Sensor_tr_2011_1st_5060.csv: {'filename': '148_SMAP_id_5_Sensor_tr_2011_1st_5060.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '2011', 'data_shape': '8044x25', 'output_length': 10000, 'label_length': 6033, 'aligned_length': 6033, 'AUC-PR': 0.8264384788893449, 'AUC-ROC': 0.994247605728298, 'VUS-PR': 0.7890047714037622, 'VUS-ROC': 0.9942830414219636, 'Standard-F1': 0.7499999999994897, 'Standard-Precision': 0.7397260273972502, 'Standard-Recall': 0.7605633802816795, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9990711434049424, 'Affiliation-P': 1.0, 'Affiliation-R': 0.9981440107577338, 'F1_T': 0.7272727272722191, 'Precision_T': 0.7868852459016265, 'Recall_T': 0.6760563380281596}
Logits metrics skipped for 148_SMAP_id_5_Sensor_tr_2011_1st_5060.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/148_SMAP_id_5_Sensor_tr_2011_1st_5060_results.pkl
Skipping file: 093_SVDB_id_10_Medical_tr_500_1st_37.csv due to filter criteria.
Skipping file: 200_Exathlon_id_27_Facility_tr_10766_1st_12590.csv due to filter criteria.
Skipping file: 041_GHL_id_10_Sensor_tr_50000_1st_57001.csv due to filter criteria.
Skipping file: 043_GHL_id_12_Sensor_tr_39938_1st_40038.csv due to filter criteria.
Skipping file: 178_Exathlon_id_5_Facility_tr_12538_1st_12638.csv due to filter criteria.
Skipping file: 079_LTDB_id_1_Medical_tr_3618_1st_3718.csv due to filter criteria.
Skipping file: 195_Exathlon_id_22_Facility_tr_10766_1st_12590.csv due to filter criteria.
Processing file: 004_MSL_id_3_Sensor_tr_530_1st_630.csv, Data shape: (2430, 55)
Label shape: (2430,)
Train data shape: (530, 55), Test data shape: (1900, 55), Label test shape: (1900,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 004_MSL_id_3_Sensor_tr_530_1st_630.csv...
Using CUDA device: 2
Data shape: (1900, 55)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=55), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda


  0%|          | 0/1 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 1900, 55])
batched_mask shape: torch.Size([1, 1900])

100%|██████████| 1/1 [00:00<00:00,  6.21it/s]
Original shapes - Output: (1900,), Label: (1900,)
Aligned shapes - Output: (1900,), Label: (1900,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:00<00:00, 436.77it/s]
Table1-only metrics completed in 0.48s
Evaluation Result:  {'AUC-PR': 0.25912796220275547, 'AUC-ROC': 0.7634152346707918, 'VUS-PR': 0.25007257581133463, 'VUS-ROC': 0.7634152346707919, 'Standard-F1': 0.4028776978412321, 'Standard-Precision': 0.35668789808916973, 'Standard-Recall': 0.4628099173553681, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8508306737162313, 'Affiliation-P': 0.749578947368421, 'Affiliation-R': 0.9837081339712919, 'F1_T': 0.39857651245502285, 'Precision_T': 0.3499999999999978, 'Recall_T': 0.4628099173553681}
Results for 004_MSL_id_3_Sensor_tr_530_1st_630.csv: {'filename': '004_MSL_id_3_Sensor_tr_530_1st_630.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '530', 'data_shape': '2430x55', 'output_length': 1900, 'label_length': 1900, 'aligned_length': 1900, 'AUC-PR': 0.25912796220275547, 'AUC-ROC': 0.7634152346707918, 'VUS-PR': 0.25007257581133463, 'VUS-ROC': 0.7634152346707919, 'Standard-F1': 0.4028776978412321, 'Standard-Precision': 0.35668789808916973, 'Standard-Recall': 0.4628099173553681, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8508306737162313, 'Affiliation-P': 0.749578947368421, 'Affiliation-R': 0.9837081339712919, 'F1_T': 0.39857651245502285, 'Precision_T': 0.3499999999999978, 'Recall_T': 0.4628099173553681}
Logits metrics skipped for 004_MSL_id_3_Sensor_tr_530_1st_630.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/004_MSL_id_3_Sensor_tr_530_1st_630_results.pkl
Processing file: 151_SMAP_id_8_Sensor_tr_1971_1st_4870.csv, Data shape: (7884, 25)
Label shape: (7884,)
Train data shape: (1971, 25), Test data shape: (5913, 25), Label test shape: (5913,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 151_SMAP_id_8_Sensor_tr_1971_1st_4870.csv...
Using CUDA device: 2
Data shape: (5913, 25)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=25), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 5913, window_size 5000, adding 4087 samples


  0%|          | 0/2 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])

100%|██████████| 2/2 [00:00<00:00,  4.47it/s]
Original shapes - Output: (10000,), Label: (5913,)
Aligned shapes - Output: (5913,), Label: (5913,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:00<00:00, 221.95it/s]
Table1-only metrics completed in 0.94s
Evaluation Result:  {'AUC-PR': 0.02606859261881407, 'AUC-ROC': 0.6675663432064896, 'VUS-PR': 0.02620647617175558, 'VUS-ROC': 0.6676192653558908, 'Standard-F1': 0.07102803738292035, 'Standard-Precision': 0.04185022026431709, 'Standard-Recall': 0.23456790123456503, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.6712116285169201, 'Affiliation-P': 0.5317407971137044, 'Affiliation-R': 0.9098596313208185, 'F1_T': 0.06944444444413943, 'Precision_T': 0.04273504273504261, 'Recall_T': 0.1851851851851829}
Results for 151_SMAP_id_8_Sensor_tr_1971_1st_4870.csv: {'filename': '151_SMAP_id_8_Sensor_tr_1971_1st_4870.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '1971', 'data_shape': '7884x25', 'output_length': 10000, 'label_length': 5913, 'aligned_length': 5913, 'AUC-PR': 0.02606859261881407, 'AUC-ROC': 0.6675663432064896, 'VUS-PR': 0.02620647617175558, 'VUS-ROC': 0.6676192653558908, 'Standard-F1': 0.07102803738292035, 'Standard-Precision': 0.04185022026431709, 'Standard-Recall': 0.23456790123456503, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.6712116285169201, 'Affiliation-P': 0.5317407971137044, 'Affiliation-R': 0.9098596313208185, 'F1_T': 0.06944444444413943, 'Precision_T': 0.04273504273504261, 'Recall_T': 0.1851851851851829}
Logits metrics skipped for 151_SMAP_id_8_Sensor_tr_1971_1st_4870.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/151_SMAP_id_8_Sensor_tr_1971_1st_4870_results.pkl
Skipping file: 026_MITDB_id_8_Medical_tr_30495_1st_30595.csv due to filter criteria.
Skipping file: 121_TAO_id_6_Environment_tr_500_1st_7.csv due to filter criteria.
Processing file: 163_SMAP_id_20_Sensor_tr_2051_1st_4575.csv, Data shape: (8205, 25)
Label shape: (8205,)
Train data shape: (2051, 25), Test data shape: (6154, 25), Label test shape: (6154,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 163_SMAP_id_20_Sensor_tr_2051_1st_4575.csv...
Using CUDA device: 2
Data shape: (6154, 25)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=25), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 6154, window_size 5000, adding 3846 samples


  0%|          | 0/2 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])

100%|██████████| 2/2 [00:00<00:00,  4.47it/s]
Original shapes - Output: (10000,), Label: (6154,)
Aligned shapes - Output: (6154,), Label: (6154,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:00<00:00, 267.06it/s]
Table1-only metrics completed in 0.79s
Evaluation Result:  {'AUC-PR': 0.027132371271130495, 'AUC-ROC': 0.49611323113955436, 'VUS-PR': 0.02718853639248501, 'VUS-ROC': 0.49626007163654184, 'Standard-F1': 0.07061224489788613, 'Standard-Precision': 0.036699193890538814, 'Standard-Recall': 0.9301075268817155, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.7120033034800137, 'Affiliation-P': 0.5528007490228267, 'Affiliation-R': 0.9999943213785246, 'F1_T': 0.07041107041099756, 'Precision_T': 0.03659052453468697, 'Recall_T': 0.9301075268817155}
Results for 163_SMAP_id_20_Sensor_tr_2051_1st_4575.csv: {'filename': '163_SMAP_id_20_Sensor_tr_2051_1st_4575.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '2051', 'data_shape': '8205x25', 'output_length': 10000, 'label_length': 6154, 'aligned_length': 6154, 'AUC-PR': 0.027132371271130495, 'AUC-ROC': 0.49611323113955436, 'VUS-PR': 0.02718853639248501, 'VUS-ROC': 0.49626007163654184, 'Standard-F1': 0.07061224489788613, 'Standard-Precision': 0.036699193890538814, 'Standard-Recall': 0.9301075268817155, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.7120033034800137, 'Affiliation-P': 0.5528007490228267, 'Affiliation-R': 0.9999943213785246, 'F1_T': 0.07041107041099756, 'Precision_T': 0.03659052453468697, 'Recall_T': 0.9301075268817155}
Logits metrics skipped for 163_SMAP_id_20_Sensor_tr_2051_1st_4575.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/163_SMAP_id_20_Sensor_tr_2051_1st_4575_results.pkl
Processing file: 016_MSL_id_15_Sensor_tr_500_1st_780.csv, Data shape: (1096, 55)
Label shape: (1096,)
Train data shape: (500, 55), Test data shape: (596, 55), Label test shape: (596,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 016_MSL_id_15_Sensor_tr_500_1st_780.csv...
Using CUDA device: 2
Data shape: (596, 55)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=55), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda


  0%|          | 0/1 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 596, 55])
batched_mask shape: torch.Size([1, 596])

100%|██████████| 1/1 [00:00<00:00, 44.32it/s]
Original shapes - Output: (596,), Label: (596,)
Aligned shapes - Output: (596,), Label: (596,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:00<00:00, 1429.96it/s]
Table1-only metrics completed in 0.16s
Evaluation Result:  {'AUC-PR': 0.5117273828833482, 'AUC-ROC': 0.7491514167650531, 'VUS-PR': 0.5094053702954286, 'VUS-ROC': 0.749049955726092, 'Standard-F1': 0.5232558139530279, 'Standard-Precision': 0.7499999999999875, 'Standard-Recall': 0.4017857142857107, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8859913327780787, 'Affiliation-P': 0.8107604337255897, 'Affiliation-R': 0.976611656470767, 'F1_T': 0.5172413793098801, 'Precision_T': 0.7258064516128915, 'Recall_T': 0.4017857142857107}
Results for 016_MSL_id_15_Sensor_tr_500_1st_780.csv: {'filename': '016_MSL_id_15_Sensor_tr_500_1st_780.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '500', 'data_shape': '1096x55', 'output_length': 596, 'label_length': 596, 'aligned_length': 596, 'AUC-PR': 0.5117273828833482, 'AUC-ROC': 0.7491514167650531, 'VUS-PR': 0.5094053702954286, 'VUS-ROC': 0.749049955726092, 'Standard-F1': 0.5232558139530279, 'Standard-Precision': 0.7499999999999875, 'Standard-Recall': 0.4017857142857107, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8859913327780787, 'Affiliation-P': 0.8107604337255897, 'Affiliation-R': 0.976611656470767, 'F1_T': 0.5172413793098801, 'Precision_T': 0.7258064516128915, 'Recall_T': 0.4017857142857107}
Logits metrics skipped for 016_MSL_id_15_Sensor_tr_500_1st_780.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/016_MSL_id_15_Sensor_tr_500_1st_780_results.pkl
Skipping file: 100_SVDB_id_17_Medical_tr_17803_1st_17903.csv due to filter criteria.
Skipping file: 140_CATSv2_id_3_Sensor_tr_28307_1st_28407.csv due to filter criteria.
Skipping file: 114_SVDB_id_31_Medical_tr_25000_1st_63090.csv due to filter criteria.
Processing file: 154_SMAP_id_11_Sensor_tr_2117_1st_4770.csv, Data shape: (8469, 25)
Label shape: (8469,)
Train data shape: (2117, 25), Test data shape: (6352, 25), Label test shape: (6352,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 154_SMAP_id_11_Sensor_tr_2117_1st_4770.csv...
Using CUDA device: 2
Data shape: (6352, 25)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=25), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 6352, window_size 5000, adding 3648 samples


  0%|          | 0/2 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])

100%|██████████| 2/2 [00:00<00:00,  4.47it/s]
Original shapes - Output: (10000,), Label: (6352,)
Aligned shapes - Output: (6352,), Label: (6352,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:00<00:00, 243.63it/s]
Table1-only metrics completed in 0.86s
Evaluation Result:  {'AUC-PR': 0.015677813481369107, 'AUC-ROC': 0.43973149448704224, 'VUS-PR': 0.01570566799973673, 'VUS-ROC': 0.43988402429335594, 'Standard-F1': 0.03812824956668702, 'Standard-Precision': 0.019434628975265014, 'Standard-Recall': 0.9999999999999918, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.6822638019830505, 'Affiliation-P': 0.5181422876945093, 'Affiliation-R': 0.9985564512771405, 'F1_T': 0.03813425780015168, 'Precision_T': 0.01943775100401606, 'Recall_T': 0.9999999999999918}
Results for 154_SMAP_id_11_Sensor_tr_2117_1st_4770.csv: {'filename': '154_SMAP_id_11_Sensor_tr_2117_1st_4770.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '2117', 'data_shape': '8469x25', 'output_length': 10000, 'label_length': 6352, 'aligned_length': 6352, 'AUC-PR': 0.015677813481369107, 'AUC-ROC': 0.43973149448704224, 'VUS-PR': 0.01570566799973673, 'VUS-ROC': 0.43988402429335594, 'Standard-F1': 0.03812824956668702, 'Standard-Precision': 0.019434628975265014, 'Standard-Recall': 0.9999999999999918, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.6822638019830505, 'Affiliation-P': 0.5181422876945093, 'Affiliation-R': 0.9985564512771405, 'F1_T': 0.03813425780015168, 'Precision_T': 0.01943775100401606, 'Recall_T': 0.9999999999999918}
Logits metrics skipped for 154_SMAP_id_11_Sensor_tr_2117_1st_4770.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/154_SMAP_id_11_Sensor_tr_2117_1st_4770_results.pkl
Skipping file: 107_SVDB_id_24_Medical_tr_32805_1st_32905.csv due to filter criteria.
Processing file: 145_SMAP_id_2_Sensor_tr_2133_1st_5400.csv, Data shape: (8532, 25)
Label shape: (8532,)
Train data shape: (2133, 25), Test data shape: (6399, 25), Label test shape: (6399,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 145_SMAP_id_2_Sensor_tr_2133_1st_5400.csv...
Using CUDA device: 2
Data shape: (6399, 25)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=25), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 6399, window_size 5000, adding 3601 samples


  0%|          | 0/2 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])

100%|██████████| 2/2 [00:00<00:00,  4.48it/s]
Original shapes - Output: (10000,), Label: (6399,)
Aligned shapes - Output: (6399,), Label: (6399,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:00<00:00, 339.26it/s]
Table1-only metrics completed in 0.63s
Evaluation Result:  {'AUC-PR': 0.1843395859427477, 'AUC-ROC': 0.6261673087953473, 'VUS-PR': 0.18209107170311623, 'VUS-ROC': 0.6261082555590632, 'Standard-F1': 0.2805290679787734, 'Standard-Precision': 0.17351598173515975, 'Standard-Recall': 0.7319422150882813, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9016369666006372, 'Affiliation-P': 0.8520033906813091, 'Affiliation-R': 0.9574110822392242, 'F1_T': 0.28003668602842885, 'Precision_T': 0.17296072507552865, 'Recall_T': 0.7351524879614755}
Results for 145_SMAP_id_2_Sensor_tr_2133_1st_5400.csv: {'filename': '145_SMAP_id_2_Sensor_tr_2133_1st_5400.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '2133', 'data_shape': '8532x25', 'output_length': 10000, 'label_length': 6399, 'aligned_length': 6399, 'AUC-PR': 0.1843395859427477, 'AUC-ROC': 0.6261673087953473, 'VUS-PR': 0.18209107170311623, 'VUS-ROC': 0.6261082555590632, 'Standard-F1': 0.2805290679787734, 'Standard-Precision': 0.17351598173515975, 'Standard-Recall': 0.7319422150882813, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9016369666006372, 'Affiliation-P': 0.8520033906813091, 'Affiliation-R': 0.9574110822392242, 'F1_T': 0.28003668602842885, 'Precision_T': 0.17296072507552865, 'Recall_T': 0.7351524879614755}
Logits metrics skipped for 145_SMAP_id_2_Sensor_tr_2133_1st_5400.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/145_SMAP_id_2_Sensor_tr_2133_1st_5400_results.pkl
Processing file: 012_MSL_id_11_Sensor_tr_539_1st_940.csv, Data shape: (2156, 55)
Label shape: (2156,)
Train data shape: (539, 55), Test data shape: (1617, 55), Label test shape: (1617,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 012_MSL_id_11_Sensor_tr_539_1st_940.csv...
Using CUDA device: 2
Data shape: (1617, 55)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=55), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda


  0%|          | 0/1 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 1617, 55])
batched_mask shape: torch.Size([1, 1617])

100%|██████████| 1/1 [00:00<00:00,  8.39it/s]
Original shapes - Output: (1617,), Label: (1617,)
Aligned shapes - Output: (1617,), Label: (1617,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:00<00:00, 642.44it/s]
Table1-only metrics completed in 0.33s
Evaluation Result:  {'AUC-PR': 0.4734469141644832, 'AUC-ROC': 0.8998406436949763, 'VUS-PR': 0.4699248589212237, 'VUS-ROC': 0.8997655372397398, 'Standard-F1': 0.5599999999995054, 'Standard-Precision': 0.6621621621621533, 'Standard-Recall': 0.48514851485148036, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9273899803156734, 'Affiliation-P': 0.871063824416594, 'Affiliation-R': 0.9915042524660629, 'F1_T': 0.5549132947971954, 'Precision_T': 0.6666666666666574, 'Recall_T': 0.47524752475247056}
Results for 012_MSL_id_11_Sensor_tr_539_1st_940.csv: {'filename': '012_MSL_id_11_Sensor_tr_539_1st_940.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '539', 'data_shape': '2156x55', 'output_length': 1617, 'label_length': 1617, 'aligned_length': 1617, 'AUC-PR': 0.4734469141644832, 'AUC-ROC': 0.8998406436949763, 'VUS-PR': 0.4699248589212237, 'VUS-ROC': 0.8997655372397398, 'Standard-F1': 0.5599999999995054, 'Standard-Precision': 0.6621621621621533, 'Standard-Recall': 0.48514851485148036, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9273899803156734, 'Affiliation-P': 0.871063824416594, 'Affiliation-R': 0.9915042524660629, 'F1_T': 0.5549132947971954, 'Precision_T': 0.6666666666666574, 'Recall_T': 0.47524752475247056}
Logits metrics skipped for 012_MSL_id_11_Sensor_tr_539_1st_940.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/012_MSL_id_11_Sensor_tr_539_1st_940_results.pkl
Processing file: 013_MSL_id_12_Sensor_tr_554_1st_1200.csv, Data shape: (2218, 55)
Label shape: (2218,)
Train data shape: (554, 55), Test data shape: (1664, 55), Label test shape: (1664,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 013_MSL_id_12_Sensor_tr_554_1st_1200.csv...
Using CUDA device: 2
Data shape: (1664, 55)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=55), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda


  0%|          | 0/1 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 1664, 55])
batched_mask shape: torch.Size([1, 1664])

100%|██████████| 1/1 [00:00<00:00,  8.18it/s]
Original shapes - Output: (1664,), Label: (1664,)
Aligned shapes - Output: (1664,), Label: (1664,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:00<00:00, 473.84it/s]
Table1-only metrics completed in 0.44s
Evaluation Result:  {'AUC-PR': 0.08965356402314749, 'AUC-ROC': 0.7212595097210481, 'VUS-PR': 0.0906917296499225, 'VUS-ROC': 0.7213886540809619, 'Standard-F1': 0.24242424242375757, 'Standard-Precision': 0.199999999999995, 'Standard-Recall': 0.3076923076922959, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.7094944953267922, 'Affiliation-P': 0.5507597870879121, 'Affiliation-R': 0.9967756102071005, 'F1_T': 0.2352941176465796, 'Precision_T': 0.19047619047618594, 'Recall_T': 0.3076923076922959}
Results for 013_MSL_id_12_Sensor_tr_554_1st_1200.csv: {'filename': '013_MSL_id_12_Sensor_tr_554_1st_1200.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '554', 'data_shape': '2218x55', 'output_length': 1664, 'label_length': 1664, 'aligned_length': 1664, 'AUC-PR': 0.08965356402314749, 'AUC-ROC': 0.7212595097210481, 'VUS-PR': 0.0906917296499225, 'VUS-ROC': 0.7213886540809619, 'Standard-F1': 0.24242424242375757, 'Standard-Precision': 0.199999999999995, 'Standard-Recall': 0.3076923076922959, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.7094944953267922, 'Affiliation-P': 0.5507597870879121, 'Affiliation-R': 0.9967756102071005, 'F1_T': 0.2352941176465796, 'Precision_T': 0.19047619047618594, 'Recall_T': 0.3076923076922959}
Logits metrics skipped for 013_MSL_id_12_Sensor_tr_554_1st_1200.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/013_MSL_id_12_Sensor_tr_554_1st_1200_results.pkl
Skipping file: 087_SVDB_id_4_Medical_tr_5421_1st_5521.csv due to filter criteria.
Processing file: 078_SMD_id_22_Facility_tr_500_1st_326.csv, Data shape: (23693, 38)
Label shape: (23693,)
Train data shape: (500, 38), Test data shape: (23193, 38), Label test shape: (23193,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 078_SMD_id_22_Facility_tr_500_1st_326.csv...
Using CUDA device: 2
Data shape: (23193, 38)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=38), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 23193, window_size 5000, adding 1807 samples


  0%|          | 0/5 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


 40%|████      | 2/5 [00:01<00:01,  1.92it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


 80%|████████  | 4/5 [00:02<00:00,  1.92it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])

100%|██████████| 5/5 [00:02<00:00,  1.92it/s]
Original shapes - Output: (25000,), Label: (23193,)
Aligned shapes - Output: (23193,), Label: (23193,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]

Calculating Affiliation Metrics:  36%|███▋      | 73/200 [00:01<00:01, 71.70it/s]

Calculating Affiliation Metrics:  72%|███████▎  | 145/200 [00:02<00:00, 58.56it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:03<00:00, 65.05it/s]
Table1-only metrics completed in 3.20s
Evaluation Result:  {'AUC-PR': 0.8698155615683433, 'AUC-ROC': 0.9661795530064666, 'VUS-PR': 0.8522532477265448, 'VUS-ROC': 0.965628733025895, 'Standard-F1': 0.8520408163260319, 'Standard-Precision': 0.9115720524017457, 'Standard-Recall': 0.7998084291187733, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8987934907656626, 'Affiliation-P': 0.8196696608219357, 'Affiliation-R': 0.9948253369426046, 'F1_T': 0.8511722731901229, 'Precision_T': 0.9095860566448791, 'Recall_T': 0.7998084291187733}
Results for 078_SMD_id_22_Facility_tr_500_1st_326.csv: {'filename': '078_SMD_id_22_Facility_tr_500_1st_326.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '500', 'data_shape': '23693x38', 'output_length': 25000, 'label_length': 23193, 'aligned_length': 23193, 'AUC-PR': 0.8698155615683433, 'AUC-ROC': 0.9661795530064666, 'VUS-PR': 0.8522532477265448, 'VUS-ROC': 0.965628733025895, 'Standard-F1': 0.8520408163260319, 'Standard-Precision': 0.9115720524017457, 'Standard-Recall': 0.7998084291187733, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8987934907656626, 'Affiliation-P': 0.8196696608219357, 'Affiliation-R': 0.9948253369426046, 'F1_T': 0.8511722731901229, 'Precision_T': 0.9095860566448791, 'Recall_T': 0.7998084291187733}
Logits metrics skipped for 078_SMD_id_22_Facility_tr_500_1st_326.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/078_SMD_id_22_Facility_tr_500_1st_326_results.pkl
Skipping file: 039_GHL_id_8_Sensor_tr_50000_1st_63030.csv due to filter criteria.
Skipping file: 135_OPPORTUNITY_id_7_HumanActivity_tr_2085_1st_2185.csv due to filter criteria.
Skipping file: 118_TAO_id_3_Environment_tr_500_1st_7.csv due to filter criteria.
Skipping file: 128_TAO_id_13_Environment_tr_500_1st_7.csv due to filter criteria.
Processing file: 011_MSL_id_10_Sensor_tr_1525_1st_4590.csv, Data shape: (6100, 55)
Label shape: (6100,)
Train data shape: (1525, 55), Test data shape: (4575, 55), Label test shape: (4575,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 011_MSL_id_10_Sensor_tr_1525_1st_4590.csv...
Using CUDA device: 2
Data shape: (4575, 55)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=55), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda


  0%|          | 0/1 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 4575, 55])
batched_mask shape: torch.Size([1, 4575])

100%|██████████| 1/1 [00:00<00:00,  1.10it/s]
Original shapes - Output: (4575,), Label: (4575,)
Aligned shapes - Output: (4575,), Label: (4575,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:00<00:00, 229.85it/s]
Table1-only metrics completed in 0.90s
Evaluation Result:  {'AUC-PR': 0.2101842810142521, 'AUC-ROC': 0.92811132258264, 'VUS-PR': 0.21156998550519146, 'VUS-ROC': 0.9281199112277638, 'Standard-F1': 0.3681592039796583, 'Standard-Precision': 0.27306273062730524, 'Standard-Recall': 0.5648854961832018, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8569986759266424, 'Affiliation-P': 0.7509843985111269, 'Affiliation-R': 0.9978642639636257, 'F1_T': 0.3681592039796583, 'Precision_T': 0.27306273062730524, 'Recall_T': 0.5648854961832018}
Results for 011_MSL_id_10_Sensor_tr_1525_1st_4590.csv: {'filename': '011_MSL_id_10_Sensor_tr_1525_1st_4590.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '1525', 'data_shape': '6100x55', 'output_length': 4575, 'label_length': 4575, 'aligned_length': 4575, 'AUC-PR': 0.2101842810142521, 'AUC-ROC': 0.92811132258264, 'VUS-PR': 0.21156998550519146, 'VUS-ROC': 0.9281199112277638, 'Standard-F1': 0.3681592039796583, 'Standard-Precision': 0.27306273062730524, 'Standard-Recall': 0.5648854961832018, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8569986759266424, 'Affiliation-P': 0.7509843985111269, 'Affiliation-R': 0.9978642639636257, 'F1_T': 0.3681592039796583, 'Precision_T': 0.27306273062730524, 'Recall_T': 0.5648854961832018}
Logits metrics skipped for 011_MSL_id_10_Sensor_tr_1525_1st_4590.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/011_MSL_id_10_Sensor_tr_1525_1st_4590_results.pkl
Processing file: 002_MSL_id_1_Sensor_tr_500_1st_900.csv, Data shape: (1827, 55)
Label shape: (1827,)
Train data shape: (500, 55), Test data shape: (1327, 55), Label test shape: (1327,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 002_MSL_id_1_Sensor_tr_500_1st_900.csv...
Using CUDA device: 2
Data shape: (1327, 55)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=55), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda


  0%|          | 0/1 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 1327, 55])
batched_mask shape: torch.Size([1, 1327])

100%|██████████| 1/1 [00:00<00:00, 12.26it/s]
Original shapes - Output: (1327,), Label: (1327,)
Aligned shapes - Output: (1327,), Label: (1327,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:00<00:00, 695.74it/s]
Table1-only metrics completed in 0.31s
Evaluation Result:  {'AUC-PR': 0.11000260851141755, 'AUC-ROC': 0.9595192042000552, 'VUS-PR': 0.11660982854241395, 'VUS-ROC': 0.9595537441282123, 'Standard-F1': 0.23728813559290893, 'Standard-Precision': 0.1458333333333303, 'Standard-Recall': 0.6363636363635785, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.7355568811946976, 'Affiliation-P': 0.5818515608715868, 'Affiliation-R': 0.9996232102486814, 'F1_T': 0.21874999999970848, 'Precision_T': 0.1320754716981107, 'Recall_T': 0.6363636363635785}
Results for 002_MSL_id_1_Sensor_tr_500_1st_900.csv: {'filename': '002_MSL_id_1_Sensor_tr_500_1st_900.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '500', 'data_shape': '1827x55', 'output_length': 1327, 'label_length': 1327, 'aligned_length': 1327, 'AUC-PR': 0.11000260851141755, 'AUC-ROC': 0.9595192042000552, 'VUS-PR': 0.11660982854241395, 'VUS-ROC': 0.9595537441282123, 'Standard-F1': 0.23728813559290893, 'Standard-Precision': 0.1458333333333303, 'Standard-Recall': 0.6363636363635785, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.7355568811946976, 'Affiliation-P': 0.5818515608715868, 'Affiliation-R': 0.9996232102486814, 'F1_T': 0.21874999999970848, 'Precision_T': 0.1320754716981107, 'Recall_T': 0.6363636363635785}
Logits metrics skipped for 002_MSL_id_1_Sensor_tr_500_1st_900.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/002_MSL_id_1_Sensor_tr_500_1st_900_results.pkl
Processing file: 164_SMAP_id_21_Sensor_tr_1976_1st_4200.csv, Data shape: (7907, 25)
Label shape: (7907,)
Train data shape: (1976, 25), Test data shape: (5931, 25), Label test shape: (5931,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 164_SMAP_id_21_Sensor_tr_1976_1st_4200.csv...
Using CUDA device: 2
Data shape: (5931, 25)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=25), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 5931, window_size 5000, adding 4069 samples


  0%|          | 0/2 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])

100%|██████████| 2/2 [00:00<00:00,  4.47it/s]
Original shapes - Output: (10000,), Label: (5931,)
Aligned shapes - Output: (5931,), Label: (5931,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:00<00:00, 252.10it/s]
Table1-only metrics completed in 0.83s
Evaluation Result:  {'AUC-PR': 0.018144055055152306, 'AUC-ROC': 0.7155895691609978, 'VUS-PR': 0.01679525058923084, 'VUS-ROC': 0.715532879818594, 'Standard-F1': 0.04187604690109068, 'Standard-Precision': 0.021872265966754137, 'Standard-Recall': 0.4901960784313629, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.6712932078898599, 'Affiliation-P': 0.5052229821327017, 'Affiliation-R': 1.0, 'F1_T': 0.0419815281275418, 'Precision_T': 0.02192982456140349, 'Recall_T': 0.4901960784313629}
Results for 164_SMAP_id_21_Sensor_tr_1976_1st_4200.csv: {'filename': '164_SMAP_id_21_Sensor_tr_1976_1st_4200.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '1976', 'data_shape': '7907x25', 'output_length': 10000, 'label_length': 5931, 'aligned_length': 5931, 'AUC-PR': 0.018144055055152306, 'AUC-ROC': 0.7155895691609978, 'VUS-PR': 0.01679525058923084, 'VUS-ROC': 0.715532879818594, 'Standard-F1': 0.04187604690109068, 'Standard-Precision': 0.021872265966754137, 'Standard-Recall': 0.4901960784313629, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.6712932078898599, 'Affiliation-P': 0.5052229821327017, 'Affiliation-R': 1.0, 'F1_T': 0.0419815281275418, 'Precision_T': 0.02192982456140349, 'Recall_T': 0.4901960784313629}
Logits metrics skipped for 164_SMAP_id_21_Sensor_tr_1976_1st_4200.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/164_SMAP_id_21_Sensor_tr_1976_1st_4200_results.pkl
Skipping file: 020_MITDB_id_2_Medical_tr_50000_1st_52315.csv due to filter criteria.
Processing file: 156_SMAP_id_13_Sensor_tr_1173_1st_2750.csv, Data shape: (4693, 25)
Label shape: (4693,)
Train data shape: (1173, 25), Test data shape: (3520, 25), Label test shape: (3520,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 156_SMAP_id_13_Sensor_tr_1173_1st_2750.csv...
Using CUDA device: 2
Data shape: (3520, 25)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=25), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda


  0%|          | 0/1 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 3520, 25])
batched_mask shape: torch.Size([1, 3520])

100%|██████████| 1/1 [00:00<00:00,  8.76it/s]
Original shapes - Output: (3520,), Label: (3520,)
Aligned shapes - Output: (3520,), Label: (3520,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:00<00:00, 202.12it/s]
Table1-only metrics completed in 1.02s
Evaluation Result:  {'AUC-PR': 0.7390697040046639, 'AUC-ROC': 0.982590903181682, 'VUS-PR': 0.7040378529568697, 'VUS-ROC': 0.9826361216149764, 'Standard-F1': 0.6666666666661605, 'Standard-Precision': 0.6031746031745936, 'Standard-Recall': 0.7450980392156716, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9991023589560597, 'Affiliation-P': 1.0, 'Affiliation-R': 0.9982063279857397, 'F1_T': 0.6611570247928897, 'Precision_T': 0.5714285714285633, 'Recall_T': 0.7843137254901806}
Results for 156_SMAP_id_13_Sensor_tr_1173_1st_2750.csv: {'filename': '156_SMAP_id_13_Sensor_tr_1173_1st_2750.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '1173', 'data_shape': '4693x25', 'output_length': 3520, 'label_length': 3520, 'aligned_length': 3520, 'AUC-PR': 0.7390697040046639, 'AUC-ROC': 0.982590903181682, 'VUS-PR': 0.7040378529568697, 'VUS-ROC': 0.9826361216149764, 'Standard-F1': 0.6666666666661605, 'Standard-Precision': 0.6031746031745936, 'Standard-Recall': 0.7450980392156716, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9991023589560597, 'Affiliation-P': 1.0, 'Affiliation-R': 0.9982063279857397, 'F1_T': 0.6611570247928897, 'Precision_T': 0.5714285714285633, 'Recall_T': 0.7843137254901806}
Logits metrics skipped for 156_SMAP_id_13_Sensor_tr_1173_1st_2750.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/156_SMAP_id_13_Sensor_tr_1173_1st_2750_results.pkl
Skipping file: 053_GHL_id_22_Sensor_tr_50000_1st_126448.csv due to filter criteria.
Processing file: 072_SMD_id_16_Facility_tr_7119_1st_15849.csv, Data shape: (28479, 38)
Label shape: (28479,)
Train data shape: (7119, 38), Test data shape: (21360, 38), Label test shape: (21360,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 072_SMD_id_16_Facility_tr_7119_1st_15849.csv...
Using CUDA device: 2
Data shape: (21360, 38)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=38), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 21360, window_size 5000, adding 3640 samples


  0%|          | 0/5 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


 40%|████      | 2/5 [00:01<00:01,  1.92it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


 80%|████████  | 4/5 [00:02<00:00,  1.92it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])

100%|██████████| 5/5 [00:02<00:00,  1.92it/s]
Original shapes - Output: (25000,), Label: (21360,)
Aligned shapes - Output: (21360,), Label: (21360,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]

Calculating Affiliation Metrics:  38%|███▊      | 77/200 [00:01<00:01, 75.16it/s]

Calculating Affiliation Metrics:  76%|███████▋  | 153/200 [00:03<00:01, 44.42it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:03<00:00, 53.03it/s]
Table1-only metrics completed in 3.88s
Evaluation Result:  {'AUC-PR': 0.30836178627408356, 'AUC-ROC': 0.5257603854926094, 'VUS-PR': 0.30271390618235233, 'VUS-ROC': 0.5257553232194658, 'Standard-F1': 0.3528129088525462, 'Standard-Precision': 0.42758985200845645, 'Standard-Recall': 0.3002969561989606, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9265251484234891, 'Affiliation-P': 0.9104936990300562, 'Affiliation-R': 0.9431312612412511, 'F1_T': 0.35242866477843027, 'Precision_T': 0.4264628360569318, 'Recall_T': 0.3002969561989606}
Results for 072_SMD_id_16_Facility_tr_7119_1st_15849.csv: {'filename': '072_SMD_id_16_Facility_tr_7119_1st_15849.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '7119', 'data_shape': '28479x38', 'output_length': 25000, 'label_length': 21360, 'aligned_length': 21360, 'AUC-PR': 0.30836178627408356, 'AUC-ROC': 0.5257603854926094, 'VUS-PR': 0.30271390618235233, 'VUS-ROC': 0.5257553232194658, 'Standard-F1': 0.3528129088525462, 'Standard-Precision': 0.42758985200845645, 'Standard-Recall': 0.3002969561989606, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9265251484234891, 'Affiliation-P': 0.9104936990300562, 'Affiliation-R': 0.9431312612412511, 'F1_T': 0.35242866477843027, 'Precision_T': 0.4264628360569318, 'Recall_T': 0.3002969561989606}
Logits metrics skipped for 072_SMD_id_16_Facility_tr_7119_1st_15849.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/072_SMD_id_16_Facility_tr_7119_1st_15849_results.pkl
Processing file: 068_SMD_id_12_Facility_tr_1099_1st_1199.csv, Data shape: (28713, 38)
Label shape: (28713,)
Train data shape: (1099, 38), Test data shape: (27614, 38), Label test shape: (27614,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 068_SMD_id_12_Facility_tr_1099_1st_1199.csv...
Using CUDA device: 2
Data shape: (27614, 38)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=38), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 27614, window_size 5000, adding 2386 samples


  0%|          | 0/6 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


 33%|███▎      | 2/6 [00:01<00:02,  1.92it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


 67%|██████▋   | 4/6 [00:02<00:01,  1.92it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


100%|██████████| 6/6 [00:03<00:00,  1.93it/s]
100%|██████████| 6/6 [00:03<00:00,  1.93it/s]
Original shapes - Output: (30000,), Label: (27614,)
Aligned shapes - Output: (27614,), Label: (27614,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]

Calculating Affiliation Metrics:  32%|███▎      | 65/200 [00:01<00:02, 64.75it/s]

Calculating Affiliation Metrics:  65%|██████▌   | 130/200 [00:02<00:01, 49.23it/s]

Calculating Affiliation Metrics:  94%|█████████▎| 187/200 [00:03<00:00, 52.10it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:03<00:00, 55.14it/s]
Table1-only metrics completed in 3.75s
Evaluation Result:  {'AUC-PR': 0.7523728459468848, 'AUC-ROC': 0.9527723871944149, 'VUS-PR': 0.7064548375471142, 'VUS-ROC': 0.9526527531007283, 'Standard-F1': 0.724137931033981, 'Standard-Precision': 0.7581227436823077, 'Standard-Recall': 0.6930693069306907, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9529185574300726, 'Affiliation-P': 0.9134712554807582, 'Affiliation-R': 0.9959266076284334, 'F1_T': 0.724137931033981, 'Precision_T': 0.7581227436823077, 'Recall_T': 0.6930693069306907}
Results for 068_SMD_id_12_Facility_tr_1099_1st_1199.csv: {'filename': '068_SMD_id_12_Facility_tr_1099_1st_1199.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '1099', 'data_shape': '28713x38', 'output_length': 30000, 'label_length': 27614, 'aligned_length': 27614, 'AUC-PR': 0.7523728459468848, 'AUC-ROC': 0.9527723871944149, 'VUS-PR': 0.7064548375471142, 'VUS-ROC': 0.9526527531007283, 'Standard-F1': 0.724137931033981, 'Standard-Precision': 0.7581227436823077, 'Standard-Recall': 0.6930693069306907, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.9529185574300726, 'Affiliation-P': 0.9134712554807582, 'Affiliation-R': 0.9959266076284334, 'F1_T': 0.724137931033981, 'Precision_T': 0.7581227436823077, 'Recall_T': 0.6930693069306907}
Logits metrics skipped for 068_SMD_id_12_Facility_tr_1099_1st_1199.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/068_SMD_id_12_Facility_tr_1099_1st_1199_results.pkl
Processing file: 157_SMAP_id_14_Sensor_tr_2126_1st_5000.csv, Data shape: (8505, 25)
Label shape: (8505,)
Train data shape: (2126, 25), Test data shape: (6379, 25), Label test shape: (6379,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 157_SMAP_id_14_Sensor_tr_2126_1st_5000.csv...
Using CUDA device: 2
Data shape: (6379, 25)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=25), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 6379, window_size 5000, adding 3621 samples


  0%|          | 0/2 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])

100%|██████████| 2/2 [00:00<00:00,  4.46it/s]
Original shapes - Output: (10000,), Label: (6379,)
Aligned shapes - Output: (6379,), Label: (6379,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:00<00:00, 271.39it/s]
Table1-only metrics completed in 0.78s
Evaluation Result:  {'AUC-PR': 0.1085267850487855, 'AUC-ROC': 0.6831431627284072, 'VUS-PR': 0.09739784809561125, 'VUS-ROC': 0.6830424117878088, 'Standard-F1': 0.17858341725391919, 'Standard-Precision': 0.10011290929619869, 'Standard-Recall': 0.8260869565217365, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8660160132108228, 'Affiliation-P': 0.7853280536021755, 'Affiliation-R': 0.9651830428619048, 'F1_T': 0.17774807885045937, 'Precision_T': 0.09958816922500932, 'Recall_T': 0.8260869565217365}
Results for 157_SMAP_id_14_Sensor_tr_2126_1st_5000.csv: {'filename': '157_SMAP_id_14_Sensor_tr_2126_1st_5000.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '2126', 'data_shape': '8505x25', 'output_length': 10000, 'label_length': 6379, 'aligned_length': 6379, 'AUC-PR': 0.1085267850487855, 'AUC-ROC': 0.6831431627284072, 'VUS-PR': 0.09739784809561125, 'VUS-ROC': 0.6830424117878088, 'Standard-F1': 0.17858341725391919, 'Standard-Precision': 0.10011290929619869, 'Standard-Recall': 0.8260869565217365, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8660160132108228, 'Affiliation-P': 0.7853280536021755, 'Affiliation-R': 0.9651830428619048, 'F1_T': 0.17774807885045937, 'Precision_T': 0.09958816922500932, 'Recall_T': 0.8260869565217365}
Logits metrics skipped for 157_SMAP_id_14_Sensor_tr_2126_1st_5000.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/157_SMAP_id_14_Sensor_tr_2126_1st_5000_results.pkl
Skipping file: 138_CATSv2_id_1_Sensor_tr_16568_1st_16668.csv due to filter criteria.
Skipping file: 047_GHL_id_16_Sensor_tr_50000_1st_77001.csv due to filter criteria.
Skipping file: 081_LTDB_id_3_Medical_tr_500_1st_26.csv due to filter criteria.
Skipping file: 131_OPPORTUNITY_id_3_HumanActivity_tr_7016_1st_26691.csv due to filter criteria.
Processing file: 115_PSM_id_1_Facility_tr_50000_1st_129872.csv, Data shape: (217624, 25)
Label shape: (217624,)
Train data shape: (50000, 25), Test data shape: (167624, 25), Label test shape: (167624,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 115_PSM_id_1_Facility_tr_50000_1st_129872.csv...
Using CUDA device: 2
Data shape: (167624, 25)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=25), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 167624, window_size 5000, adding 2376 samples


  0%|          | 0/34 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])


 15%|█▍        | 5/34 [00:01<00:06,  4.48it/s]batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])


 29%|██▉       | 10/34 [00:02<00:05,  4.49it/s]batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])


 44%|████▍     | 15/34 [00:03<00:04,  4.50it/s]batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])


 59%|█████▉    | 20/34 [00:04<00:03,  4.51it/s]batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])


 74%|███████▎  | 25/34 [00:05<00:01,  4.51it/s]batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])


 88%|████████▊ | 30/34 [00:06<00:00,  4.52it/s]batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])

100%|██████████| 34/34 [00:07<00:00,  4.51it/s]
Original shapes - Output: (170000,), Label: (167624,)
Aligned shapes - Output: (167624,), Label: (167624,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]

Calculating Affiliation Metrics:   4%|▍         | 9/200 [00:01<00:21,  8.92it/s]

Calculating Affiliation Metrics:   9%|▉         | 18/200 [00:02<00:31,  5.72it/s]

Calculating Affiliation Metrics:  12%|█▎        | 25/200 [00:05<00:38,  4.50it/s]

Calculating Affiliation Metrics:  16%|█▌        | 31/200 [00:07<00:44,  3.76it/s]

Calculating Affiliation Metrics:  18%|█▊        | 36/200 [00:09<00:49,  3.29it/s]

Calculating Affiliation Metrics:  20%|██        | 40/200 [00:10<00:53,  2.97it/s]

Calculating Affiliation Metrics:  22%|██▏       | 44/200 [00:12<00:57,  2.70it/s]

Calculating Affiliation Metrics:  24%|██▎       | 47/200 [00:14<01:00,  2.52it/s]

Calculating Affiliation Metrics:  25%|██▌       | 50/200 [00:15<01:03,  2.37it/s]

Calculating Affiliation Metrics:  26%|██▋       | 53/200 [00:17<01:05,  2.24it/s]

Calculating Affiliation Metrics:  28%|██▊       | 56/200 [00:18<01:07,  2.13it/s]

Calculating Affiliation Metrics:  30%|██▉       | 59/200 [00:20<01:09,  2.03it/s]

Calculating Affiliation Metrics:  31%|███       | 62/200 [00:22<01:11,  1.92it/s]

Calculating Affiliation Metrics:  32%|███▏      | 64/200 [00:23<01:12,  1.88it/s]

Calculating Affiliation Metrics:  33%|███▎      | 66/200 [00:24<01:13,  1.84it/s]

Calculating Affiliation Metrics:  34%|███▍      | 68/200 [00:25<01:13,  1.79it/s]

Calculating Affiliation Metrics:  35%|███▌      | 70/200 [00:27<01:13,  1.76it/s]

Calculating Affiliation Metrics:  36%|███▌      | 72/200 [00:28<01:14,  1.73it/s]

Calculating Affiliation Metrics:  37%|███▋      | 74/200 [00:29<01:13,  1.70it/s]

Calculating Affiliation Metrics:  38%|███▊      | 76/200 [00:30<01:13,  1.68it/s]

Calculating Affiliation Metrics:  39%|███▉      | 78/200 [00:32<01:13,  1.66it/s]

Calculating Affiliation Metrics:  40%|████      | 80/200 [00:33<01:13,  1.64it/s]

Calculating Affiliation Metrics:  41%|████      | 82/200 [00:34<01:12,  1.63it/s]

Calculating Affiliation Metrics:  42%|████▏     | 84/200 [00:35<01:11,  1.61it/s]

Calculating Affiliation Metrics:  43%|████▎     | 86/200 [00:37<01:11,  1.60it/s]

Calculating Affiliation Metrics:  44%|████▍     | 88/200 [00:38<01:10,  1.59it/s]

Calculating Affiliation Metrics:  45%|████▌     | 90/200 [00:39<01:09,  1.59it/s]

Calculating Affiliation Metrics:  46%|████▌     | 92/200 [00:40<01:08,  1.58it/s]

Calculating Affiliation Metrics:  47%|████▋     | 94/200 [00:42<01:07,  1.58it/s]

Calculating Affiliation Metrics:  48%|████▊     | 96/200 [00:43<01:05,  1.58it/s]

Calculating Affiliation Metrics:  49%|████▉     | 98/200 [00:44<01:04,  1.58it/s]

Calculating Affiliation Metrics:  50%|█████     | 100/200 [00:45<01:03,  1.59it/s]

Calculating Affiliation Metrics:  51%|█████     | 102/200 [00:47<01:01,  1.59it/s]

Calculating Affiliation Metrics:  52%|█████▏    | 104/200 [00:48<01:00,  1.60it/s]

Calculating Affiliation Metrics:  53%|█████▎    | 106/200 [00:49<00:58,  1.60it/s]

Calculating Affiliation Metrics:  54%|█████▍    | 108/200 [00:50<00:56,  1.62it/s]

Calculating Affiliation Metrics:  55%|█████▌    | 110/200 [00:52<00:55,  1.63it/s]

Calculating Affiliation Metrics:  56%|█████▌    | 112/200 [00:53<00:53,  1.64it/s]

Calculating Affiliation Metrics:  57%|█████▋    | 114/200 [00:54<00:52,  1.65it/s]

Calculating Affiliation Metrics:  58%|█████▊    | 116/200 [00:55<00:50,  1.66it/s]

Calculating Affiliation Metrics:  59%|█████▉    | 118/200 [00:56<00:49,  1.67it/s]

Calculating Affiliation Metrics:  60%|██████    | 120/200 [00:57<00:47,  1.68it/s]

Calculating Affiliation Metrics:  61%|██████    | 122/200 [00:59<00:45,  1.70it/s]

Calculating Affiliation Metrics:  62%|██████▏   | 124/200 [01:00<00:44,  1.71it/s]

Calculating Affiliation Metrics:  63%|██████▎   | 126/200 [01:01<00:42,  1.73it/s]

Calculating Affiliation Metrics:  64%|██████▍   | 128/200 [01:02<00:41,  1.75it/s]

Calculating Affiliation Metrics:  65%|██████▌   | 130/200 [01:03<00:39,  1.77it/s]

Calculating Affiliation Metrics:  66%|██████▌   | 132/200 [01:04<00:37,  1.79it/s]

Calculating Affiliation Metrics:  67%|██████▋   | 134/200 [01:05<00:36,  1.82it/s]

Calculating Affiliation Metrics:  68%|██████▊   | 136/200 [01:06<00:34,  1.85it/s]

Calculating Affiliation Metrics:  69%|██████▉   | 138/200 [01:07<00:32,  1.88it/s]

Calculating Affiliation Metrics:  70%|███████   | 141/200 [01:09<00:30,  1.94it/s]

Calculating Affiliation Metrics:  72%|███████▏  | 144/200 [01:10<00:28,  2.00it/s]

Calculating Affiliation Metrics:  74%|███████▎  | 147/200 [01:12<00:25,  2.05it/s]

Calculating Affiliation Metrics:  75%|███████▌  | 150/200 [01:13<00:23,  2.11it/s]

Calculating Affiliation Metrics:  76%|███████▋  | 153/200 [01:14<00:21,  2.18it/s]

Calculating Affiliation Metrics:  78%|███████▊  | 156/200 [01:15<00:19,  2.27it/s]

Calculating Affiliation Metrics:  80%|███████▉  | 159/200 [01:17<00:17,  2.37it/s]

Calculating Affiliation Metrics:  81%|████████  | 162/200 [01:18<00:15,  2.47it/s]

Calculating Affiliation Metrics:  82%|████████▎ | 165/200 [01:19<00:13,  2.58it/s]

Calculating Affiliation Metrics:  84%|████████▍ | 169/200 [01:20<00:11,  2.73it/s]

Calculating Affiliation Metrics:  86%|████████▋ | 173/200 [01:21<00:09,  2.89it/s]

Calculating Affiliation Metrics:  88%|████████▊ | 177/200 [01:22<00:07,  3.09it/s]

Calculating Affiliation Metrics:  91%|█████████ | 182/200 [01:24<00:05,  3.45it/s]

Calculating Affiliation Metrics:  94%|█████████▎| 187/200 [01:25<00:03,  3.85it/s]

Calculating Affiliation Metrics:  98%|█████████▊| 195/200 [01:26<00:01,  4.93it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [01:26<00:00,  2.32it/s]
Table1-only metrics completed in 87.23s
Evaluation Result:  {'AUC-PR': 0.2319898418806022, 'AUC-ROC': 0.5974934764651723, 'VUS-PR': 0.21692047124466282, 'VUS-ROC': 0.5846719065692294, 'Standard-F1': 0.28254810501175803, 'Standard-Precision': 0.18431509759328218, 'Standard-Recall': 0.6049792871498297, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8216609949431958, 'Affiliation-P': 0.7282995394094962, 'Affiliation-R': 0.942478234279199, 'F1_T': 0.282490093607866, 'Precision_T': 0.18423911007902918, 'Recall_T': 0.6052663959640704}
Results for 115_PSM_id_1_Facility_tr_50000_1st_129872.csv: {'filename': '115_PSM_id_1_Facility_tr_50000_1st_129872.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '50000', 'data_shape': '217624x25', 'output_length': 170000, 'label_length': 167624, 'aligned_length': 167624, 'AUC-PR': 0.2319898418806022, 'AUC-ROC': 0.5974934764651723, 'VUS-PR': 0.21692047124466282, 'VUS-ROC': 0.5846719065692294, 'Standard-F1': 0.28254810501175803, 'Standard-Precision': 0.18431509759328218, 'Standard-Recall': 0.6049792871498297, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8216609949431958, 'Affiliation-P': 0.7282995394094962, 'Affiliation-R': 0.942478234279199, 'F1_T': 0.282490093607866, 'Precision_T': 0.18423911007902918, 'Recall_T': 0.6052663959640704}
Logits metrics skipped for 115_PSM_id_1_Facility_tr_50000_1st_129872.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/115_PSM_id_1_Facility_tr_50000_1st_129872_results.pkl
Skipping file: 191_Exathlon_id_18_Facility_tr_11665_1st_13484.csv due to filter criteria.
Processing file: 066_SMD_id_10_Facility_tr_2634_1st_2734.csv, Data shape: (23687, 38)
Label shape: (23687,)
Train data shape: (2634, 38), Test data shape: (21053, 38), Label test shape: (21053,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 066_SMD_id_10_Facility_tr_2634_1st_2734.csv...
Using CUDA device: 2
Data shape: (21053, 38)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=38), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 21053, window_size 5000, adding 3947 samples


  0%|          | 0/5 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


 40%|████      | 2/5 [00:01<00:01,  1.92it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


 80%|████████  | 4/5 [00:02<00:00,  1.92it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])

100%|██████████| 5/5 [00:02<00:00,  1.92it/s]
Original shapes - Output: (25000,), Label: (21053,)
Aligned shapes - Output: (21053,), Label: (21053,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]

Calculating Affiliation Metrics:  41%|████      | 82/200 [00:01<00:01, 81.93it/s]

Calculating Affiliation Metrics:  82%|████████▏ | 164/200 [00:02<00:00, 61.01it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:02<00:00, 68.03it/s]
Table1-only metrics completed in 3.05s
Evaluation Result:  {'AUC-PR': 0.2556073418149325, 'AUC-ROC': 0.6186537217937242, 'VUS-PR': 0.23689333213032504, 'VUS-ROC': 0.6185987649439805, 'Standard-F1': 0.28381374722797964, 'Standard-Precision': 0.510638297872339, 'Standard-Recall': 0.19651995905834166, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.898826357515197, 'Affiliation-P': 0.8319242375845991, 'Affiliation-R': 0.9774298218851801, 'F1_T': 0.2810035842289706, 'Precision_T': 0.4688995215310993, 'Recall_T': 0.2006141248720571}
Results for 066_SMD_id_10_Facility_tr_2634_1st_2734.csv: {'filename': '066_SMD_id_10_Facility_tr_2634_1st_2734.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '2634', 'data_shape': '23687x38', 'output_length': 25000, 'label_length': 21053, 'aligned_length': 21053, 'AUC-PR': 0.2556073418149325, 'AUC-ROC': 0.6186537217937242, 'VUS-PR': 0.23689333213032504, 'VUS-ROC': 0.6185987649439805, 'Standard-F1': 0.28381374722797964, 'Standard-Precision': 0.510638297872339, 'Standard-Recall': 0.19651995905834166, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.898826357515197, 'Affiliation-P': 0.8319242375845991, 'Affiliation-R': 0.9774298218851801, 'F1_T': 0.2810035842289706, 'Precision_T': 0.4688995215310993, 'Recall_T': 0.2006141248720571}
Logits metrics skipped for 066_SMD_id_10_Facility_tr_2634_1st_2734.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/066_SMD_id_10_Facility_tr_2634_1st_2734_results.pkl
Skipping file: 183_Exathlon_id_10_Facility_tr_11665_1st_13484.csv due to filter criteria.
Processing file: 007_MSL_id_6_Sensor_tr_980_1st_3550.csv, Data shape: (3922, 55)
Label shape: (3922,)
Train data shape: (980, 55), Test data shape: (2942, 55), Label test shape: (2942,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 007_MSL_id_6_Sensor_tr_980_1st_3550.csv...
Using CUDA device: 2
Data shape: (2942, 55)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=55), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda


  0%|          | 0/1 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 2942, 55])
batched_mask shape: torch.Size([1, 2942])

100%|██████████| 1/1 [00:00<00:00,  2.68it/s]
Original shapes - Output: (2942,), Label: (2942,)
Aligned shapes - Output: (2942,), Label: (2942,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:00<00:00, 338.43it/s]
Table1-only metrics completed in 0.62s
Evaluation Result:  {'AUC-PR': 0.28696361569959533, 'AUC-ROC': 0.821476790345505, 'VUS-PR': 0.29154105999166113, 'VUS-ROC': 0.8214483166089679, 'Standard-F1': 0.4984615384610378, 'Standard-Precision': 0.46551724137930767, 'Standard-Recall': 0.5364238410595991, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8545625951190494, 'Affiliation-P': 0.7482453561651387, 'Affiliation-R': 0.9960967220568968, 'F1_T': 0.49693251533692295, 'Precision_T': 0.46285714285714025, 'Recall_T': 0.5364238410595991}
Results for 007_MSL_id_6_Sensor_tr_980_1st_3550.csv: {'filename': '007_MSL_id_6_Sensor_tr_980_1st_3550.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '980', 'data_shape': '3922x55', 'output_length': 2942, 'label_length': 2942, 'aligned_length': 2942, 'AUC-PR': 0.28696361569959533, 'AUC-ROC': 0.821476790345505, 'VUS-PR': 0.29154105999166113, 'VUS-ROC': 0.8214483166089679, 'Standard-F1': 0.4984615384610378, 'Standard-Precision': 0.46551724137930767, 'Standard-Recall': 0.5364238410595991, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8545625951190494, 'Affiliation-P': 0.7482453561651387, 'Affiliation-R': 0.9960967220568968, 'F1_T': 0.49693251533692295, 'Precision_T': 0.46285714285714025, 'Recall_T': 0.5364238410595991}
Logits metrics skipped for 007_MSL_id_6_Sensor_tr_980_1st_3550.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/007_MSL_id_6_Sensor_tr_980_1st_3550_results.pkl
Skipping file: 028_MITDB_id_10_Medical_tr_37500_1st_39948.csv due to filter criteria.
Skipping file: 111_SVDB_id_28_Medical_tr_1791_1st_1891.csv due to filter criteria.
Processing file: 069_SMD_id_13_Facility_tr_5925_1st_12760.csv, Data shape: (23700, 38)
Label shape: (23700,)
Train data shape: (5925, 38), Test data shape: (17775, 38), Label test shape: (17775,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 069_SMD_id_13_Facility_tr_5925_1st_12760.csv...
Using CUDA device: 2
Data shape: (17775, 38)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=38), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 17775, window_size 5000, adding 2225 samples


  0%|          | 0/4 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


 50%|█████     | 2/4 [00:01<00:01,  1.92it/s]batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 38])
batched_mask shape: torch.Size([1, 5000])


100%|██████████| 4/4 [00:02<00:00,  1.91it/s]
100%|██████████| 4/4 [00:02<00:00,  1.91it/s]
Original shapes - Output: (20000,), Label: (17775,)
Aligned shapes - Output: (17775,), Label: (17775,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]

Calculating Affiliation Metrics:  34%|███▍      | 69/200 [00:01<00:01, 68.46it/s]

Calculating Affiliation Metrics:  69%|██████▉   | 138/200 [00:02<00:01, 61.22it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:02<00:00, 66.85it/s]
Table1-only metrics completed in 3.10s
Evaluation Result:  {'AUC-PR': 0.458248887351302, 'AUC-ROC': 0.8401748320355592, 'VUS-PR': 0.45677363189149106, 'VUS-ROC': 0.8399496021310962, 'Standard-F1': 0.6174459176547904, 'Standard-Precision': 0.51061865189289, 'Standard-Recall': 0.7807977409106951, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8839666400985446, 'Affiliation-P': 0.8223629884046411, 'Affiliation-R': 0.9555471917630292, 'F1_T': 0.6172189103003685, 'Precision_T': 0.5082191780821916, 'Recall_T': 0.7857394987645603}
Results for 069_SMD_id_13_Facility_tr_5925_1st_12760.csv: {'filename': '069_SMD_id_13_Facility_tr_5925_1st_12760.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '5925', 'data_shape': '23700x38', 'output_length': 20000, 'label_length': 17775, 'aligned_length': 17775, 'AUC-PR': 0.458248887351302, 'AUC-ROC': 0.8401748320355592, 'VUS-PR': 0.45677363189149106, 'VUS-ROC': 0.8399496021310962, 'Standard-F1': 0.6174459176547904, 'Standard-Precision': 0.51061865189289, 'Standard-Recall': 0.7807977409106951, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8839666400985446, 'Affiliation-P': 0.8223629884046411, 'Affiliation-R': 0.9555471917630292, 'F1_T': 0.6172189103003685, 'Precision_T': 0.5082191780821916, 'Recall_T': 0.7857394987645603}
Logits metrics skipped for 069_SMD_id_13_Facility_tr_5925_1st_12760.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/069_SMD_id_13_Facility_tr_5925_1st_12760_results.pkl
Skipping file: 092_SVDB_id_9_Medical_tr_2674_1st_2774.csv due to filter criteria.
Skipping file: 196_Exathlon_id_23_Facility_tr_11665_1st_13484.csv due to filter criteria.
Processing file: 146_SMAP_id_3_Sensor_tr_2128_1st_5000.csv, Data shape: (8514, 25)
Label shape: (8514,)
Train data shape: (2128, 25), Test data shape: (6386, 25), Label test shape: (6386,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 146_SMAP_id_3_Sensor_tr_2128_1st_5000.csv...
Using CUDA device: 2
Data shape: (6386, 25)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=25), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda
Padding AnomalyClipDataset: original length 6386, window_size 5000, adding 3614 samples


  0%|          | 0/2 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])
batched_time_series shape: torch.Size([1, 5000, 25])
batched_mask shape: torch.Size([1, 5000])

100%|██████████| 2/2 [00:00<00:00,  4.47it/s]
Original shapes - Output: (10000,), Label: (6386,)
Aligned shapes - Output: (6386,), Label: (6386,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:00<00:00, 296.04it/s]
Table1-only metrics completed in 0.72s
Evaluation Result:  {'AUC-PR': 0.08408724960451464, 'AUC-ROC': 0.7027417501231324, 'VUS-PR': 0.08343406169897132, 'VUS-ROC': 0.7027384109369479, 'Standard-F1': 0.19494320263813608, 'Standard-Precision': 0.10928512736236644, 'Standard-Recall': 0.9016949152542342, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8411058890366913, 'Affiliation-P': 0.7463178312379719, 'Affiliation-R': 0.9634745003804177, 'F1_T': 0.19426048565102044, 'Precision_T': 0.10895583986793228, 'Recall_T': 0.894915254237285}
Results for 146_SMAP_id_3_Sensor_tr_2128_1st_5000.csv: {'filename': '146_SMAP_id_3_Sensor_tr_2128_1st_5000.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '2128', 'data_shape': '8514x25', 'output_length': 10000, 'label_length': 6386, 'aligned_length': 6386, 'AUC-PR': 0.08408724960451464, 'AUC-ROC': 0.7027417501231324, 'VUS-PR': 0.08343406169897132, 'VUS-ROC': 0.7027384109369479, 'Standard-F1': 0.19494320263813608, 'Standard-Precision': 0.10928512736236644, 'Standard-Recall': 0.9016949152542342, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.8411058890366913, 'Affiliation-P': 0.7463178312379719, 'Affiliation-R': 0.9634745003804177, 'F1_T': 0.19426048565102044, 'Precision_T': 0.10895583986793228, 'Recall_T': 0.894915254237285}
Logits metrics skipped for 146_SMAP_id_3_Sensor_tr_2128_1st_5000.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/146_SMAP_id_3_Sensor_tr_2128_1st_5000_results.pkl
Skipping file: 024_MITDB_id_6_Medical_tr_50000_1st_58118.csv due to filter criteria.
Processing file: 015_MSL_id_14_Sensor_tr_575_1st_1250.csv, Data shape: (2303, 55)
Label shape: (2303,)
Train data shape: (575, 55), Test data shape: (1728, 55), Label test shape: (1728,)
Optimal Hyperparameters for Time_RCD: {'win_size': 5000, 'batch_size': 1}
Running Time_RCD on 015_MSL_id_14_Sensor_tr_575_1st_1250.csv...
Using CUDA device: 2
Data shape: (1728, 55)
Config: TimeRCDConfig(ts_config=TimeSeriesConfig(d_model=512, d_proj=256, patch_size=16, num_query_tokens=1, num_layers=8, num_heads=8, d_ff_dropout=0.1, use_rope=True, activation='gelu', num_features=55), llm_config=LLMConfig(model_name='deepseek-ai/DeepSeek-R1-Distill-Llama-8B', max_batch_size=20, num_fixed_tokens=30, num_prototype=1000, num_heads=8), deepspeed_config_path='experiments/configs/deepspeed_config.json', accelerate_config_path='experiments/configs/accelerate_config.yaml', batch_size=1, learning_rate=0.0001, num_epochs=1000, max_seq_len=512, dropout=0.1, accumulation_steps=1, weight_decay=1e-05, enable_ts_train=False, data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_dataset_20250614_133353', test_data_path='/home3/lantian/ChatTS/data/ChatTSQA/time_rcd_test_dataset_20250621_090751', pretrain_data_path='/home2/lijinbo/Projects/dataset/ChatTS/result', load_data=True, load_pretrain_path='experiments/checkpoints/pretrain/pretrain_checkpoint_latest.pth', load_path='experiments/checkpoints/checkpoint_step.pth', checkpoint_dir='experiments/checkpoints', log_freq=100, save_freq=10, save_step_freq=100, model_prefix='time_rcd_qa_by_pretrain', test_batch_limit=20, early_stopping_patience=7, seed=72, cuda_devices='2', dist_port='12355', device='cuda', use_multi_gpu=True, mixed_precision=True)
Checkpoint path: best_model/pretrain_checkpoint_best_multi.pth
Successfully loaded checkpoint from best_model/pretrain_checkpoint_best_multi.pth
Model loaded on device: cuda


  0%|          | 0/1 [00:00<?, ?it/s]batched_time_series shape: torch.Size([1, 1728, 55])
batched_mask shape: torch.Size([1, 1728])

100%|██████████| 1/1 [00:00<00:00,  7.58it/s]
Original shapes - Output: (1728,), Label: (1728,)
Aligned shapes - Output: (1728,), Label: (1728,)


Calculating Affiliation Metrics:   0%|          | 0/200 [00:00<?, ?it/s]
Calculating Affiliation Metrics: 100%|██████████| 200/200 [00:00<00:00, 467.85it/s]
Table1-only metrics completed in 0.45s
Evaluation Result:  {'AUC-PR': 0.26406088523693183, 'AUC-ROC': 0.6657323055360898, 'VUS-PR': 0.2620367646000384, 'VUS-ROC': 0.6656950552584587, 'Standard-F1': 0.37913741223628783, 'Standard-Precision': 0.27155172413793066, 'Standard-Recall': 0.6279069767441839, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.798759577484934, 'Affiliation-P': 0.6716517857142859, 'Affiliation-R': 0.9852065260858865, 'F1_T': 0.37837837837795657, 'Precision_T': 0.270773638968481, 'Recall_T': 0.6279069767441839}
Results for 015_MSL_id_14_Sensor_tr_575_1st_1250.csv: {'filename': '015_MSL_id_14_Sensor_tr_575_1st_1250.csv', 'AD_Name': 'Time_RCD', 'sliding_window': 0, 'train_index': '575', 'data_shape': '2303x55', 'output_length': 1728, 'label_length': 1728, 'aligned_length': 1728, 'AUC-PR': 0.26406088523693183, 'AUC-ROC': 0.6657323055360898, 'VUS-PR': 0.2620367646000384, 'VUS-ROC': 0.6656950552584587, 'Standard-F1': 0.37913741223628783, 'Standard-Precision': 0.27155172413793066, 'Standard-Recall': 0.6279069767441839, 'PA-F1': 0.0, 'PA-Precision': 0.0, 'PA-Recall': 0.0, 'Affiliation-F': 0.798759577484934, 'Affiliation-P': 0.6716517857142859, 'Affiliation-R': 0.9852065260858865, 'F1_T': 0.37837837837795657, 'Precision_T': 0.270773638968481, 'Recall_T': 0.6279069767441839}
Logits metrics skipped for 015_MSL_id_14_Sensor_tr_575_1st_1250.csv
Results saved to /timercd_full_eval_work/code/Time-RCD-main/Multi_Time_RCD/015_MSL_id_14_Sensor_tr_575_1st_1250_results.pkl

All results saved to Multi_Time_RCD.csv
Total file processed: 68
Results shape: (68, 24)