Using device: mps
Loading Oculus from checkpoints/oculus_detection_v2/final...
  ✓ Loaded projector from checkpoints/oculus_detection_v2/final
[Oculus] Loading language model...
Using a slow image processor as `use_fast` is unset and a slow processor was saved with this model. `use_fast=True` will be the default behavior in v4.52, even if the model was saved with a slow processor. This will result in minor differences in outputs. You'll still be able to use a slow processor with `use_fast=False`.

Loading weights:   0%|          | 0/472 [00:00<?, ?it/s]
Loading weights:   0%|          | 1/472 [00:00<00:00, 4593.98it/s, Materializing param=text_decoder.bert.embeddings.LayerNorm.bias]
Loading weights:   0%|          | 1/472 [00:00<00:00, 1314.42it/s, Materializing param=text_decoder.bert.embeddings.LayerNorm.bias]
Loading weights:   0%|          | 2/472 [00:00<00:00, 1144.89it/s, Materializing param=text_decoder.bert.embeddings.LayerNorm.weight]
Loading weights:   0%|          | 2/472 [00:00<00:00, 934.25it/s, Materializing param=text_decoder.bert.embeddings.LayerNorm.weight] 
Loading weights:   1%|          | 3/472 [00:00<00:00, 1160.68it/s, Materializing param=text_decoder.bert.embeddings.position_embeddings.weight]
Loading weights:   1%|          | 3/472 [00:00<00:00, 997.14it/s, Materializing param=text_decoder.bert.embeddings.position_embeddings.weight] 
Loading weights:   1%|          | 4/472 [00:00<00:00, 1134.52it/s, Materializing param=text_decoder.bert.embeddings.word_embeddings.weight]   
Loading weights:   1%|          | 4/472 [00:00<00:00, 1067.05it/s, Materializing param=text_decoder.bert.embeddings.word_embeddings.weight]
Loading weights:   1%|          | 5/472 [00:00<00:00, 1087.00it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.output.LayerNorm.bias]
Loading weights:   1%|          | 5/472 [00:00<00:00, 1060.45it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.output.LayerNorm.bias]
Loading weights:   1%|▏         | 6/472 [00:00<00:00, 995.25it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.output.LayerNorm.weight]
Loading weights:   1%|▏         | 6/472 [00:00<00:00, 933.00it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.output.LayerNorm.weight]
Loading weights:   1%|▏         | 7/472 [00:00<00:00, 898.03it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.output.dense.bias]      
Loading weights:   1%|▏         | 7/472 [00:00<00:00, 832.58it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.output.dense.bias]
Loading weights:   2%|▏         | 8/472 [00:00<00:00, 814.25it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.output.dense.weight]
Loading weights:   2%|▏         | 8/472 [00:00<00:00, 785.80it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.output.dense.weight]
Loading weights:   2%|▏         | 9/472 [00:00<00:00, 778.36it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.self.key.bias]      
Loading weights:   2%|▏         | 9/472 [00:00<00:00, 759.56it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.self.key.bias]
Loading weights:   2%|▏         | 10/472 [00:00<00:00, 801.23it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.self.key.weight]
Loading weights:   2%|▏         | 10/472 [00:00<00:00, 755.46it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.self.key.weight]
Loading weights:   2%|▏         | 11/472 [00:00<00:00, 794.29it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.self.query.bias]
Loading weights:   2%|▏         | 11/472 [00:00<00:00, 788.77it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.self.query.bias]
Loading weights:   3%|▎         | 12/472 [00:00<00:00, 836.89it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.self.query.weight]
Loading weights:   3%|▎         | 12/472 [00:00<00:00, 817.90it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.self.query.weight]
Loading weights:   3%|▎         | 13/472 [00:00<00:00, 866.78it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.self.value.bias]  
Loading weights:   3%|▎         | 13/472 [00:00<00:00, 849.90it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.self.value.bias]
Loading weights:   3%|▎         | 14/472 [00:00<00:00, 909.87it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.self.value.weight]
Loading weights:   3%|▎         | 14/472 [00:00<00:00, 886.20it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.self.value.weight]
Loading weights:   3%|▎         | 15/472 [00:00<00:00, 924.34it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.output.LayerNorm.bias]
Loading weights:   3%|▎         | 15/472 [00:00<00:00, 921.66it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.output.LayerNorm.bias]
Loading weights:   3%|▎         | 16/472 [00:00<00:00, 977.78it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.output.LayerNorm.weight]
Loading weights:   3%|▎         | 16/472 [00:00<00:00, 973.90it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.output.LayerNorm.weight]
Loading weights:   4%|▎         | 17/472 [00:00<00:00, 1004.33it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.output.dense.bias]     
Loading weights:   4%|▎         | 17/472 [00:00<00:00, 990.17it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.output.dense.bias] 
Loading weights:   4%|▍         | 18/472 [00:00<00:00, 1001.51it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.output.dense.weight]
Loading weights:   4%|▍         | 18/472 [00:00<00:00, 998.62it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.output.dense.weight] 
Loading weights:   4%|▍         | 19/472 [00:00<00:00, 1044.55it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.self.key.bias]     
Loading weights:   4%|▍         | 19/472 [00:00<00:00, 1018.39it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.self.key.bias]
Loading weights:   4%|▍         | 20/472 [00:00<00:00, 1034.74it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.self.key.weight]
Loading weights:   4%|▍         | 20/472 [00:00<00:00, 1016.53it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.self.key.weight]
Loading weights:   4%|▍         | 21/472 [00:00<00:00, 1034.54it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.self.query.bias]
Loading weights:   4%|▍         | 21/472 [00:00<00:00, 978.16it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.self.query.bias] 
Loading weights:   5%|▍         | 22/472 [00:00<00:00, 1018.06it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.self.query.weight]
Loading weights:   5%|▍         | 22/472 [00:00<00:00, 1001.02it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.self.query.weight]
Loading weights:   5%|▍         | 23/472 [00:00<00:00, 1027.25it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.self.value.bias]  
Loading weights:   5%|▍         | 23/472 [00:00<00:00, 1002.19it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.self.value.bias]
Loading weights:   5%|▌         | 24/472 [00:00<00:00, 1030.19it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.self.value.weight]
Loading weights:   5%|▌         | 24/472 [00:00<00:00, 1026.13it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.self.value.weight]
Loading weights:   5%|▌         | 25/472 [00:00<00:00, 1031.79it/s, Materializing param=text_decoder.bert.encoder.layer.0.intermediate.dense.bias]         
Loading weights:   5%|▌         | 25/472 [00:00<00:00, 1004.71it/s, Materializing param=text_decoder.bert.encoder.layer.0.intermediate.dense.bias]
Loading weights:   6%|▌         | 26/472 [00:00<00:00, 1008.39it/s, Materializing param=text_decoder.bert.encoder.layer.0.intermediate.dense.weight]
Loading weights:   6%|▌         | 26/472 [00:00<00:00, 987.18it/s, Materializing param=text_decoder.bert.encoder.layer.0.intermediate.dense.weight] 
Loading weights:   6%|▌         | 27/472 [00:00<00:00, 1003.06it/s, Materializing param=text_decoder.bert.encoder.layer.0.output.LayerNorm.bias]   
Loading weights:   6%|▌         | 27/472 [00:00<00:00, 991.52it/s, Materializing param=text_decoder.bert.encoder.layer.0.output.LayerNorm.bias] 
Loading weights:   6%|▌         | 28/472 [00:00<00:00, 1007.56it/s, Materializing param=text_decoder.bert.encoder.layer.0.output.LayerNorm.weight]
Loading weights:   6%|▌         | 28/472 [00:00<00:00, 983.40it/s, Materializing param=text_decoder.bert.encoder.layer.0.output.LayerNorm.weight] 
Loading weights:   6%|▌         | 29/472 [00:00<00:00, 1004.47it/s, Materializing param=text_decoder.bert.encoder.layer.0.output.dense.bias]     
Loading weights:   6%|▌         | 29/472 [00:00<00:00, 1002.84it/s, Materializing param=text_decoder.bert.encoder.layer.0.output.dense.bias]
Loading weights:   6%|▋         | 30/472 [00:00<00:00, 1026.10it/s, Materializing param=text_decoder.bert.encoder.layer.0.output.dense.weight]
Loading weights:   6%|▋         | 30/472 [00:00<00:00, 1006.34it/s, Materializing param=text_decoder.bert.encoder.layer.0.output.dense.weight]
Loading weights:   7%|▋         | 31/472 [00:00<00:00, 1018.54it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.output.LayerNorm.bias]
Loading weights:   7%|▋         | 31/472 [00:00<00:00, 1017.00it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.output.LayerNorm.bias]
Loading weights:   7%|▋         | 32/472 [00:00<00:00, 1036.18it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.output.LayerNorm.weight]
Loading weights:   7%|▋         | 32/472 [00:00<00:00, 1033.76it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.output.LayerNorm.weight]
Loading weights:   7%|▋         | 33/472 [00:00<00:00, 1051.29it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.output.dense.bias]      
Loading weights:   7%|▋         | 33/472 [00:00<00:00, 1049.78it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.output.dense.bias]
Loading weights:   7%|▋         | 34/472 [00:00<00:00, 1050.36it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.output.dense.weight]
Loading weights:   7%|▋         | 34/472 [00:00<00:00, 1049.03it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.output.dense.weight]
Loading weights:   7%|▋         | 35/472 [00:00<00:00, 1069.00it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.self.key.bias]      
Loading weights:   7%|▋         | 35/472 [00:00<00:00, 1067.73it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.self.key.bias]
Loading weights:   8%|▊         | 36/472 [00:00<00:00, 1092.54it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.self.key.weight]
Loading weights:   8%|▊         | 36/472 [00:00<00:00, 1077.72it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.self.key.weight]
Loading weights:   8%|▊         | 37/472 [00:00<00:00, 1084.41it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.self.query.bias]
Loading weights:   8%|▊         | 37/472 [00:00<00:00, 1076.34it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.self.query.bias]
Loading weights:   8%|▊         | 38/472 [00:00<00:00, 1088.92it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.self.query.weight]
Loading weights:   8%|▊         | 38/472 [00:00<00:00, 1087.40it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.self.query.weight]
Loading weights:   8%|▊         | 39/472 [00:00<00:00, 1102.76it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.self.value.bias]  
Loading weights:   8%|▊         | 39/472 [00:00<00:00, 1101.21it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.self.value.bias]
Loading weights:   8%|▊         | 40/472 [00:00<00:00, 1089.56it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.self.value.weight]
Loading weights:   8%|▊         | 40/472 [00:00<00:00, 1088.26it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.self.value.weight]
Loading weights:   9%|▊         | 41/472 [00:00<00:00, 1089.65it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.output.LayerNorm.bias]
Loading weights:   9%|▊         | 41/472 [00:00<00:00, 1070.72it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.output.LayerNorm.bias]
Loading weights:   9%|▉         | 42/472 [00:00<00:00, 1092.36it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.output.LayerNorm.weight]
Loading weights:   9%|▉         | 42/472 [00:00<00:00, 1081.01it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.output.LayerNorm.weight]
Loading weights:   9%|▉         | 43/472 [00:00<00:00, 1098.63it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.output.dense.bias]      
Loading weights:   9%|▉         | 43/472 [00:00<00:00, 1097.48it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.output.dense.bias]
Loading weights:   9%|▉         | 44/472 [00:00<00:00, 1111.00it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.output.dense.weight]
Loading weights:   9%|▉         | 44/472 [00:00<00:00, 1097.84it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.output.dense.weight]
Loading weights:  10%|▉         | 45/472 [00:00<00:00, 1109.06it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.self.key.bias]      
Loading weights:  10%|▉         | 45/472 [00:00<00:00, 1107.21it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.self.key.bias]
Loading weights:  10%|▉         | 46/472 [00:00<00:00, 1104.16it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.self.key.weight]
Loading weights:  10%|▉         | 46/472 [00:00<00:00, 1101.99it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.self.key.weight]
Loading weights:  10%|▉         | 47/472 [00:00<00:00, 1113.23it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.self.query.bias]
Loading weights:  10%|▉         | 47/472 [00:00<00:00, 1111.90it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.self.query.bias]
Loading weights:  10%|█         | 48/472 [00:00<00:00, 1043.98it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.self.query.weight]
Loading weights:  10%|█         | 48/472 [00:00<00:00, 1041.56it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.self.query.weight]
Loading weights:  10%|█         | 49/472 [00:00<00:00, 1048.85it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.self.value.bias]  
Loading weights:  10%|█         | 49/472 [00:00<00:00, 1035.29it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.self.value.bias]
Loading weights:  11%|█         | 50/472 [00:00<00:00, 1046.62it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.self.value.weight]
Loading weights:  11%|█         | 50/472 [00:00<00:00, 1043.82it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.self.value.weight]
Loading weights:  11%|█         | 51/472 [00:00<00:00, 1056.04it/s, Materializing param=text_decoder.bert.encoder.layer.1.intermediate.dense.bias]         
Loading weights:  11%|█         | 51/472 [00:00<00:00, 1041.61it/s, Materializing param=text_decoder.bert.encoder.layer.1.intermediate.dense.bias]
Loading weights:  11%|█         | 52/472 [00:00<00:00, 1059.32it/s, Materializing param=text_decoder.bert.encoder.layer.1.intermediate.dense.weight]
Loading weights:  11%|█         | 52/472 [00:00<00:00, 1058.31it/s, Materializing param=text_decoder.bert.encoder.layer.1.intermediate.dense.weight]
Loading weights:  11%|█         | 53/472 [00:00<00:00, 1069.81it/s, Materializing param=text_decoder.bert.encoder.layer.1.output.LayerNorm.bias]    
Loading weights:  11%|█         | 53/472 [00:00<00:00, 1067.52it/s, Materializing param=text_decoder.bert.encoder.layer.1.output.LayerNorm.bias]
Loading weights:  11%|█▏        | 54/472 [00:00<00:00, 1072.15it/s, Materializing param=text_decoder.bert.encoder.layer.1.output.LayerNorm.weight]
Loading weights:  11%|█▏        | 54/472 [00:00<00:00, 1070.56it/s, Materializing param=text_decoder.bert.encoder.layer.1.output.LayerNorm.weight]
Loading weights:  12%|█▏        | 55/472 [00:00<00:00, 1075.69it/s, Materializing param=text_decoder.bert.encoder.layer.1.output.dense.bias]      
Loading weights:  12%|█▏        | 55/472 [00:00<00:00, 1074.19it/s, Materializing param=text_decoder.bert.encoder.layer.1.output.dense.bias]
Loading weights:  12%|█▏        | 56/472 [00:00<00:00, 1089.67it/s, Materializing param=text_decoder.bert.encoder.layer.1.output.dense.weight]
Loading weights:  12%|█▏        | 56/472 [00:00<00:00, 1083.47it/s, Materializing param=text_decoder.bert.encoder.layer.1.output.dense.weight]
Loading weights:  12%|█▏        | 57/472 [00:00<00:00, 1093.28it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.output.LayerNorm.bias]
Loading weights:  12%|█▏        | 57/472 [00:00<00:00, 1091.77it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.output.LayerNorm.bias]
Loading weights:  12%|█▏        | 58/472 [00:00<00:00, 1096.39it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.output.LayerNorm.weight]
Loading weights:  12%|█▏        | 58/472 [00:00<00:00, 1094.38it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.output.LayerNorm.weight]
Loading weights:  12%|█▎        | 59/472 [00:00<00:00, 1104.54it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.output.dense.bias]      
Loading weights:  12%|█▎        | 59/472 [00:00<00:00, 1084.12it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.output.dense.bias]
Loading weights:  13%|█▎        | 60/472 [00:00<00:00, 1084.03it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.output.dense.weight]
Loading weights:  13%|█▎        | 60/472 [00:00<00:00, 1076.14it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.output.dense.weight]
Loading weights:  13%|█▎        | 61/472 [00:00<00:00, 1067.52it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.self.key.bias]      
Loading weights:  13%|█▎        | 61/472 [00:00<00:00, 1066.15it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.self.key.bias]
Loading weights:  13%|█▎        | 62/472 [00:00<00:00, 1072.76it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.self.key.weight]
Loading weights:  13%|█▎        | 62/472 [00:00<00:00, 1061.52it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.self.key.weight]
Loading weights:  13%|█▎        | 63/472 [00:00<00:00, 1065.99it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.self.query.bias]
Loading weights:  13%|█▎        | 63/472 [00:00<00:00, 1065.20it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.self.query.bias]
Loading weights:  14%|█▎        | 64/472 [00:00<00:00, 1075.85it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.self.query.weight]
Loading weights:  14%|█▎        | 64/472 [00:00<00:00, 1074.59it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.self.query.weight]
Loading weights:  14%|█▍        | 65/472 [00:00<00:00, 1083.55it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.self.value.bias]  
Loading weights:  14%|█▍        | 65/472 [00:00<00:00, 1076.05it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.self.value.bias]
Loading weights:  14%|█▍        | 66/472 [00:00<00:00, 1080.38it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.self.value.weight]
Loading weights:  14%|█▍        | 66/472 [00:00<00:00, 1072.23it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.self.value.weight]
Loading weights:  14%|█▍        | 67/472 [00:00<00:00, 1076.81it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.output.LayerNorm.bias]
Loading weights:  14%|█▍        | 67/472 [00:00<00:00, 1075.17it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.output.LayerNorm.bias]
Loading weights:  14%|█▍        | 68/472 [00:00<00:00, 1083.82it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.output.LayerNorm.weight]
Loading weights:  14%|█▍        | 68/472 [00:00<00:00, 1082.74it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.output.LayerNorm.weight]
Loading weights:  15%|█▍        | 69/472 [00:00<00:00, 1097.12it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.output.dense.bias]      
Loading weights:  15%|█▍        | 69/472 [00:00<00:00, 1096.48it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.output.dense.bias]
Loading weights:  15%|█▍        | 70/472 [00:00<00:00, 1111.20it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.output.dense.weight]
Loading weights:  15%|█▍        | 70/472 [00:00<00:00, 1110.58it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.output.dense.weight]
Loading weights:  15%|█▌        | 71/472 [00:00<00:00, 1125.31it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.self.key.bias]      
Loading weights:  15%|█▌        | 71/472 [00:00<00:00, 1124.74it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.self.key.bias]
Loading weights:  15%|█▌        | 72/472 [00:00<00:00, 1139.57it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.self.key.weight]
Loading weights:  15%|█▌        | 72/472 [00:00<00:00, 1138.97it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.self.key.weight]
Loading weights:  15%|█▌        | 73/472 [00:00<00:00, 1153.79it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.self.query.bias]
Loading weights:  15%|█▌        | 73/472 [00:00<00:00, 1153.21it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.self.query.bias]
Loading weights:  16%|█▌        | 74/472 [00:00<00:00, 1168.01it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.self.query.weight]
Loading weights:  16%|█▌        | 74/472 [00:00<00:00, 1167.44it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.self.query.weight]
Loading weights:  16%|█▌        | 75/472 [00:00<00:00, 1182.22it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.self.value.bias]  
Loading weights:  16%|█▌        | 75/472 [00:00<00:00, 1181.63it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.self.value.bias]
Loading weights:  16%|█▌        | 76/472 [00:00<00:00, 1196.38it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.self.value.weight]
Loading weights:  16%|█▌        | 76/472 [00:00<00:00, 1195.78it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.self.value.weight]
Loading weights:  16%|█▋        | 77/472 [00:00<00:00, 1210.53it/s, Materializing param=text_decoder.bert.encoder.layer.2.intermediate.dense.bias]         
Loading weights:  16%|█▋        | 77/472 [00:00<00:00, 1209.94it/s, Materializing param=text_decoder.bert.encoder.layer.2.intermediate.dense.bias]
Loading weights:  17%|█▋        | 78/472 [00:00<00:00, 1224.61it/s, Materializing param=text_decoder.bert.encoder.layer.2.intermediate.dense.weight]
Loading weights:  17%|█▋        | 78/472 [00:00<00:00, 1224.00it/s, Materializing param=text_decoder.bert.encoder.layer.2.intermediate.dense.weight]
Loading weights:  17%|█▋        | 79/472 [00:00<00:00, 1238.45it/s, Materializing param=text_decoder.bert.encoder.layer.2.output.LayerNorm.bias]    
Loading weights:  17%|█▋        | 79/472 [00:00<00:00, 1237.84it/s, Materializing param=text_decoder.bert.encoder.layer.2.output.LayerNorm.bias]
Loading weights:  17%|█▋        | 80/472 [00:00<00:00, 1252.51it/s, Materializing param=text_decoder.bert.encoder.layer.2.output.LayerNorm.weight]
Loading weights:  17%|█▋        | 80/472 [00:00<00:00, 1251.90it/s, Materializing param=text_decoder.bert.encoder.layer.2.output.LayerNorm.weight]
Loading weights:  17%|█▋        | 81/472 [00:00<00:00, 1266.42it/s, Materializing param=text_decoder.bert.encoder.layer.2.output.dense.bias]      
Loading weights:  17%|█▋        | 81/472 [00:00<00:00, 1265.81it/s, Materializing param=text_decoder.bert.encoder.layer.2.output.dense.bias]
Loading weights:  17%|█▋        | 82/472 [00:00<00:00, 1280.37it/s, Materializing param=text_decoder.bert.encoder.layer.2.output.dense.weight]
Loading weights:  17%|█▋        | 82/472 [00:00<00:00, 1279.75it/s, Materializing param=text_decoder.bert.encoder.layer.2.output.dense.weight]
Loading weights:  18%|█▊        | 83/472 [00:00<00:00, 1294.25it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.output.LayerNorm.bias]
Loading weights:  18%|█▊        | 83/472 [00:00<00:00, 1293.62it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.output.LayerNorm.bias]
Loading weights:  18%|█▊        | 84/472 [00:00<00:00, 1307.95it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.output.LayerNorm.weight]
Loading weights:  18%|█▊        | 84/472 [00:00<00:00, 1307.30it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.output.LayerNorm.weight]
Loading weights:  18%|█▊        | 85/472 [00:00<00:00, 1321.75it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.output.dense.bias]      
Loading weights:  18%|█▊        | 85/472 [00:00<00:00, 1321.13it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.output.dense.bias]
Loading weights:  18%|█▊        | 86/472 [00:00<00:00, 1335.37it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.output.dense.weight]
Loading weights:  18%|█▊        | 86/472 [00:00<00:00, 1334.72it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.output.dense.weight]
Loading weights:  18%|█▊        | 87/472 [00:00<00:00, 1349.15it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.self.key.bias]      
Loading weights:  18%|█▊        | 87/472 [00:00<00:00, 1348.51it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.self.key.bias]
Loading weights:  19%|█▊        | 88/472 [00:00<00:00, 1362.93it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.self.key.weight]
Loading weights:  19%|█▊        | 88/472 [00:00<00:00, 1362.29it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.self.key.weight]
Loading weights:  19%|█▉        | 89/472 [00:00<00:00, 1376.65it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.self.query.bias]
Loading weights:  19%|█▉        | 89/472 [00:00<00:00, 1376.01it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.self.query.bias]
Loading weights:  19%|█▉        | 90/472 [00:00<00:00, 1390.37it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.self.query.weight]
Loading weights:  19%|█▉        | 90/472 [00:00<00:00, 1389.73it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.self.query.weight]
Loading weights:  19%|█▉        | 91/472 [00:00<00:00, 1404.05it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.self.value.bias]  
Loading weights:  19%|█▉        | 91/472 [00:00<00:00, 1403.39it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.self.value.bias]
Loading weights:  19%|█▉        | 92/472 [00:00<00:00, 1417.46it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.self.value.weight]
Loading weights:  19%|█▉        | 92/472 [00:00<00:00, 1416.78it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.self.value.weight]
Loading weights:  20%|█▉        | 93/472 [00:00<00:00, 1431.08it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.output.LayerNorm.bias]
Loading weights:  20%|█▉        | 93/472 [00:00<00:00, 1430.40it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.output.LayerNorm.bias]
Loading weights:  20%|█▉        | 94/472 [00:00<00:00, 1444.52it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.output.LayerNorm.weight]
Loading weights:  20%|█▉        | 94/472 [00:00<00:00, 1443.78it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.output.LayerNorm.weight]
Loading weights:  20%|██        | 95/472 [00:00<00:00, 1457.87it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.output.dense.bias]      
Loading weights:  20%|██        | 95/472 [00:00<00:00, 1457.17it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.output.dense.bias]
Loading weights:  20%|██        | 96/472 [00:00<00:00, 1471.31it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.output.dense.weight]
Loading weights:  20%|██        | 96/472 [00:00<00:00, 1470.57it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.output.dense.weight]
Loading weights:  21%|██        | 97/472 [00:00<00:00, 1484.67it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.self.key.bias]      
Loading weights:  21%|██        | 97/472 [00:00<00:00, 1483.96it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.self.key.bias]
Loading weights:  21%|██        | 98/472 [00:00<00:00, 1497.97it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.self.key.weight]
Loading weights:  21%|██        | 98/472 [00:00<00:00, 1497.27it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.self.key.weight]
Loading weights:  21%|██        | 99/472 [00:00<00:00, 1511.11it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.self.query.bias]
Loading weights:  21%|██        | 99/472 [00:00<00:00, 1510.39it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.self.query.bias]
Loading weights:  21%|██        | 100/472 [00:00<00:00, 1524.12it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.self.query.weight]
Loading weights:  21%|██        | 100/472 [00:00<00:00, 1523.40it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.self.query.weight]
Loading weights:  21%|██▏       | 101/472 [00:00<00:00, 1537.13it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.self.value.bias]  
Loading weights:  21%|██▏       | 101/472 [00:00<00:00, 1536.40it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.self.value.bias]
Loading weights:  22%|██▏       | 102/472 [00:00<00:00, 1550.33it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.self.value.weight]
Loading weights:  22%|██▏       | 102/472 [00:00<00:00, 1549.56it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.self.value.weight]
Loading weights:  22%|██▏       | 103/472 [00:00<00:00, 1563.50it/s, Materializing param=text_decoder.bert.encoder.layer.3.intermediate.dense.bias]         
Loading weights:  22%|██▏       | 103/472 [00:00<00:00, 1562.76it/s, Materializing param=text_decoder.bert.encoder.layer.3.intermediate.dense.bias]
Loading weights:  22%|██▏       | 104/472 [00:00<00:00, 1576.69it/s, Materializing param=text_decoder.bert.encoder.layer.3.intermediate.dense.weight]
Loading weights:  22%|██▏       | 104/472 [00:00<00:00, 1575.98it/s, Materializing param=text_decoder.bert.encoder.layer.3.intermediate.dense.weight]
Loading weights:  22%|██▏       | 105/472 [00:00<00:00, 1589.78it/s, Materializing param=text_decoder.bert.encoder.layer.3.output.LayerNorm.bias]    
Loading weights:  22%|██▏       | 105/472 [00:00<00:00, 1589.06it/s, Materializing param=text_decoder.bert.encoder.layer.3.output.LayerNorm.bias]
Loading weights:  22%|██▏       | 106/472 [00:00<00:00, 1602.79it/s, Materializing param=text_decoder.bert.encoder.layer.3.output.LayerNorm.weight]
Loading weights:  22%|██▏       | 106/472 [00:00<00:00, 1602.03it/s, Materializing param=text_decoder.bert.encoder.layer.3.output.LayerNorm.weight]
Loading weights:  23%|██▎       | 107/472 [00:00<00:00, 1615.76it/s, Materializing param=text_decoder.bert.encoder.layer.3.output.dense.bias]      
Loading weights:  23%|██▎       | 107/472 [00:00<00:00, 1615.03it/s, Materializing param=text_decoder.bert.encoder.layer.3.output.dense.bias]
Loading weights:  23%|██▎       | 108/472 [00:00<00:00, 1628.84it/s, Materializing param=text_decoder.bert.encoder.layer.3.output.dense.weight]
Loading weights:  23%|██▎       | 108/472 [00:00<00:00, 1628.13it/s, Materializing param=text_decoder.bert.encoder.layer.3.output.dense.weight]
Loading weights:  23%|██▎       | 109/472 [00:00<00:00, 1641.89it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.output.LayerNorm.bias]
Loading weights:  23%|██▎       | 109/472 [00:00<00:00, 1641.13it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.output.LayerNorm.bias]
Loading weights:  23%|██▎       | 110/472 [00:00<00:00, 1654.78it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.output.LayerNorm.weight]
Loading weights:  23%|██▎       | 110/472 [00:00<00:00, 1654.01it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.output.LayerNorm.weight]
Loading weights:  24%|██▎       | 111/472 [00:00<00:00, 1667.15it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.output.dense.bias]      
Loading weights:  24%|██▎       | 111/472 [00:00<00:00, 1666.34it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.output.dense.bias]
Loading weights:  24%|██▎       | 112/472 [00:00<00:00, 1679.98it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.output.dense.weight]
Loading weights:  24%|██▎       | 112/472 [00:00<00:00, 1679.19it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.output.dense.weight]
Loading weights:  24%|██▍       | 113/472 [00:00<00:00, 1692.89it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.self.key.bias]      
Loading weights:  24%|██▍       | 113/472 [00:00<00:00, 1692.10it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.self.key.bias]
Loading weights:  24%|██▍       | 114/472 [00:00<00:00, 1705.77it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.self.key.weight]
Loading weights:  24%|██▍       | 114/472 [00:00<00:00, 1704.98it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.self.key.weight]
Loading weights:  24%|██▍       | 115/472 [00:00<00:00, 1718.45it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.self.query.bias]
Loading weights:  24%|██▍       | 115/472 [00:00<00:00, 1717.68it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.self.query.bias]
Loading weights:  25%|██▍       | 116/472 [00:00<00:00, 1731.19it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.self.query.weight]
Loading weights:  25%|██▍       | 116/472 [00:00<00:00, 1730.39it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.self.query.weight]
Loading weights:  25%|██▍       | 117/472 [00:00<00:00, 1743.99it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.self.value.bias]  
Loading weights:  25%|██▍       | 117/472 [00:00<00:00, 1743.15it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.self.value.bias]
Loading weights:  25%|██▌       | 118/472 [00:00<00:00, 1756.53it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.self.value.weight]
Loading weights:  25%|██▌       | 118/472 [00:00<00:00, 1755.72it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.self.value.weight]
Loading weights:  25%|██▌       | 119/472 [00:00<00:00, 1769.18it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.output.LayerNorm.bias]
Loading weights:  25%|██▌       | 119/472 [00:00<00:00, 1768.36it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.output.LayerNorm.bias]
Loading weights:  25%|██▌       | 120/472 [00:00<00:00, 1779.71it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.output.LayerNorm.weight]
Loading weights:  25%|██▌       | 120/472 [00:00<00:00, 1778.36it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.output.LayerNorm.weight]
Loading weights:  26%|██▌       | 121/472 [00:00<00:00, 1791.17it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.output.dense.bias]      
Loading weights:  26%|██▌       | 121/472 [00:00<00:00, 1790.19it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.output.dense.bias]
Loading weights:  26%|██▌       | 122/472 [00:00<00:00, 1803.30it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.output.dense.weight]
Loading weights:  26%|██▌       | 122/472 [00:00<00:00, 1802.36it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.output.dense.weight]
Loading weights:  26%|██▌       | 123/472 [00:00<00:00, 1815.56it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.self.key.bias]      
Loading weights:  26%|██▌       | 123/472 [00:00<00:00, 1814.62it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.self.key.bias]
Loading weights:  26%|██▋       | 124/472 [00:00<00:00, 1827.65it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.self.key.weight]
Loading weights:  26%|██▋       | 124/472 [00:00<00:00, 1826.70it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.self.key.weight]
Loading weights:  26%|██▋       | 125/472 [00:00<00:00, 1839.84it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.self.query.bias]
Loading weights:  26%|██▋       | 125/472 [00:00<00:00, 1838.97it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.self.query.bias]
Loading weights:  27%|██▋       | 126/472 [00:00<00:00, 1852.15it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.self.query.weight]
Loading weights:  27%|██▋       | 126/472 [00:00<00:00, 1851.28it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.self.query.weight]
Loading weights:  27%|██▋       | 127/472 [00:00<00:00, 1864.50it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.self.value.bias]  
Loading weights:  27%|██▋       | 127/472 [00:00<00:00, 1863.65it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.self.value.bias]
Loading weights:  27%|██▋       | 128/472 [00:00<00:00, 1876.84it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.self.value.weight]
Loading weights:  27%|██▋       | 128/472 [00:00<00:00, 1875.96it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.self.value.weight]
Loading weights:  27%|██▋       | 129/472 [00:00<00:00, 1889.17it/s, Materializing param=text_decoder.bert.encoder.layer.4.intermediate.dense.bias]         
Loading weights:  27%|██▋       | 129/472 [00:00<00:00, 1888.32it/s, Materializing param=text_decoder.bert.encoder.layer.4.intermediate.dense.bias]
Loading weights:  28%|██▊       | 130/472 [00:00<00:00, 1901.46it/s, Materializing param=text_decoder.bert.encoder.layer.4.intermediate.dense.weight]
Loading weights:  28%|██▊       | 130/472 [00:00<00:00, 1900.40it/s, Materializing param=text_decoder.bert.encoder.layer.4.intermediate.dense.weight]
Loading weights:  28%|██▊       | 131/472 [00:00<00:00, 1913.59it/s, Materializing param=text_decoder.bert.encoder.layer.4.output.LayerNorm.bias]    
Loading weights:  28%|██▊       | 131/472 [00:00<00:00, 1912.75it/s, Materializing param=text_decoder.bert.encoder.layer.4.output.LayerNorm.bias]
Loading weights:  28%|██▊       | 132/472 [00:00<00:00, 1925.92it/s, Materializing param=text_decoder.bert.encoder.layer.4.output.LayerNorm.weight]
Loading weights:  28%|██▊       | 132/472 [00:00<00:00, 1925.05it/s, Materializing param=text_decoder.bert.encoder.layer.4.output.LayerNorm.weight]
Loading weights:  28%|██▊       | 133/472 [00:00<00:00, 1938.22it/s, Materializing param=text_decoder.bert.encoder.layer.4.output.dense.bias]      
Loading weights:  28%|██▊       | 133/472 [00:00<00:00, 1937.29it/s, Materializing param=text_decoder.bert.encoder.layer.4.output.dense.bias]
Loading weights:  28%|██▊       | 134/472 [00:00<00:00, 1950.40it/s, Materializing param=text_decoder.bert.encoder.layer.4.output.dense.weight]
Loading weights:  28%|██▊       | 134/472 [00:00<00:00, 1949.49it/s, Materializing param=text_decoder.bert.encoder.layer.4.output.dense.weight]
Loading weights:  29%|██▊       | 135/472 [00:00<00:00, 1962.56it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.output.LayerNorm.bias]
Loading weights:  29%|██▊       | 135/472 [00:00<00:00, 1961.58it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.output.LayerNorm.bias]
Loading weights:  29%|██▉       | 136/472 [00:00<00:00, 1974.31it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.output.LayerNorm.weight]
Loading weights:  29%|██▉       | 136/472 [00:00<00:00, 1973.31it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.output.LayerNorm.weight]
Loading weights:  29%|██▉       | 137/472 [00:00<00:00, 1986.15it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.output.dense.bias]      
Loading weights:  29%|██▉       | 137/472 [00:00<00:00, 1985.25it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.output.dense.bias]
Loading weights:  29%|██▉       | 138/472 [00:00<00:00, 1998.24it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.output.dense.weight]
Loading weights:  29%|██▉       | 138/472 [00:00<00:00, 1997.29it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.output.dense.weight]
Loading weights:  29%|██▉       | 139/472 [00:00<00:00, 2010.21it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.self.key.bias]      
Loading weights:  29%|██▉       | 139/472 [00:00<00:00, 2009.29it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.self.key.bias]
Loading weights:  30%|██▉       | 140/472 [00:00<00:00, 2022.22it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.self.key.weight]
Loading weights:  30%|██▉       | 140/472 [00:00<00:00, 2021.20it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.self.key.weight]
Loading weights:  30%|██▉       | 141/472 [00:00<00:00, 2034.05it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.self.query.bias]
Loading weights:  30%|██▉       | 141/472 [00:00<00:00, 2033.14it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.self.query.bias]
Loading weights:  30%|███       | 142/472 [00:00<00:00, 2046.06it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.self.query.weight]
Loading weights:  30%|███       | 142/472 [00:00<00:00, 2045.12it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.self.query.weight]
Loading weights:  30%|███       | 143/472 [00:00<00:00, 2057.89it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.self.value.bias]  
Loading weights:  30%|███       | 143/472 [00:00<00:00, 2056.79it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.self.value.bias]
Loading weights:  31%|███       | 144/472 [00:00<00:00, 2069.62it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.self.value.weight]
Loading weights:  31%|███       | 144/472 [00:00<00:00, 2068.70it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.self.value.weight]
Loading weights:  31%|███       | 145/472 [00:00<00:00, 2081.51it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.output.LayerNorm.bias]
Loading weights:  31%|███       | 145/472 [00:00<00:00, 2080.53it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.output.LayerNorm.bias]
Loading weights:  31%|███       | 146/472 [00:00<00:00, 2093.14it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.output.LayerNorm.weight]
Loading weights:  31%|███       | 146/472 [00:00<00:00, 2092.18it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.output.LayerNorm.weight]
Loading weights:  31%|███       | 147/472 [00:00<00:00, 2104.88it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.output.dense.bias]      
Loading weights:  31%|███       | 147/472 [00:00<00:00, 2103.94it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.output.dense.bias]
Loading weights:  31%|███▏      | 148/472 [00:00<00:00, 2116.67it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.output.dense.weight]
Loading weights:  31%|███▏      | 148/472 [00:00<00:00, 2115.74it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.output.dense.weight]
Loading weights:  32%|███▏      | 149/472 [00:00<00:00, 2128.49it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.self.key.bias]      
Loading weights:  32%|███▏      | 149/472 [00:00<00:00, 2127.54it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.self.key.bias]
Loading weights:  32%|███▏      | 150/472 [00:00<00:00, 2140.11it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.self.key.weight]
Loading weights:  32%|███▏      | 150/472 [00:00<00:00, 2139.17it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.self.key.weight]
Loading weights:  32%|███▏      | 151/472 [00:00<00:00, 2151.86it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.self.query.bias]
Loading weights:  32%|███▏      | 151/472 [00:00<00:00, 2150.91it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.self.query.bias]
Loading weights:  32%|███▏      | 152/472 [00:00<00:00, 2163.55it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.self.query.weight]
Loading weights:  32%|███▏      | 152/472 [00:00<00:00, 2162.60it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.self.query.weight]
Loading weights:  32%|███▏      | 153/472 [00:00<00:00, 2175.19it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.self.value.bias]  
Loading weights:  32%|███▏      | 153/472 [00:00<00:00, 2174.23it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.self.value.bias]
Loading weights:  33%|███▎      | 154/472 [00:00<00:00, 2186.82it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.self.value.weight]
Loading weights:  33%|███▎      | 154/472 [00:00<00:00, 2185.83it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.self.value.weight]
Loading weights:  33%|███▎      | 155/472 [00:00<00:00, 2198.40it/s, Materializing param=text_decoder.bert.encoder.layer.5.intermediate.dense.bias]         
Loading weights:  33%|███▎      | 155/472 [00:00<00:00, 2197.43it/s, Materializing param=text_decoder.bert.encoder.layer.5.intermediate.dense.bias]
Loading weights:  33%|███▎      | 156/472 [00:00<00:00, 2209.86it/s, Materializing param=text_decoder.bert.encoder.layer.5.intermediate.dense.weight]
Loading weights:  33%|███▎      | 156/472 [00:00<00:00, 2208.88it/s, Materializing param=text_decoder.bert.encoder.layer.5.intermediate.dense.weight]
Loading weights:  33%|███▎      | 157/472 [00:00<00:00, 2221.48it/s, Materializing param=text_decoder.bert.encoder.layer.5.output.LayerNorm.bias]    
Loading weights:  33%|███▎      | 157/472 [00:00<00:00, 2220.53it/s, Materializing param=text_decoder.bert.encoder.layer.5.output.LayerNorm.bias]
Loading weights:  33%|███▎      | 158/472 [00:00<00:00, 2232.75it/s, Materializing param=text_decoder.bert.encoder.layer.5.output.LayerNorm.weight]
Loading weights:  33%|███▎      | 158/472 [00:00<00:00, 2231.77it/s, Materializing param=text_decoder.bert.encoder.layer.5.output.LayerNorm.weight]
Loading weights:  34%|███▎      | 159/472 [00:00<00:00, 2244.22it/s, Materializing param=text_decoder.bert.encoder.layer.5.output.dense.bias]      
Loading weights:  34%|███▎      | 159/472 [00:00<00:00, 2243.26it/s, Materializing param=text_decoder.bert.encoder.layer.5.output.dense.bias]
Loading weights:  34%|███▍      | 160/472 [00:00<00:00, 2255.81it/s, Materializing param=text_decoder.bert.encoder.layer.5.output.dense.weight]
Loading weights:  34%|███▍      | 160/472 [00:00<00:00, 2254.90it/s, Materializing param=text_decoder.bert.encoder.layer.5.output.dense.weight]
Loading weights:  34%|███▍      | 161/472 [00:00<00:00, 2267.27it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.output.LayerNorm.bias]
Loading weights:  34%|███▍      | 161/472 [00:00<00:00, 2266.31it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.output.LayerNorm.bias]
Loading weights:  34%|███▍      | 162/472 [00:00<00:00, 2278.52it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.output.LayerNorm.weight]
Loading weights:  34%|███▍      | 162/472 [00:00<00:00, 2277.50it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.output.LayerNorm.weight]
Loading weights:  35%|███▍      | 163/472 [00:00<00:00, 2289.71it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.output.dense.bias]      
Loading weights:  35%|███▍      | 163/472 [00:00<00:00, 2288.72it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.output.dense.bias]
Loading weights:  35%|███▍      | 164/472 [00:00<00:00, 2301.02it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.output.dense.weight]
Loading weights:  35%|███▍      | 164/472 [00:00<00:00, 2299.99it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.output.dense.weight]
Loading weights:  35%|███▍      | 165/472 [00:00<00:00, 2312.29it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.self.key.bias]      
Loading weights:  35%|███▍      | 165/472 [00:00<00:00, 2311.29it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.self.key.bias]
Loading weights:  35%|███▌      | 166/472 [00:00<00:00, 2323.57it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.self.key.weight]
Loading weights:  35%|███▌      | 166/472 [00:00<00:00, 2322.60it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.self.key.weight]
Loading weights:  35%|███▌      | 167/472 [00:00<00:00, 2334.95it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.self.query.bias]
Loading weights:  35%|███▌      | 167/472 [00:00<00:00, 2333.97it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.self.query.bias]
Loading weights:  36%|███▌      | 168/472 [00:00<00:00, 2346.31it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.self.query.weight]
Loading weights:  36%|███▌      | 168/472 [00:00<00:00, 2345.33it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.self.query.weight]
Loading weights:  36%|███▌      | 169/472 [00:00<00:00, 2357.51it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.self.value.bias]  
Loading weights:  36%|███▌      | 169/472 [00:00<00:00, 2356.49it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.self.value.bias]
Loading weights:  36%|███▌      | 170/472 [00:00<00:00, 2368.72it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.self.value.weight]
Loading weights:  36%|███▌      | 170/472 [00:00<00:00, 2367.74it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.self.value.weight]
Loading weights:  36%|███▌      | 171/472 [00:00<00:00, 2379.90it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.output.LayerNorm.bias]
Loading weights:  36%|███▌      | 171/472 [00:00<00:00, 2378.87it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.output.LayerNorm.bias]
Loading weights:  36%|███▋      | 172/472 [00:00<00:00, 2390.99it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.output.LayerNorm.weight]
Loading weights:  36%|███▋      | 172/472 [00:00<00:00, 2389.92it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.output.LayerNorm.weight]
Loading weights:  37%|███▋      | 173/472 [00:00<00:00, 2402.02it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.output.dense.bias]      
Loading weights:  37%|███▋      | 173/472 [00:00<00:00, 2400.98it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.output.dense.bias]
Loading weights:  37%|███▋      | 174/472 [00:00<00:00, 2413.09it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.output.dense.weight]
Loading weights:  37%|███▋      | 174/472 [00:00<00:00, 2412.05it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.output.dense.weight]
Loading weights:  37%|███▋      | 175/472 [00:00<00:00, 2424.09it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.self.key.bias]      
Loading weights:  37%|███▋      | 175/472 [00:00<00:00, 2422.99it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.self.key.bias]
Loading weights:  37%|███▋      | 176/472 [00:00<00:00, 2435.08it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.self.key.weight]
Loading weights:  37%|███▋      | 176/472 [00:00<00:00, 2433.97it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.self.key.weight]
Loading weights:  38%|███▊      | 177/472 [00:00<00:00, 2446.08it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.self.query.bias]
Loading weights:  38%|███▊      | 177/472 [00:00<00:00, 2444.99it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.self.query.bias]
Loading weights:  38%|███▊      | 178/472 [00:00<00:00, 2457.04it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.self.query.weight]
Loading weights:  38%|███▊      | 178/472 [00:00<00:00, 2455.99it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.self.query.weight]
Loading weights:  38%|███▊      | 179/472 [00:00<00:00, 2468.06it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.self.value.bias]  
Loading weights:  38%|███▊      | 179/472 [00:00<00:00, 2466.93it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.self.value.bias]
Loading weights:  38%|███▊      | 180/472 [00:00<00:00, 2478.83it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.self.value.weight]
Loading weights:  38%|███▊      | 180/472 [00:00<00:00, 2477.78it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.self.value.weight]
Loading weights:  38%|███▊      | 181/472 [00:00<00:00, 2489.72it/s, Materializing param=text_decoder.bert.encoder.layer.6.intermediate.dense.bias]         
Loading weights:  38%|███▊      | 181/472 [00:00<00:00, 2488.49it/s, Materializing param=text_decoder.bert.encoder.layer.6.intermediate.dense.bias]
Loading weights:  39%|███▊      | 182/472 [00:00<00:00, 2500.25it/s, Materializing param=text_decoder.bert.encoder.layer.6.intermediate.dense.weight]
Loading weights:  39%|███▊      | 182/472 [00:00<00:00, 2499.18it/s, Materializing param=text_decoder.bert.encoder.layer.6.intermediate.dense.weight]
Loading weights:  39%|███▉      | 183/472 [00:00<00:00, 2511.06it/s, Materializing param=text_decoder.bert.encoder.layer.6.output.LayerNorm.bias]    
Loading weights:  39%|███▉      | 183/472 [00:00<00:00, 2509.99it/s, Materializing param=text_decoder.bert.encoder.layer.6.output.LayerNorm.bias]
Loading weights:  39%|███▉      | 184/472 [00:00<00:00, 2521.97it/s, Materializing param=text_decoder.bert.encoder.layer.6.output.LayerNorm.weight]
Loading weights:  39%|███▉      | 184/472 [00:00<00:00, 2520.90it/s, Materializing param=text_decoder.bert.encoder.layer.6.output.LayerNorm.weight]
Loading weights:  39%|███▉      | 185/472 [00:00<00:00, 2532.76it/s, Materializing param=text_decoder.bert.encoder.layer.6.output.dense.bias]      
Loading weights:  39%|███▉      | 185/472 [00:00<00:00, 2531.68it/s, Materializing param=text_decoder.bert.encoder.layer.6.output.dense.bias]
Loading weights:  39%|███▉      | 186/472 [00:00<00:00, 2543.56it/s, Materializing param=text_decoder.bert.encoder.layer.6.output.dense.weight]
Loading weights:  39%|███▉      | 186/472 [00:00<00:00, 2542.52it/s, Materializing param=text_decoder.bert.encoder.layer.6.output.dense.weight]
Loading weights:  40%|███▉      | 187/472 [00:00<00:00, 2554.44it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.output.LayerNorm.bias]
Loading weights:  40%|███▉      | 187/472 [00:00<00:00, 2553.36it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.output.LayerNorm.bias]
Loading weights:  40%|███▉      | 188/472 [00:00<00:00, 2565.09it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.output.LayerNorm.weight]
Loading weights:  40%|███▉      | 188/472 [00:00<00:00, 2564.00it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.output.LayerNorm.weight]
Loading weights:  40%|████      | 189/472 [00:00<00:00, 2575.85it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.output.dense.bias]      
Loading weights:  40%|████      | 189/472 [00:00<00:00, 2574.76it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.output.dense.bias]
Loading weights:  40%|████      | 190/472 [00:00<00:00, 2586.48it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.output.dense.weight]
Loading weights:  40%|████      | 190/472 [00:00<00:00, 2585.32it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.output.dense.weight]
Loading weights:  40%|████      | 191/472 [00:00<00:00, 2596.98it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.self.key.bias]      
Loading weights:  40%|████      | 191/472 [00:00<00:00, 2595.83it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.self.key.bias]
Loading weights:  41%|████      | 192/472 [00:00<00:00, 2607.43it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.self.key.weight]
Loading weights:  41%|████      | 192/472 [00:00<00:00, 2606.37it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.self.key.weight]
Loading weights:  41%|████      | 193/472 [00:00<00:00, 2617.77it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.self.query.bias]
Loading weights:  41%|████      | 193/472 [00:00<00:00, 2616.64it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.self.query.bias]
Loading weights:  41%|████      | 194/472 [00:00<00:00, 2628.38it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.self.query.weight]
Loading weights:  41%|████      | 194/472 [00:00<00:00, 2626.88it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.self.query.weight]
Loading weights:  41%|████▏     | 195/472 [00:00<00:00, 2638.60it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.self.value.bias]  
Loading weights:  41%|████▏     | 195/472 [00:00<00:00, 2637.50it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.self.value.bias]
Loading weights:  42%|████▏     | 196/472 [00:00<00:00, 2649.12it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.self.value.weight]
Loading weights:  42%|████▏     | 196/472 [00:00<00:00, 2648.01it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.self.value.weight]
Loading weights:  42%|████▏     | 197/472 [00:00<00:00, 2659.59it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.output.LayerNorm.bias]
Loading weights:  42%|████▏     | 197/472 [00:00<00:00, 2658.47it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.output.LayerNorm.bias]
Loading weights:  42%|████▏     | 198/472 [00:00<00:00, 2669.84it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.output.LayerNorm.weight]
Loading weights:  42%|████▏     | 198/472 [00:00<00:00, 2668.65it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.output.LayerNorm.weight]
Loading weights:  42%|████▏     | 199/472 [00:00<00:00, 2680.18it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.output.dense.bias]      
Loading weights:  42%|████▏     | 199/472 [00:00<00:00, 2679.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.output.dense.bias]
Loading weights:  42%|████▏     | 200/472 [00:00<00:00, 2690.31it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.output.dense.weight]
Loading weights:  42%|████▏     | 200/472 [00:00<00:00, 2689.16it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.output.dense.weight]
Loading weights:  43%|████▎     | 201/472 [00:00<00:00, 2700.71it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.self.key.bias]      
Loading weights:  43%|████▎     | 201/472 [00:00<00:00, 2699.59it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.self.key.bias]
Loading weights:  43%|████▎     | 202/472 [00:00<00:00, 2711.02it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.self.key.weight]
Loading weights:  43%|████▎     | 202/472 [00:00<00:00, 2709.82it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.self.key.weight]
Loading weights:  43%|████▎     | 203/472 [00:00<00:00, 2721.26it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.self.query.bias]
Loading weights:  43%|████▎     | 203/472 [00:00<00:00, 2720.06it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.self.query.bias]
Loading weights:  43%|████▎     | 204/472 [00:00<00:00, 2731.56it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.self.query.weight]
Loading weights:  43%|████▎     | 204/472 [00:00<00:00, 2730.31it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.self.query.weight]
Loading weights:  43%|████▎     | 205/472 [00:00<00:00, 2741.67it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.self.value.bias]  
Loading weights:  43%|████▎     | 205/472 [00:00<00:00, 2740.57it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.self.value.bias]
Loading weights:  44%|████▎     | 206/472 [00:00<00:00, 2751.82it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.self.value.weight]
Loading weights:  44%|████▎     | 206/472 [00:00<00:00, 2750.16it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.self.value.weight]
Loading weights:  44%|████▍     | 207/472 [00:00<00:00, 2760.56it/s, Materializing param=text_decoder.bert.encoder.layer.7.intermediate.dense.bias]         
Loading weights:  44%|████▍     | 207/472 [00:00<00:00, 2759.38it/s, Materializing param=text_decoder.bert.encoder.layer.7.intermediate.dense.bias]
Loading weights:  44%|████▍     | 208/472 [00:00<00:00, 2770.72it/s, Materializing param=text_decoder.bert.encoder.layer.7.intermediate.dense.weight]
Loading weights:  44%|████▍     | 208/472 [00:00<00:00, 2769.54it/s, Materializing param=text_decoder.bert.encoder.layer.7.intermediate.dense.weight]
Loading weights:  44%|████▍     | 209/472 [00:00<00:00, 2781.00it/s, Materializing param=text_decoder.bert.encoder.layer.7.output.LayerNorm.bias]    
Loading weights:  44%|████▍     | 209/472 [00:00<00:00, 2779.82it/s, Materializing param=text_decoder.bert.encoder.layer.7.output.LayerNorm.bias]
Loading weights:  44%|████▍     | 210/472 [00:00<00:00, 2790.96it/s, Materializing param=text_decoder.bert.encoder.layer.7.output.LayerNorm.weight]
Loading weights:  44%|████▍     | 210/472 [00:00<00:00, 2789.81it/s, Materializing param=text_decoder.bert.encoder.layer.7.output.LayerNorm.weight]
Loading weights:  45%|████▍     | 211/472 [00:00<00:00, 2801.17it/s, Materializing param=text_decoder.bert.encoder.layer.7.output.dense.bias]      
Loading weights:  45%|████▍     | 211/472 [00:00<00:00, 2800.05it/s, Materializing param=text_decoder.bert.encoder.layer.7.output.dense.bias]
Loading weights:  45%|████▍     | 212/472 [00:00<00:00, 2811.50it/s, Materializing param=text_decoder.bert.encoder.layer.7.output.dense.weight]
Loading weights:  45%|████▍     | 212/472 [00:00<00:00, 2810.37it/s, Materializing param=text_decoder.bert.encoder.layer.7.output.dense.weight]
Loading weights:  45%|████▌     | 213/472 [00:00<00:00, 2821.65it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.output.LayerNorm.bias]
Loading weights:  45%|████▌     | 213/472 [00:00<00:00, 2817.95it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.output.LayerNorm.bias]
Loading weights:  45%|████▌     | 214/472 [00:00<00:00, 2827.26it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.output.LayerNorm.weight]
Loading weights:  45%|████▌     | 214/472 [00:00<00:00, 2825.72it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.output.LayerNorm.weight]
Loading weights:  46%|████▌     | 215/472 [00:00<00:00, 2836.27it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.output.dense.bias]      
Loading weights:  46%|████▌     | 215/472 [00:00<00:00, 2834.89it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.output.dense.bias]
Loading weights:  46%|████▌     | 216/472 [00:00<00:00, 2845.70it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.output.dense.weight]
Loading weights:  46%|████▌     | 216/472 [00:00<00:00, 2844.44it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.output.dense.weight]
Loading weights:  46%|████▌     | 217/472 [00:00<00:00, 2855.31it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.self.key.bias]      
Loading weights:  46%|████▌     | 217/472 [00:00<00:00, 2854.04it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.self.key.bias]
Loading weights:  46%|████▌     | 218/472 [00:00<00:00, 2864.96it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.self.key.weight]
Loading weights:  46%|████▌     | 218/472 [00:00<00:00, 2863.76it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.self.key.weight]
Loading weights:  46%|████▋     | 219/472 [00:00<00:00, 2874.82it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.self.query.bias]
Loading weights:  46%|████▋     | 219/472 [00:00<00:00, 2873.01it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.self.query.bias]
Loading weights:  47%|████▋     | 220/472 [00:00<00:00, 2883.85it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.self.query.weight]
Loading weights:  47%|████▋     | 220/472 [00:00<00:00, 2882.42it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.self.query.weight]
Loading weights:  47%|████▋     | 221/472 [00:00<00:00, 2893.47it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.self.value.bias]  
Loading weights:  47%|████▋     | 221/472 [00:00<00:00, 2892.15it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.self.value.bias]
Loading weights:  47%|████▋     | 222/472 [00:00<00:00, 2902.84it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.self.value.weight]
Loading weights:  47%|████▋     | 222/472 [00:00<00:00, 2901.55it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.self.value.weight]
Loading weights:  47%|████▋     | 223/472 [00:00<00:00, 2912.52it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.output.LayerNorm.bias]
Loading weights:  47%|████▋     | 223/472 [00:00<00:00, 2911.24it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.output.LayerNorm.bias]
Loading weights:  47%|████▋     | 224/472 [00:00<00:00, 2921.96it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.output.LayerNorm.weight]
Loading weights:  47%|████▋     | 224/472 [00:00<00:00, 2920.67it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.output.LayerNorm.weight]
Loading weights:  48%|████▊     | 225/472 [00:00<00:00, 2931.60it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.output.dense.bias]      
Loading weights:  48%|████▊     | 225/472 [00:00<00:00, 2930.38it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.output.dense.bias]
Loading weights:  48%|████▊     | 226/472 [00:00<00:00, 2941.34it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.output.dense.weight]
Loading weights:  48%|████▊     | 226/472 [00:00<00:00, 2940.08it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.output.dense.weight]
Loading weights:  48%|████▊     | 227/472 [00:00<00:00, 2951.05it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.self.key.bias]      
Loading weights:  48%|████▊     | 227/472 [00:00<00:00, 2949.83it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.self.key.bias]
Loading weights:  48%|████▊     | 228/472 [00:00<00:00, 2960.74it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.self.key.weight]
Loading weights:  48%|████▊     | 228/472 [00:00<00:00, 2959.51it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.self.key.weight]
Loading weights:  49%|████▊     | 229/472 [00:00<00:00, 2970.45it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.self.query.bias]
Loading weights:  49%|████▊     | 229/472 [00:00<00:00, 2969.21it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.self.query.bias]
Loading weights:  49%|████▊     | 230/472 [00:00<00:00, 2980.17it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.self.query.weight]
Loading weights:  49%|████▊     | 230/472 [00:00<00:00, 2978.94it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.self.query.weight]
Loading weights:  49%|████▉     | 231/472 [00:00<00:00, 2989.88it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.self.value.bias]  
Loading weights:  49%|████▉     | 231/472 [00:00<00:00, 2988.48it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.self.value.bias]
Loading weights:  49%|████▉     | 232/472 [00:00<00:00, 2999.37it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.self.value.weight]
Loading weights:  49%|████▉     | 232/472 [00:00<00:00, 2998.12it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.self.value.weight]
Loading weights:  49%|████▉     | 233/472 [00:00<00:00, 3008.86it/s, Materializing param=text_decoder.bert.encoder.layer.8.intermediate.dense.bias]         
Loading weights:  49%|████▉     | 233/472 [00:00<00:00, 3007.62it/s, Materializing param=text_decoder.bert.encoder.layer.8.intermediate.dense.bias]
Loading weights:  50%|████▉     | 234/472 [00:00<00:00, 3018.51it/s, Materializing param=text_decoder.bert.encoder.layer.8.intermediate.dense.weight]
Loading weights:  50%|████▉     | 234/472 [00:00<00:00, 3017.26it/s, Materializing param=text_decoder.bert.encoder.layer.8.intermediate.dense.weight]
Loading weights:  50%|████▉     | 235/472 [00:00<00:00, 3028.20it/s, Materializing param=text_decoder.bert.encoder.layer.8.output.LayerNorm.bias]    
Loading weights:  50%|████▉     | 235/472 [00:00<00:00, 3026.96it/s, Materializing param=text_decoder.bert.encoder.layer.8.output.LayerNorm.bias]
Loading weights:  50%|█████     | 236/472 [00:00<00:00, 3037.80it/s, Materializing param=text_decoder.bert.encoder.layer.8.output.LayerNorm.weight]
Loading weights:  50%|█████     | 236/472 [00:00<00:00, 3036.55it/s, Materializing param=text_decoder.bert.encoder.layer.8.output.LayerNorm.weight]
Loading weights:  50%|█████     | 237/472 [00:00<00:00, 3047.41it/s, Materializing param=text_decoder.bert.encoder.layer.8.output.dense.bias]      
Loading weights:  50%|█████     | 237/472 [00:00<00:00, 3046.16it/s, Materializing param=text_decoder.bert.encoder.layer.8.output.dense.bias]
Loading weights:  50%|█████     | 238/472 [00:00<00:00, 3056.89it/s, Materializing param=text_decoder.bert.encoder.layer.8.output.dense.weight]
Loading weights:  50%|█████     | 238/472 [00:00<00:00, 3055.64it/s, Materializing param=text_decoder.bert.encoder.layer.8.output.dense.weight]
Loading weights:  51%|█████     | 239/472 [00:00<00:00, 3066.47it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.output.LayerNorm.bias]
Loading weights:  51%|█████     | 239/472 [00:00<00:00, 3065.21it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.output.LayerNorm.bias]
Loading weights:  51%|█████     | 240/472 [00:00<00:00, 3075.63it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.output.LayerNorm.weight]
Loading weights:  51%|█████     | 240/472 [00:00<00:00, 3074.33it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.output.LayerNorm.weight]
Loading weights:  51%|█████     | 241/472 [00:00<00:00, 3085.08it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.output.dense.bias]      
Loading weights:  51%|█████     | 241/472 [00:00<00:00, 3083.82it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.output.dense.bias]
Loading weights:  51%|█████▏    | 242/472 [00:00<00:00, 3094.36it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.output.dense.weight]
Loading weights:  51%|█████▏    | 242/472 [00:00<00:00, 3093.09it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.output.dense.weight]
Loading weights:  51%|█████▏    | 243/472 [00:00<00:00, 3103.81it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.self.key.bias]      
Loading weights:  51%|█████▏    | 243/472 [00:00<00:00, 3102.54it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.self.key.bias]
Loading weights:  52%|█████▏    | 244/472 [00:00<00:00, 3113.24it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.self.key.weight]
Loading weights:  52%|█████▏    | 244/472 [00:00<00:00, 3111.97it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.self.key.weight]
Loading weights:  52%|█████▏    | 245/472 [00:00<00:00, 3122.70it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.self.query.bias]
Loading weights:  52%|█████▏    | 245/472 [00:00<00:00, 3121.46it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.self.query.bias]
Loading weights:  52%|█████▏    | 246/472 [00:00<00:00, 3132.05it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.self.query.weight]
Loading weights:  52%|█████▏    | 246/472 [00:00<00:00, 3130.73it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.self.query.weight]
Loading weights:  52%|█████▏    | 247/472 [00:00<00:00, 3141.38it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.self.value.bias]  
Loading weights:  52%|█████▏    | 247/472 [00:00<00:00, 3139.98it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.self.value.bias]
Loading weights:  53%|█████▎    | 248/472 [00:00<00:00, 3150.45it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.self.value.weight]
Loading weights:  53%|█████▎    | 248/472 [00:00<00:00, 3149.13it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.self.value.weight]
Loading weights:  53%|█████▎    | 249/472 [00:00<00:00, 3159.63it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.output.LayerNorm.bias]
Loading weights:  53%|█████▎    | 249/472 [00:00<00:00, 3158.30it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.output.LayerNorm.bias]
Loading weights:  53%|█████▎    | 250/472 [00:00<00:00, 3168.77it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.output.LayerNorm.weight]
Loading weights:  53%|█████▎    | 250/472 [00:00<00:00, 3167.45it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.output.LayerNorm.weight]
Loading weights:  53%|█████▎    | 251/472 [00:00<00:00, 3177.91it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.output.dense.bias]      
Loading weights:  53%|█████▎    | 251/472 [00:00<00:00, 3176.58it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.output.dense.bias]
Loading weights:  53%|█████▎    | 252/472 [00:00<00:00, 3186.78it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.output.dense.weight]
Loading weights:  53%|█████▎    | 252/472 [00:00<00:00, 3185.40it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.output.dense.weight]
Loading weights:  54%|█████▎    | 253/472 [00:00<00:00, 3195.87it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.self.key.bias]      
Loading weights:  54%|█████▎    | 253/472 [00:00<00:00, 3194.53it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.self.key.bias]
Loading weights:  54%|█████▍    | 254/472 [00:00<00:00, 3205.01it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.self.key.weight]
Loading weights:  54%|█████▍    | 254/472 [00:00<00:00, 3203.76it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.self.key.weight]
Loading weights:  54%|█████▍    | 255/472 [00:00<00:00, 3214.27it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.self.query.bias]
Loading weights:  54%|█████▍    | 255/472 [00:00<00:00, 3213.01it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.self.query.bias]
Loading weights:  54%|█████▍    | 256/472 [00:00<00:00, 3223.54it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.self.query.weight]
Loading weights:  54%|█████▍    | 256/472 [00:00<00:00, 3222.12it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.self.query.weight]
Loading weights:  54%|█████▍    | 257/472 [00:00<00:00, 3232.59it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.self.value.bias]  
Loading weights:  54%|█████▍    | 257/472 [00:00<00:00, 3231.21it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.self.value.bias]
Loading weights:  55%|█████▍    | 258/472 [00:00<00:00, 3241.54it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.self.value.weight]
Loading weights:  55%|█████▍    | 258/472 [00:00<00:00, 3240.15it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.self.value.weight]
Loading weights:  55%|█████▍    | 259/472 [00:00<00:00, 3250.50it/s, Materializing param=text_decoder.bert.encoder.layer.9.intermediate.dense.bias]         
Loading weights:  55%|█████▍    | 259/472 [00:00<00:00, 3249.20it/s, Materializing param=text_decoder.bert.encoder.layer.9.intermediate.dense.bias]
Loading weights:  55%|█████▌    | 260/472 [00:00<00:00, 3259.50it/s, Materializing param=text_decoder.bert.encoder.layer.9.intermediate.dense.weight]
Loading weights:  55%|█████▌    | 260/472 [00:00<00:00, 3258.20it/s, Materializing param=text_decoder.bert.encoder.layer.9.intermediate.dense.weight]
Loading weights:  55%|█████▌    | 261/472 [00:00<00:00, 3268.63it/s, Materializing param=text_decoder.bert.encoder.layer.9.output.LayerNorm.bias]    
Loading weights:  55%|█████▌    | 261/472 [00:00<00:00, 3267.28it/s, Materializing param=text_decoder.bert.encoder.layer.9.output.LayerNorm.bias]
Loading weights:  56%|█████▌    | 262/472 [00:00<00:00, 3277.71it/s, Materializing param=text_decoder.bert.encoder.layer.9.output.LayerNorm.weight]
Loading weights:  56%|█████▌    | 262/472 [00:00<00:00, 3276.36it/s, Materializing param=text_decoder.bert.encoder.layer.9.output.LayerNorm.weight]
Loading weights:  56%|█████▌    | 263/472 [00:00<00:00, 3286.68it/s, Materializing param=text_decoder.bert.encoder.layer.9.output.dense.bias]      
Loading weights:  56%|█████▌    | 263/472 [00:00<00:00, 3285.37it/s, Materializing param=text_decoder.bert.encoder.layer.9.output.dense.bias]
Loading weights:  56%|█████▌    | 264/472 [00:00<00:00, 3295.81it/s, Materializing param=text_decoder.bert.encoder.layer.9.output.dense.weight]
Loading weights:  56%|█████▌    | 264/472 [00:00<00:00, 3294.53it/s, Materializing param=text_decoder.bert.encoder.layer.9.output.dense.weight]
Loading weights:  56%|█████▌    | 265/472 [00:00<00:00, 3304.95it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.output.LayerNorm.bias]
Loading weights:  56%|█████▌    | 265/472 [00:00<00:00, 3303.66it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.output.LayerNorm.bias]
Loading weights:  56%|█████▋    | 266/472 [00:00<00:00, 3313.91it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.output.LayerNorm.weight]
Loading weights:  56%|█████▋    | 266/472 [00:00<00:00, 3312.59it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.output.LayerNorm.weight]
Loading weights:  57%|█████▋    | 267/472 [00:00<00:00, 3322.84it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.output.dense.bias]      
Loading weights:  57%|█████▋    | 267/472 [00:00<00:00, 3321.53it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.output.dense.bias]
Loading weights:  57%|█████▋    | 268/472 [00:00<00:00, 3331.80it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.output.dense.weight]
Loading weights:  57%|█████▋    | 268/472 [00:00<00:00, 3330.52it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.output.dense.weight]
Loading weights:  57%|█████▋    | 269/472 [00:00<00:00, 3340.84it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.self.key.bias]      
Loading weights:  57%|█████▋    | 269/472 [00:00<00:00, 3339.50it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.self.key.bias]
Loading weights:  57%|█████▋    | 270/472 [00:00<00:00, 3349.76it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.self.key.weight]
Loading weights:  57%|█████▋    | 270/472 [00:00<00:00, 3348.43it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.self.key.weight]
Loading weights:  57%|█████▋    | 271/472 [00:00<00:00, 3358.67it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.self.query.bias]
Loading weights:  57%|█████▋    | 271/472 [00:00<00:00, 3357.34it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.self.query.bias]
Loading weights:  58%|█████▊    | 272/472 [00:00<00:00, 3367.56it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.self.query.weight]
Loading weights:  58%|█████▊    | 272/472 [00:00<00:00, 3366.19it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.self.query.weight]
Loading weights:  58%|█████▊    | 273/472 [00:00<00:00, 3376.43it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.self.value.bias]  
Loading weights:  58%|█████▊    | 273/472 [00:00<00:00, 3375.13it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.self.value.bias]
Loading weights:  58%|█████▊    | 274/472 [00:00<00:00, 3385.31it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.self.value.weight]
Loading weights:  58%|█████▊    | 274/472 [00:00<00:00, 3384.02it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.self.value.weight]
Loading weights:  58%|█████▊    | 275/472 [00:00<00:00, 3394.23it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.output.LayerNorm.bias]
Loading weights:  58%|█████▊    | 275/472 [00:00<00:00, 3392.81it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.output.LayerNorm.bias]
Loading weights:  58%|█████▊    | 276/472 [00:00<00:00, 3402.79it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.output.LayerNorm.weight]
Loading weights:  58%|█████▊    | 276/472 [00:00<00:00, 3401.45it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.output.LayerNorm.weight]
Loading weights:  59%|█████▊    | 277/472 [00:00<00:00, 3411.47it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.output.dense.bias]      
Loading weights:  59%|█████▊    | 277/472 [00:00<00:00, 3410.16it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.output.dense.bias]
Loading weights:  59%|█████▉    | 278/472 [00:00<00:00, 3420.25it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.output.dense.weight]
Loading weights:  59%|█████▉    | 278/472 [00:00<00:00, 3418.86it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.output.dense.weight]
Loading weights:  59%|█████▉    | 279/472 [00:00<00:00, 3428.92it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.self.key.bias]      
Loading weights:  59%|█████▉    | 279/472 [00:00<00:00, 3427.57it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.self.key.bias]
Loading weights:  59%|█████▉    | 280/472 [00:00<00:00, 3437.69it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.self.key.weight]
Loading weights:  59%|█████▉    | 280/472 [00:00<00:00, 3436.39it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.self.key.weight]
Loading weights:  60%|█████▉    | 281/472 [00:00<00:00, 3446.47it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.self.query.bias]
Loading weights:  60%|█████▉    | 281/472 [00:00<00:00, 3445.11it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.self.query.bias]
Loading weights:  60%|█████▉    | 282/472 [00:00<00:00, 3450.18it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.self.query.weight]
Loading weights:  60%|█████▉    | 282/472 [00:00<00:00, 3448.17it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.self.query.weight]
Loading weights:  60%|█████▉    | 283/472 [00:00<00:00, 3453.93it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.self.value.bias]  
Loading weights:  60%|█████▉    | 283/472 [00:00<00:00, 3452.24it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.self.value.bias]
Loading weights:  60%|██████    | 284/472 [00:00<00:00, 3461.81it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.self.value.weight]
Loading weights:  60%|██████    | 284/472 [00:00<00:00, 3460.39it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.self.value.weight]
Loading weights:  60%|██████    | 285/472 [00:00<00:00, 3470.29it/s, Materializing param=text_decoder.bert.encoder.layer.10.intermediate.dense.bias]         
Loading weights:  60%|██████    | 285/472 [00:00<00:00, 3468.94it/s, Materializing param=text_decoder.bert.encoder.layer.10.intermediate.dense.bias]
Loading weights:  61%|██████    | 286/472 [00:00<00:00, 3478.91it/s, Materializing param=text_decoder.bert.encoder.layer.10.intermediate.dense.weight]
Loading weights:  61%|██████    | 286/472 [00:00<00:00, 3477.60it/s, Materializing param=text_decoder.bert.encoder.layer.10.intermediate.dense.weight]
Loading weights:  61%|██████    | 287/472 [00:00<00:00, 3487.55it/s, Materializing param=text_decoder.bert.encoder.layer.10.output.LayerNorm.bias]    
Loading weights:  61%|██████    | 287/472 [00:00<00:00, 3486.14it/s, Materializing param=text_decoder.bert.encoder.layer.10.output.LayerNorm.bias]
Loading weights:  61%|██████    | 288/472 [00:00<00:00, 3496.01it/s, Materializing param=text_decoder.bert.encoder.layer.10.output.LayerNorm.weight]
Loading weights:  61%|██████    | 288/472 [00:00<00:00, 3494.69it/s, Materializing param=text_decoder.bert.encoder.layer.10.output.LayerNorm.weight]
Loading weights:  61%|██████    | 289/472 [00:00<00:00, 3504.61it/s, Materializing param=text_decoder.bert.encoder.layer.10.output.dense.bias]      
Loading weights:  61%|██████    | 289/472 [00:00<00:00, 3503.29it/s, Materializing param=text_decoder.bert.encoder.layer.10.output.dense.bias]
Loading weights:  61%|██████▏   | 290/472 [00:00<00:00, 3513.28it/s, Materializing param=text_decoder.bert.encoder.layer.10.output.dense.weight]
Loading weights:  61%|██████▏   | 290/472 [00:00<00:00, 3511.66it/s, Materializing param=text_decoder.bert.encoder.layer.10.output.dense.weight]
Loading weights:  62%|██████▏   | 291/472 [00:00<00:00, 3521.65it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.output.LayerNorm.bias]
Loading weights:  62%|██████▏   | 291/472 [00:00<00:00, 3520.23it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.output.LayerNorm.bias]
Loading weights:  62%|██████▏   | 292/472 [00:00<00:00, 3529.91it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.output.LayerNorm.weight]
Loading weights:  62%|██████▏   | 292/472 [00:00<00:00, 3528.54it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.output.LayerNorm.weight]
Loading weights:  62%|██████▏   | 293/472 [00:00<00:00, 3538.37it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.output.dense.bias]      
Loading weights:  62%|██████▏   | 293/472 [00:00<00:00, 3536.99it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.output.dense.bias]
Loading weights:  62%|██████▏   | 294/472 [00:00<00:00, 3546.83it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.output.dense.weight]
Loading weights:  62%|██████▏   | 294/472 [00:00<00:00, 3545.50it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.output.dense.weight]
Loading weights:  62%|██████▎   | 295/472 [00:00<00:00, 3555.38it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.self.key.bias]      
Loading weights:  62%|██████▎   | 295/472 [00:00<00:00, 3553.96it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.self.key.bias]
Loading weights:  63%|██████▎   | 296/472 [00:00<00:00, 3563.78it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.self.key.weight]
Loading weights:  63%|██████▎   | 296/472 [00:00<00:00, 3562.41it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.self.key.weight]
Loading weights:  63%|██████▎   | 297/472 [00:00<00:00, 3572.21it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.self.query.bias]
Loading weights:  63%|██████▎   | 297/472 [00:00<00:00, 3570.88it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.self.query.bias]
Loading weights:  63%|██████▎   | 298/472 [00:00<00:00, 3580.72it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.self.query.weight]
Loading weights:  63%|██████▎   | 298/472 [00:00<00:00, 3579.25it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.self.query.weight]
Loading weights:  63%|██████▎   | 299/472 [00:00<00:00, 3588.93it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.self.value.bias]  
Loading weights:  63%|██████▎   | 299/472 [00:00<00:00, 3587.59it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.self.value.bias]
Loading weights:  64%|██████▎   | 300/472 [00:00<00:00, 3597.21it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.self.value.weight]
Loading weights:  64%|██████▎   | 300/472 [00:00<00:00, 3595.80it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.self.value.weight]
Loading weights:  64%|██████▍   | 301/472 [00:00<00:00, 3605.49it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.output.LayerNorm.bias]
Loading weights:  64%|██████▍   | 301/472 [00:00<00:00, 3604.10it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.output.LayerNorm.bias]
Loading weights:  64%|██████▍   | 302/472 [00:00<00:00, 3613.65it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.output.LayerNorm.weight]
Loading weights:  64%|██████▍   | 302/472 [00:00<00:00, 3612.27it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.output.LayerNorm.weight]
Loading weights:  64%|██████▍   | 303/472 [00:00<00:00, 3621.94it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.output.dense.bias]      
Loading weights:  64%|██████▍   | 303/472 [00:00<00:00, 3620.56it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.output.dense.bias]
Loading weights:  64%|██████▍   | 304/472 [00:00<00:00, 3630.25it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.output.dense.weight]
Loading weights:  64%|██████▍   | 304/472 [00:00<00:00, 3628.82it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.output.dense.weight]
Loading weights:  65%|██████▍   | 305/472 [00:00<00:00, 3638.54it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.self.key.bias]      
Loading weights:  65%|██████▍   | 305/472 [00:00<00:00, 3636.98it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.self.key.bias]
Loading weights:  65%|██████▍   | 306/472 [00:00<00:00, 3646.60it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.self.key.weight]
Loading weights:  65%|██████▍   | 306/472 [00:00<00:00, 3645.20it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.self.key.weight]
Loading weights:  65%|██████▌   | 307/472 [00:00<00:00, 3654.81it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.self.query.bias]
Loading weights:  65%|██████▌   | 307/472 [00:00<00:00, 3653.42it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.self.query.bias]
Loading weights:  65%|██████▌   | 308/472 [00:00<00:00, 3663.10it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.self.query.weight]
Loading weights:  65%|██████▌   | 308/472 [00:00<00:00, 3661.70it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.self.query.weight]
Loading weights:  65%|██████▌   | 309/472 [00:00<00:00, 3671.32it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.self.value.bias]  
Loading weights:  65%|██████▌   | 309/472 [00:00<00:00, 3669.84it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.self.value.bias]
Loading weights:  66%|██████▌   | 310/472 [00:00<00:00, 3679.50it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.self.value.weight]
Loading weights:  66%|██████▌   | 310/472 [00:00<00:00, 3678.14it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.self.value.weight]
Loading weights:  66%|██████▌   | 311/472 [00:00<00:00, 3687.73it/s, Materializing param=text_decoder.bert.encoder.layer.11.intermediate.dense.bias]         
Loading weights:  66%|██████▌   | 311/472 [00:00<00:00, 3686.34it/s, Materializing param=text_decoder.bert.encoder.layer.11.intermediate.dense.bias]
Loading weights:  66%|██████▌   | 312/472 [00:00<00:00, 3695.86it/s, Materializing param=text_decoder.bert.encoder.layer.11.intermediate.dense.weight]
Loading weights:  66%|██████▌   | 312/472 [00:00<00:00, 3694.50it/s, Materializing param=text_decoder.bert.encoder.layer.11.intermediate.dense.weight]
Loading weights:  66%|██████▋   | 313/472 [00:00<00:00, 3704.02it/s, Materializing param=text_decoder.bert.encoder.layer.11.output.LayerNorm.bias]    
Loading weights:  66%|██████▋   | 313/472 [00:00<00:00, 3702.70it/s, Materializing param=text_decoder.bert.encoder.layer.11.output.LayerNorm.bias]
Loading weights:  67%|██████▋   | 314/472 [00:00<00:00, 3712.24it/s, Materializing param=text_decoder.bert.encoder.layer.11.output.LayerNorm.weight]
Loading weights:  67%|██████▋   | 314/472 [00:00<00:00, 3710.89it/s, Materializing param=text_decoder.bert.encoder.layer.11.output.LayerNorm.weight]
Loading weights:  67%|██████▋   | 315/472 [00:00<00:00, 3720.47it/s, Materializing param=text_decoder.bert.encoder.layer.11.output.dense.bias]      
Loading weights:  67%|██████▋   | 315/472 [00:00<00:00, 3719.11it/s, Materializing param=text_decoder.bert.encoder.layer.11.output.dense.bias]
Loading weights:  67%|██████▋   | 316/472 [00:00<00:00, 3728.72it/s, Materializing param=text_decoder.bert.encoder.layer.11.output.dense.weight]
Loading weights:  67%|██████▋   | 316/472 [00:00<00:00, 3727.35it/s, Materializing param=text_decoder.bert.encoder.layer.11.output.dense.weight]
Loading weights:  67%|██████▋   | 317/472 [00:00<00:00, 3736.94it/s, Materializing param=text_decoder.cls.predictions.bias]                     
Loading weights:  67%|██████▋   | 317/472 [00:00<00:00, 3735.49it/s, Materializing param=text_decoder.cls.predictions.bias]
Loading weights:  67%|██████▋   | 318/472 [00:00<00:00, 3745.06it/s, Materializing param=text_decoder.cls.predictions.decoder.weight]
Loading weights:  67%|██████▋   | 318/472 [00:00<00:00, 3743.13it/s, Materializing param=text_decoder.cls.predictions.decoder.weight]
Loading weights:  68%|██████▊   | 319/472 [00:00<00:00, 3752.60it/s, Materializing param=text_decoder.cls.predictions.transform.LayerNorm.bias]
Loading weights:  68%|██████▊   | 319/472 [00:00<00:00, 3751.18it/s, Materializing param=text_decoder.cls.predictions.transform.LayerNorm.bias]
Loading weights:  68%|██████▊   | 320/472 [00:00<00:00, 3760.74it/s, Materializing param=text_decoder.cls.predictions.transform.LayerNorm.weight]
Loading weights:  68%|██████▊   | 320/472 [00:00<00:00, 3759.33it/s, Materializing param=text_decoder.cls.predictions.transform.LayerNorm.weight]
Loading weights:  68%|██████▊   | 321/472 [00:00<00:00, 3768.90it/s, Materializing param=text_decoder.cls.predictions.transform.dense.bias]      
Loading weights:  68%|██████▊   | 321/472 [00:00<00:00, 3767.57it/s, Materializing param=text_decoder.cls.predictions.transform.dense.bias]
Loading weights:  68%|██████▊   | 322/472 [00:00<00:00, 3776.73it/s, Materializing param=text_decoder.cls.predictions.transform.dense.weight]
Loading weights:  68%|██████▊   | 322/472 [00:00<00:00, 3775.41it/s, Materializing param=text_decoder.cls.predictions.transform.dense.weight]
Loading weights:  68%|██████▊   | 323/472 [00:00<00:00, 3784.87it/s, Materializing param=vision_model.embeddings.class_embedding]            
Loading weights:  68%|██████▊   | 323/472 [00:00<00:00, 3780.79it/s, Materializing param=vision_model.embeddings.class_embedding]
Loading weights:  69%|██████▊   | 324/472 [00:00<00:00, 3785.91it/s, Materializing param=vision_model.embeddings.patch_embedding.bias]
Loading weights:  69%|██████▊   | 324/472 [00:00<00:00, 3784.00it/s, Materializing param=vision_model.embeddings.patch_embedding.bias]
Loading weights:  69%|██████▉   | 325/472 [00:00<00:00, 3792.27it/s, Materializing param=vision_model.embeddings.patch_embedding.weight]
Loading weights:  69%|██████▉   | 325/472 [00:00<00:00, 3790.67it/s, Materializing param=vision_model.embeddings.patch_embedding.weight]
Loading weights:  69%|██████▉   | 326/472 [00:00<00:00, 3799.77it/s, Materializing param=vision_model.embeddings.position_embedding]    
Loading weights:  69%|██████▉   | 326/472 [00:00<00:00, 3798.26it/s, Materializing param=vision_model.embeddings.position_embedding]
Loading weights:  69%|██████▉   | 327/472 [00:00<00:00, 3807.47it/s, Materializing param=vision_model.encoder.layers.0.layer_norm1.bias]
Loading weights:  69%|██████▉   | 327/472 [00:00<00:00, 3806.01it/s, Materializing param=vision_model.encoder.layers.0.layer_norm1.bias]
Loading weights:  69%|██████▉   | 328/472 [00:00<00:00, 3814.85it/s, Materializing param=vision_model.encoder.layers.0.layer_norm1.weight]
Loading weights:  69%|██████▉   | 328/472 [00:00<00:00, 3813.43it/s, Materializing param=vision_model.encoder.layers.0.layer_norm1.weight]
Loading weights:  70%|██████▉   | 329/472 [00:00<00:00, 3822.74it/s, Materializing param=vision_model.encoder.layers.0.layer_norm2.bias]  
Loading weights:  70%|██████▉   | 329/472 [00:00<00:00, 3821.28it/s, Materializing param=vision_model.encoder.layers.0.layer_norm2.bias]
Loading weights:  70%|██████▉   | 330/472 [00:00<00:00, 3830.68it/s, Materializing param=vision_model.encoder.layers.0.layer_norm2.weight]
Loading weights:  70%|██████▉   | 330/472 [00:00<00:00, 3829.33it/s, Materializing param=vision_model.encoder.layers.0.layer_norm2.weight]
Loading weights:  70%|███████   | 331/472 [00:00<00:00, 3838.76it/s, Materializing param=vision_model.encoder.layers.0.mlp.fc1.bias]      
Loading weights:  70%|███████   | 331/472 [00:00<00:00, 3837.38it/s, Materializing param=vision_model.encoder.layers.0.mlp.fc1.bias]
Loading weights:  70%|███████   | 332/472 [00:00<00:00, 3846.38it/s, Materializing param=vision_model.encoder.layers.0.mlp.fc1.weight]
Loading weights:  70%|███████   | 332/472 [00:00<00:00, 3845.00it/s, Materializing param=vision_model.encoder.layers.0.mlp.fc1.weight]
Loading weights:  71%|███████   | 333/472 [00:00<00:00, 3854.30it/s, Materializing param=vision_model.encoder.layers.0.mlp.fc2.bias]  
Loading weights:  71%|███████   | 333/472 [00:00<00:00, 3852.97it/s, Materializing param=vision_model.encoder.layers.0.mlp.fc2.bias]
Loading weights:  71%|███████   | 334/472 [00:00<00:00, 3862.26it/s, Materializing param=vision_model.encoder.layers.0.mlp.fc2.weight]
Loading weights:  71%|███████   | 334/472 [00:00<00:00, 3860.84it/s, Materializing param=vision_model.encoder.layers.0.mlp.fc2.weight]
Loading weights:  71%|███████   | 335/472 [00:00<00:00, 3870.10it/s, Materializing param=vision_model.encoder.layers.0.self_attn.projection.bias]
Loading weights:  71%|███████   | 335/472 [00:00<00:00, 3868.69it/s, Materializing param=vision_model.encoder.layers.0.self_attn.projection.bias]
Loading weights:  71%|███████   | 336/472 [00:00<00:00, 3877.82it/s, Materializing param=vision_model.encoder.layers.0.self_attn.projection.weight]
Loading weights:  71%|███████   | 336/472 [00:00<00:00, 3876.38it/s, Materializing param=vision_model.encoder.layers.0.self_attn.projection.weight]
Loading weights:  71%|███████▏  | 337/472 [00:00<00:00, 3885.55it/s, Materializing param=vision_model.encoder.layers.0.self_attn.qkv.bias]         
Loading weights:  71%|███████▏  | 337/472 [00:00<00:00, 3884.15it/s, Materializing param=vision_model.encoder.layers.0.self_attn.qkv.bias]
Loading weights:  72%|███████▏  | 338/472 [00:00<00:00, 3893.34it/s, Materializing param=vision_model.encoder.layers.0.self_attn.qkv.weight]
Loading weights:  72%|███████▏  | 338/472 [00:00<00:00, 3891.91it/s, Materializing param=vision_model.encoder.layers.0.self_attn.qkv.weight]
Loading weights:  72%|███████▏  | 339/472 [00:00<00:00, 3901.23it/s, Materializing param=vision_model.encoder.layers.1.layer_norm1.bias]    
Loading weights:  72%|███████▏  | 339/472 [00:00<00:00, 3899.84it/s, Materializing param=vision_model.encoder.layers.1.layer_norm1.bias]
Loading weights:  72%|███████▏  | 340/472 [00:00<00:00, 3909.05it/s, Materializing param=vision_model.encoder.layers.1.layer_norm1.weight]
Loading weights:  72%|███████▏  | 340/472 [00:00<00:00, 3907.64it/s, Materializing param=vision_model.encoder.layers.1.layer_norm1.weight]
Loading weights:  72%|███████▏  | 341/472 [00:00<00:00, 3916.94it/s, Materializing param=vision_model.encoder.layers.1.layer_norm2.bias]  
Loading weights:  72%|███████▏  | 341/472 [00:00<00:00, 3915.54it/s, Materializing param=vision_model.encoder.layers.1.layer_norm2.bias]
Loading weights:  72%|███████▏  | 342/472 [00:00<00:00, 3924.40it/s, Materializing param=vision_model.encoder.layers.1.layer_norm2.weight]
Loading weights:  72%|███████▏  | 342/472 [00:00<00:00, 3923.01it/s, Materializing param=vision_model.encoder.layers.1.layer_norm2.weight]
Loading weights:  73%|███████▎  | 343/472 [00:00<00:00, 3932.27it/s, Materializing param=vision_model.encoder.layers.1.mlp.fc1.bias]      
Loading weights:  73%|███████▎  | 343/472 [00:00<00:00, 3930.93it/s, Materializing param=vision_model.encoder.layers.1.mlp.fc1.bias]
Loading weights:  73%|███████▎  | 344/472 [00:00<00:00, 3940.12it/s, Materializing param=vision_model.encoder.layers.1.mlp.fc1.weight]
Loading weights:  73%|███████▎  | 344/472 [00:00<00:00, 3938.78it/s, Materializing param=vision_model.encoder.layers.1.mlp.fc1.weight]
Loading weights:  73%|███████▎  | 345/472 [00:00<00:00, 3947.96it/s, Materializing param=vision_model.encoder.layers.1.mlp.fc2.bias]  
Loading weights:  73%|███████▎  | 345/472 [00:00<00:00, 3946.56it/s, Materializing param=vision_model.encoder.layers.1.mlp.fc2.bias]
Loading weights:  73%|███████▎  | 346/472 [00:00<00:00, 3955.70it/s, Materializing param=vision_model.encoder.layers.1.mlp.fc2.weight]
Loading weights:  73%|███████▎  | 346/472 [00:00<00:00, 3954.34it/s, Materializing param=vision_model.encoder.layers.1.mlp.fc2.weight]
Loading weights:  74%|███████▎  | 347/472 [00:00<00:00, 3963.55it/s, Materializing param=vision_model.encoder.layers.1.self_attn.projection.bias]
Loading weights:  74%|███████▎  | 347/472 [00:00<00:00, 3962.05it/s, Materializing param=vision_model.encoder.layers.1.self_attn.projection.bias]
Loading weights:  74%|███████▎  | 348/472 [00:00<00:00, 3970.98it/s, Materializing param=vision_model.encoder.layers.1.self_attn.projection.weight]
Loading weights:  74%|███████▎  | 348/472 [00:00<00:00, 3969.52it/s, Materializing param=vision_model.encoder.layers.1.self_attn.projection.weight]
Loading weights:  74%|███████▍  | 349/472 [00:00<00:00, 3978.63it/s, Materializing param=vision_model.encoder.layers.1.self_attn.qkv.bias]         
Loading weights:  74%|███████▍  | 349/472 [00:00<00:00, 3977.26it/s, Materializing param=vision_model.encoder.layers.1.self_attn.qkv.bias]
Loading weights:  74%|███████▍  | 350/472 [00:00<00:00, 3986.26it/s, Materializing param=vision_model.encoder.layers.1.self_attn.qkv.weight]
Loading weights:  74%|███████▍  | 350/472 [00:00<00:00, 3984.85it/s, Materializing param=vision_model.encoder.layers.1.self_attn.qkv.weight]
Loading weights:  74%|███████▍  | 351/472 [00:00<00:00, 3994.00it/s, Materializing param=vision_model.encoder.layers.2.layer_norm1.bias]    
Loading weights:  74%|███████▍  | 351/472 [00:00<00:00, 3992.54it/s, Materializing param=vision_model.encoder.layers.2.layer_norm1.bias]
Loading weights:  75%|███████▍  | 352/472 [00:00<00:00, 4001.50it/s, Materializing param=vision_model.encoder.layers.2.layer_norm1.weight]
Loading weights:  75%|███████▍  | 352/472 [00:00<00:00, 4000.15it/s, Materializing param=vision_model.encoder.layers.2.layer_norm1.weight]
Loading weights:  75%|███████▍  | 353/472 [00:00<00:00, 4009.29it/s, Materializing param=vision_model.encoder.layers.2.layer_norm2.bias]  
Loading weights:  75%|███████▍  | 353/472 [00:00<00:00, 4007.95it/s, Materializing param=vision_model.encoder.layers.2.layer_norm2.bias]
Loading weights:  75%|███████▌  | 354/472 [00:00<00:00, 4016.89it/s, Materializing param=vision_model.encoder.layers.2.layer_norm2.weight]
Loading weights:  75%|███████▌  | 354/472 [00:00<00:00, 4015.52it/s, Materializing param=vision_model.encoder.layers.2.layer_norm2.weight]
Loading weights:  75%|███████▌  | 355/472 [00:00<00:00, 4024.55it/s, Materializing param=vision_model.encoder.layers.2.mlp.fc1.bias]      
Loading weights:  75%|███████▌  | 355/472 [00:00<00:00, 4023.13it/s, Materializing param=vision_model.encoder.layers.2.mlp.fc1.bias]
Loading weights:  75%|███████▌  | 356/472 [00:00<00:00, 4031.95it/s, Materializing param=vision_model.encoder.layers.2.mlp.fc1.weight]
Loading weights:  75%|███████▌  | 356/472 [00:00<00:00, 4030.54it/s, Materializing param=vision_model.encoder.layers.2.mlp.fc1.weight]
Loading weights:  76%|███████▌  | 357/472 [00:00<00:00, 4039.52it/s, Materializing param=vision_model.encoder.layers.2.mlp.fc2.bias]  
Loading weights:  76%|███████▌  | 357/472 [00:00<00:00, 4038.11it/s, Materializing param=vision_model.encoder.layers.2.mlp.fc2.bias]
Loading weights:  76%|███████▌  | 358/472 [00:00<00:00, 4046.85it/s, Materializing param=vision_model.encoder.layers.2.mlp.fc2.weight]
Loading weights:  76%|███████▌  | 358/472 [00:00<00:00, 4045.44it/s, Materializing param=vision_model.encoder.layers.2.mlp.fc2.weight]
Loading weights:  76%|███████▌  | 359/472 [00:00<00:00, 4054.49it/s, Materializing param=vision_model.encoder.layers.2.self_attn.projection.bias]
Loading weights:  76%|███████▌  | 359/472 [00:00<00:00, 4053.02it/s, Materializing param=vision_model.encoder.layers.2.self_attn.projection.bias]
Loading weights:  76%|███████▋  | 360/472 [00:00<00:00, 4062.07it/s, Materializing param=vision_model.encoder.layers.2.self_attn.projection.weight]
Loading weights:  76%|███████▋  | 360/472 [00:00<00:00, 4060.60it/s, Materializing param=vision_model.encoder.layers.2.self_attn.projection.weight]
Loading weights:  76%|███████▋  | 361/472 [00:00<00:00, 4069.63it/s, Materializing param=vision_model.encoder.layers.2.self_attn.qkv.bias]         
Loading weights:  76%|███████▋  | 361/472 [00:00<00:00, 4068.20it/s, Materializing param=vision_model.encoder.layers.2.self_attn.qkv.bias]
Loading weights:  77%|███████▋  | 362/472 [00:00<00:00, 4077.08it/s, Materializing param=vision_model.encoder.layers.2.self_attn.qkv.weight]
Loading weights:  77%|███████▋  | 362/472 [00:00<00:00, 4075.62it/s, Materializing param=vision_model.encoder.layers.2.self_attn.qkv.weight]
Loading weights:  77%|███████▋  | 363/472 [00:00<00:00, 4084.63it/s, Materializing param=vision_model.encoder.layers.3.layer_norm1.bias]    
Loading weights:  77%|███████▋  | 363/472 [00:00<00:00, 4083.06it/s, Materializing param=vision_model.encoder.layers.3.layer_norm1.bias]
Loading weights:  77%|███████▋  | 364/472 [00:00<00:00, 4091.73it/s, Materializing param=vision_model.encoder.layers.3.layer_norm1.weight]
Loading weights:  77%|███████▋  | 364/472 [00:00<00:00, 4090.30it/s, Materializing param=vision_model.encoder.layers.3.layer_norm1.weight]
Loading weights:  77%|███████▋  | 365/472 [00:00<00:00, 4099.25it/s, Materializing param=vision_model.encoder.layers.3.layer_norm2.bias]  
Loading weights:  77%|███████▋  | 365/472 [00:00<00:00, 4097.78it/s, Materializing param=vision_model.encoder.layers.3.layer_norm2.bias]
Loading weights:  78%|███████▊  | 366/472 [00:00<00:00, 4106.51it/s, Materializing param=vision_model.encoder.layers.3.layer_norm2.weight]
Loading weights:  78%|███████▊  | 366/472 [00:00<00:00, 4105.04it/s, Materializing param=vision_model.encoder.layers.3.layer_norm2.weight]
Loading weights:  78%|███████▊  | 367/472 [00:00<00:00, 4113.95it/s, Materializing param=vision_model.encoder.layers.3.mlp.fc1.bias]      
Loading weights:  78%|███████▊  | 367/472 [00:00<00:00, 4112.57it/s, Materializing param=vision_model.encoder.layers.3.mlp.fc1.bias]
Loading weights:  78%|███████▊  | 368/472 [00:00<00:00, 4121.55it/s, Materializing param=vision_model.encoder.layers.3.mlp.fc1.weight]
Loading weights:  78%|███████▊  | 368/472 [00:00<00:00, 4120.12it/s, Materializing param=vision_model.encoder.layers.3.mlp.fc1.weight]
Loading weights:  78%|███████▊  | 369/472 [00:00<00:00, 4129.01it/s, Materializing param=vision_model.encoder.layers.3.mlp.fc2.bias]  
Loading weights:  78%|███████▊  | 369/472 [00:00<00:00, 4127.58it/s, Materializing param=vision_model.encoder.layers.3.mlp.fc2.bias]
Loading weights:  78%|███████▊  | 370/472 [00:00<00:00, 4136.35it/s, Materializing param=vision_model.encoder.layers.3.mlp.fc2.weight]
Loading weights:  78%|███████▊  | 370/472 [00:00<00:00, 4134.96it/s, Materializing param=vision_model.encoder.layers.3.mlp.fc2.weight]
Loading weights:  79%|███████▊  | 371/472 [00:00<00:00, 4143.92it/s, Materializing param=vision_model.encoder.layers.3.self_attn.projection.bias]
Loading weights:  79%|███████▊  | 371/472 [00:00<00:00, 4142.44it/s, Materializing param=vision_model.encoder.layers.3.self_attn.projection.bias]
Loading weights:  79%|███████▉  | 372/472 [00:00<00:00, 4151.24it/s, Materializing param=vision_model.encoder.layers.3.self_attn.projection.weight]
Loading weights:  79%|███████▉  | 372/472 [00:00<00:00, 4149.79it/s, Materializing param=vision_model.encoder.layers.3.self_attn.projection.weight]
Loading weights:  79%|███████▉  | 373/472 [00:00<00:00, 4158.45it/s, Materializing param=vision_model.encoder.layers.3.self_attn.qkv.bias]         
Loading weights:  79%|███████▉  | 373/472 [00:00<00:00, 4157.01it/s, Materializing param=vision_model.encoder.layers.3.self_attn.qkv.bias]
Loading weights:  79%|███████▉  | 374/472 [00:00<00:00, 4165.93it/s, Materializing param=vision_model.encoder.layers.3.self_attn.qkv.weight]
Loading weights:  79%|███████▉  | 374/472 [00:00<00:00, 4164.55it/s, Materializing param=vision_model.encoder.layers.3.self_attn.qkv.weight]
Loading weights:  79%|███████▉  | 375/472 [00:00<00:00, 4173.39it/s, Materializing param=vision_model.encoder.layers.4.layer_norm1.bias]    
Loading weights:  79%|███████▉  | 375/472 [00:00<00:00, 4172.01it/s, Materializing param=vision_model.encoder.layers.4.layer_norm1.bias]
Loading weights:  80%|███████▉  | 376/472 [00:00<00:00, 4180.72it/s, Materializing param=vision_model.encoder.layers.4.layer_norm1.weight]
Loading weights:  80%|███████▉  | 376/472 [00:00<00:00, 4179.32it/s, Materializing param=vision_model.encoder.layers.4.layer_norm1.weight]
Loading weights:  80%|███████▉  | 377/472 [00:00<00:00, 4188.15it/s, Materializing param=vision_model.encoder.layers.4.layer_norm2.bias]  
Loading weights:  80%|███████▉  | 377/472 [00:00<00:00, 4186.71it/s, Materializing param=vision_model.encoder.layers.4.layer_norm2.bias]
Loading weights:  80%|████████  | 378/472 [00:00<00:00, 4195.44it/s, Materializing param=vision_model.encoder.layers.4.layer_norm2.weight]
Loading weights:  80%|████████  | 378/472 [00:00<00:00, 4193.99it/s, Materializing param=vision_model.encoder.layers.4.layer_norm2.weight]
Loading weights:  80%|████████  | 379/472 [00:00<00:00, 4202.81it/s, Materializing param=vision_model.encoder.layers.4.mlp.fc1.bias]      
Loading weights:  80%|████████  | 379/472 [00:00<00:00, 4201.41it/s, Materializing param=vision_model.encoder.layers.4.mlp.fc1.bias]
Loading weights:  81%|████████  | 380/472 [00:00<00:00, 4210.01it/s, Materializing param=vision_model.encoder.layers.4.mlp.fc1.weight]
Loading weights:  81%|████████  | 380/472 [00:00<00:00, 4208.57it/s, Materializing param=vision_model.encoder.layers.4.mlp.fc1.weight]
Loading weights:  81%|████████  | 381/472 [00:00<00:00, 4217.32it/s, Materializing param=vision_model.encoder.layers.4.mlp.fc2.bias]  
Loading weights:  81%|████████  | 381/472 [00:00<00:00, 4215.87it/s, Materializing param=vision_model.encoder.layers.4.mlp.fc2.bias]
Loading weights:  81%|████████  | 382/472 [00:00<00:00, 4224.56it/s, Materializing param=vision_model.encoder.layers.4.mlp.fc2.weight]
Loading weights:  81%|████████  | 382/472 [00:00<00:00, 4223.06it/s, Materializing param=vision_model.encoder.layers.4.mlp.fc2.weight]
Loading weights:  81%|████████  | 383/472 [00:00<00:00, 4231.86it/s, Materializing param=vision_model.encoder.layers.4.self_attn.projection.bias]
Loading weights:  81%|████████  | 383/472 [00:00<00:00, 4230.37it/s, Materializing param=vision_model.encoder.layers.4.self_attn.projection.bias]
Loading weights:  81%|████████▏ | 384/472 [00:00<00:00, 4239.12it/s, Materializing param=vision_model.encoder.layers.4.self_attn.projection.weight]
Loading weights:  81%|████████▏ | 384/472 [00:00<00:00, 4237.67it/s, Materializing param=vision_model.encoder.layers.4.self_attn.projection.weight]
Loading weights:  82%|████████▏ | 385/472 [00:00<00:00, 4246.41it/s, Materializing param=vision_model.encoder.layers.4.self_attn.qkv.bias]         
Loading weights:  82%|████████▏ | 385/472 [00:00<00:00, 4245.00it/s, Materializing param=vision_model.encoder.layers.4.self_attn.qkv.bias]
Loading weights:  82%|████████▏ | 386/472 [00:00<00:00, 4253.74it/s, Materializing param=vision_model.encoder.layers.4.self_attn.qkv.weight]
Loading weights:  82%|████████▏ | 386/472 [00:00<00:00, 4252.28it/s, Materializing param=vision_model.encoder.layers.4.self_attn.qkv.weight]
Loading weights:  82%|████████▏ | 387/472 [00:00<00:00, 4260.95it/s, Materializing param=vision_model.encoder.layers.5.layer_norm1.bias]    
Loading weights:  82%|████████▏ | 387/472 [00:00<00:00, 4259.54it/s, Materializing param=vision_model.encoder.layers.5.layer_norm1.bias]
Loading weights:  82%|████████▏ | 388/472 [00:00<00:00, 4268.16it/s, Materializing param=vision_model.encoder.layers.5.layer_norm1.weight]
Loading weights:  82%|████████▏ | 388/472 [00:00<00:00, 4266.71it/s, Materializing param=vision_model.encoder.layers.5.layer_norm1.weight]
Loading weights:  82%|████████▏ | 389/472 [00:00<00:00, 4275.48it/s, Materializing param=vision_model.encoder.layers.5.layer_norm2.bias]  
Loading weights:  82%|████████▏ | 389/472 [00:00<00:00, 4273.80it/s, Materializing param=vision_model.encoder.layers.5.layer_norm2.bias]
Loading weights:  83%|████████▎ | 390/472 [00:00<00:00, 4282.48it/s, Materializing param=vision_model.encoder.layers.5.layer_norm2.weight]
Loading weights:  83%|████████▎ | 390/472 [00:00<00:00, 4281.02it/s, Materializing param=vision_model.encoder.layers.5.layer_norm2.weight]
Loading weights:  83%|████████▎ | 391/472 [00:00<00:00, 4289.82it/s, Materializing param=vision_model.encoder.layers.5.mlp.fc1.bias]      
Loading weights:  83%|████████▎ | 391/472 [00:00<00:00, 4288.42it/s, Materializing param=vision_model.encoder.layers.5.mlp.fc1.bias]
Loading weights:  83%|████████▎ | 392/472 [00:00<00:00, 4297.13it/s, Materializing param=vision_model.encoder.layers.5.mlp.fc1.weight]
Loading weights:  83%|████████▎ | 392/472 [00:00<00:00, 4295.70it/s, Materializing param=vision_model.encoder.layers.5.mlp.fc1.weight]
Loading weights:  83%|████████▎ | 393/472 [00:00<00:00, 4300.17it/s, Materializing param=vision_model.encoder.layers.5.mlp.fc2.bias]  
Loading weights:  83%|████████▎ | 393/472 [00:00<00:00, 4297.07it/s, Materializing param=vision_model.encoder.layers.5.mlp.fc2.bias]
Loading weights:  83%|████████▎ | 394/472 [00:00<00:00, 4303.34it/s, Materializing param=vision_model.encoder.layers.5.mlp.fc2.weight]
Loading weights:  83%|████████▎ | 394/472 [00:00<00:00, 4301.47it/s, Materializing param=vision_model.encoder.layers.5.mlp.fc2.weight]
Loading weights:  84%|████████▎ | 395/472 [00:00<00:00, 4309.13it/s, Materializing param=vision_model.encoder.layers.5.self_attn.projection.bias]
Loading weights:  84%|████████▎ | 395/472 [00:00<00:00, 4307.40it/s, Materializing param=vision_model.encoder.layers.5.self_attn.projection.bias]
Loading weights:  84%|████████▍ | 396/472 [00:00<00:00, 4315.37it/s, Materializing param=vision_model.encoder.layers.5.self_attn.projection.weight]
Loading weights:  84%|████████▍ | 396/472 [00:00<00:00, 4313.78it/s, Materializing param=vision_model.encoder.layers.5.self_attn.projection.weight]
Loading weights:  84%|████████▍ | 397/472 [00:00<00:00, 4321.95it/s, Materializing param=vision_model.encoder.layers.5.self_attn.qkv.bias]         
Loading weights:  84%|████████▍ | 397/472 [00:00<00:00, 4320.34it/s, Materializing param=vision_model.encoder.layers.5.self_attn.qkv.bias]
Loading weights:  84%|████████▍ | 398/472 [00:00<00:00, 4328.68it/s, Materializing param=vision_model.encoder.layers.5.self_attn.qkv.weight]
Loading weights:  84%|████████▍ | 398/472 [00:00<00:00, 4327.13it/s, Materializing param=vision_model.encoder.layers.5.self_attn.qkv.weight]
Loading weights:  85%|████████▍ | 399/472 [00:00<00:00, 4335.55it/s, Materializing param=vision_model.encoder.layers.6.layer_norm1.bias]    
Loading weights:  85%|████████▍ | 399/472 [00:00<00:00, 4334.05it/s, Materializing param=vision_model.encoder.layers.6.layer_norm1.bias]
Loading weights:  85%|████████▍ | 400/472 [00:00<00:00, 4342.22it/s, Materializing param=vision_model.encoder.layers.6.layer_norm1.weight]
Loading weights:  85%|████████▍ | 400/472 [00:00<00:00, 4340.75it/s, Materializing param=vision_model.encoder.layers.6.layer_norm1.weight]
Loading weights:  85%|████████▍ | 401/472 [00:00<00:00, 4349.25it/s, Materializing param=vision_model.encoder.layers.6.layer_norm2.bias]  
Loading weights:  85%|████████▍ | 401/472 [00:00<00:00, 4347.79it/s, Materializing param=vision_model.encoder.layers.6.layer_norm2.bias]
Loading weights:  85%|████████▌ | 402/472 [00:00<00:00, 4356.07it/s, Materializing param=vision_model.encoder.layers.6.layer_norm2.weight]
Loading weights:  85%|████████▌ | 402/472 [00:00<00:00, 4354.57it/s, Materializing param=vision_model.encoder.layers.6.layer_norm2.weight]
Loading weights:  85%|████████▌ | 403/472 [00:00<00:00, 4363.00it/s, Materializing param=vision_model.encoder.layers.6.mlp.fc1.bias]      
Loading weights:  85%|████████▌ | 403/472 [00:00<00:00, 4361.58it/s, Materializing param=vision_model.encoder.layers.6.mlp.fc1.bias]
Loading weights:  86%|████████▌ | 404/472 [00:00<00:00, 4370.02it/s, Materializing param=vision_model.encoder.layers.6.mlp.fc1.weight]
Loading weights:  86%|████████▌ | 404/472 [00:00<00:00, 4368.62it/s, Materializing param=vision_model.encoder.layers.6.mlp.fc1.weight]
Loading weights:  86%|████████▌ | 405/472 [00:00<00:00, 4377.01it/s, Materializing param=vision_model.encoder.layers.6.mlp.fc2.bias]  
Loading weights:  86%|████████▌ | 405/472 [00:00<00:00, 4375.61it/s, Materializing param=vision_model.encoder.layers.6.mlp.fc2.bias]
Loading weights:  86%|████████▌ | 406/472 [00:00<00:00, 4384.07it/s, Materializing param=vision_model.encoder.layers.6.mlp.fc2.weight]
Loading weights:  86%|████████▌ | 406/472 [00:00<00:00, 4382.66it/s, Materializing param=vision_model.encoder.layers.6.mlp.fc2.weight]
Loading weights:  86%|████████▌ | 407/472 [00:00<00:00, 4390.99it/s, Materializing param=vision_model.encoder.layers.6.self_attn.projection.bias]
Loading weights:  86%|████████▌ | 407/472 [00:00<00:00, 4389.52it/s, Materializing param=vision_model.encoder.layers.6.self_attn.projection.bias]
Loading weights:  86%|████████▋ | 408/472 [00:00<00:00, 4397.93it/s, Materializing param=vision_model.encoder.layers.6.self_attn.projection.weight]
Loading weights:  86%|████████▋ | 408/472 [00:00<00:00, 4396.38it/s, Materializing param=vision_model.encoder.layers.6.self_attn.projection.weight]
Loading weights:  87%|████████▋ | 409/472 [00:00<00:00, 4404.73it/s, Materializing param=vision_model.encoder.layers.6.self_attn.qkv.bias]         
Loading weights:  87%|████████▋ | 409/472 [00:00<00:00, 4403.21it/s, Materializing param=vision_model.encoder.layers.6.self_attn.qkv.bias]
Loading weights:  87%|████████▋ | 410/472 [00:00<00:00, 4411.47it/s, Materializing param=vision_model.encoder.layers.6.self_attn.qkv.weight]
Loading weights:  87%|████████▋ | 410/472 [00:00<00:00, 4409.98it/s, Materializing param=vision_model.encoder.layers.6.self_attn.qkv.weight]
Loading weights:  87%|████████▋ | 411/472 [00:00<00:00, 4418.36it/s, Materializing param=vision_model.encoder.layers.7.layer_norm1.bias]    
Loading weights:  87%|████████▋ | 411/472 [00:00<00:00, 4416.89it/s, Materializing param=vision_model.encoder.layers.7.layer_norm1.bias]
Loading weights:  87%|████████▋ | 412/472 [00:00<00:00, 4425.12it/s, Materializing param=vision_model.encoder.layers.7.layer_norm1.weight]
Loading weights:  87%|████████▋ | 412/472 [00:00<00:00, 4423.60it/s, Materializing param=vision_model.encoder.layers.7.layer_norm1.weight]
Loading weights:  88%|████████▊ | 413/472 [00:00<00:00, 4431.97it/s, Materializing param=vision_model.encoder.layers.7.layer_norm2.bias]  
Loading weights:  88%|████████▊ | 413/472 [00:00<00:00, 4430.49it/s, Materializing param=vision_model.encoder.layers.7.layer_norm2.bias]
Loading weights:  88%|████████▊ | 414/472 [00:00<00:00, 4434.88it/s, Materializing param=vision_model.encoder.layers.7.layer_norm2.weight]
Loading weights:  88%|████████▊ | 414/472 [00:00<00:00, 4431.65it/s, Materializing param=vision_model.encoder.layers.7.layer_norm2.weight]
Loading weights:  88%|████████▊ | 415/472 [00:00<00:00, 4438.61it/s, Materializing param=vision_model.encoder.layers.7.mlp.fc1.bias]      
Loading weights:  88%|████████▊ | 415/472 [00:00<00:00, 4436.43it/s, Materializing param=vision_model.encoder.layers.7.mlp.fc1.bias]
Loading weights:  88%|████████▊ | 416/472 [00:00<00:00, 4444.02it/s, Materializing param=vision_model.encoder.layers.7.mlp.fc1.weight]
Loading weights:  88%|████████▊ | 416/472 [00:00<00:00, 4442.13it/s, Materializing param=vision_model.encoder.layers.7.mlp.fc1.weight]
Loading weights:  88%|████████▊ | 417/472 [00:00<00:00, 4449.62it/s, Materializing param=vision_model.encoder.layers.7.mlp.fc2.bias]  
Loading weights:  88%|████████▊ | 417/472 [00:00<00:00, 4447.67it/s, Materializing param=vision_model.encoder.layers.7.mlp.fc2.bias]
Loading weights:  89%|████████▊ | 418/472 [00:00<00:00, 4455.60it/s, Materializing param=vision_model.encoder.layers.7.mlp.fc2.weight]
Loading weights:  89%|████████▊ | 418/472 [00:00<00:00, 4453.88it/s, Materializing param=vision_model.encoder.layers.7.mlp.fc2.weight]
Loading weights:  89%|████████▉ | 419/472 [00:00<00:00, 4461.78it/s, Materializing param=vision_model.encoder.layers.7.self_attn.projection.bias]
Loading weights:  89%|████████▉ | 419/472 [00:00<00:00, 4459.83it/s, Materializing param=vision_model.encoder.layers.7.self_attn.projection.bias]
Loading weights:  89%|████████▉ | 420/472 [00:00<00:00, 4467.53it/s, Materializing param=vision_model.encoder.layers.7.self_attn.projection.weight]
Loading weights:  89%|████████▉ | 420/472 [00:00<00:00, 4465.76it/s, Materializing param=vision_model.encoder.layers.7.self_attn.projection.weight]
Loading weights:  89%|████████▉ | 421/472 [00:00<00:00, 4473.50it/s, Materializing param=vision_model.encoder.layers.7.self_attn.qkv.bias]         
Loading weights:  89%|████████▉ | 421/472 [00:00<00:00, 4471.60it/s, Materializing param=vision_model.encoder.layers.7.self_attn.qkv.bias]
Loading weights:  89%|████████▉ | 422/472 [00:00<00:00, 4479.27it/s, Materializing param=vision_model.encoder.layers.7.self_attn.qkv.weight]
Loading weights:  89%|████████▉ | 422/472 [00:00<00:00, 4477.56it/s, Materializing param=vision_model.encoder.layers.7.self_attn.qkv.weight]
Loading weights:  90%|████████▉ | 423/472 [00:00<00:00, 4484.59it/s, Materializing param=vision_model.encoder.layers.8.layer_norm1.bias]    
Loading weights:  90%|████████▉ | 423/472 [00:00<00:00, 4482.89it/s, Materializing param=vision_model.encoder.layers.8.layer_norm1.bias]
Loading weights:  90%|████████▉ | 424/472 [00:00<00:00, 4490.35it/s, Materializing param=vision_model.encoder.layers.8.layer_norm1.weight]
Loading weights:  90%|████████▉ | 424/472 [00:00<00:00, 4488.58it/s, Materializing param=vision_model.encoder.layers.8.layer_norm1.weight]
Loading weights:  90%|█████████ | 425/472 [00:00<00:00, 4496.50it/s, Materializing param=vision_model.encoder.layers.8.layer_norm2.bias]  
Loading weights:  90%|█████████ | 425/472 [00:00<00:00, 4494.80it/s, Materializing param=vision_model.encoder.layers.8.layer_norm2.bias]
Loading weights:  90%|█████████ | 426/472 [00:00<00:00, 4502.80it/s, Materializing param=vision_model.encoder.layers.8.layer_norm2.weight]
Loading weights:  90%|█████████ | 426/472 [00:00<00:00, 4500.95it/s, Materializing param=vision_model.encoder.layers.8.layer_norm2.weight]
Loading weights:  90%|█████████ | 427/472 [00:00<00:00, 4508.89it/s, Materializing param=vision_model.encoder.layers.8.mlp.fc1.bias]      
Loading weights:  90%|█████████ | 427/472 [00:00<00:00, 4507.33it/s, Materializing param=vision_model.encoder.layers.8.mlp.fc1.bias]
Loading weights:  91%|█████████ | 428/472 [00:00<00:00, 4515.21it/s, Materializing param=vision_model.encoder.layers.8.mlp.fc1.weight]
Loading weights:  91%|█████████ | 428/472 [00:00<00:00, 4513.49it/s, Materializing param=vision_model.encoder.layers.8.mlp.fc1.weight]
Loading weights:  91%|█████████ | 429/472 [00:00<00:00, 4521.37it/s, Materializing param=vision_model.encoder.layers.8.mlp.fc2.bias]  
Loading weights:  91%|█████████ | 429/472 [00:00<00:00, 4519.55it/s, Materializing param=vision_model.encoder.layers.8.mlp.fc2.bias]
Loading weights:  91%|█████████ | 430/472 [00:00<00:00, 4527.46it/s, Materializing param=vision_model.encoder.layers.8.mlp.fc2.weight]
Loading weights:  91%|█████████ | 430/472 [00:00<00:00, 4525.66it/s, Materializing param=vision_model.encoder.layers.8.mlp.fc2.weight]
Loading weights:  91%|█████████▏| 431/472 [00:00<00:00, 4533.42it/s, Materializing param=vision_model.encoder.layers.8.self_attn.projection.bias]
Loading weights:  91%|█████████▏| 431/472 [00:00<00:00, 4531.61it/s, Materializing param=vision_model.encoder.layers.8.self_attn.projection.bias]
Loading weights:  92%|█████████▏| 432/472 [00:00<00:00, 4539.54it/s, Materializing param=vision_model.encoder.layers.8.self_attn.projection.weight]
Loading weights:  92%|█████████▏| 432/472 [00:00<00:00, 4537.82it/s, Materializing param=vision_model.encoder.layers.8.self_attn.projection.weight]
Loading weights:  92%|█████████▏| 433/472 [00:00<00:00, 4545.56it/s, Materializing param=vision_model.encoder.layers.8.self_attn.qkv.bias]         
Loading weights:  92%|█████████▏| 433/472 [00:00<00:00, 4543.84it/s, Materializing param=vision_model.encoder.layers.8.self_attn.qkv.bias]
Loading weights:  92%|█████████▏| 434/472 [00:00<00:00, 4551.86it/s, Materializing param=vision_model.encoder.layers.8.self_attn.qkv.weight]
Loading weights:  92%|█████████▏| 434/472 [00:00<00:00, 4550.32it/s, Materializing param=vision_model.encoder.layers.8.self_attn.qkv.weight]
Loading weights:  92%|█████████▏| 435/472 [00:00<00:00, 4558.05it/s, Materializing param=vision_model.encoder.layers.9.layer_norm1.bias]    
Loading weights:  92%|█████████▏| 435/472 [00:00<00:00, 4556.27it/s, Materializing param=vision_model.encoder.layers.9.layer_norm1.bias]
Loading weights:  92%|█████████▏| 436/472 [00:00<00:00, 4563.78it/s, Materializing param=vision_model.encoder.layers.9.layer_norm1.weight]
Loading weights:  92%|█████████▏| 436/472 [00:00<00:00, 4562.21it/s, Materializing param=vision_model.encoder.layers.9.layer_norm1.weight]
Loading weights:  93%|█████████▎| 437/472 [00:00<00:00, 4570.00it/s, Materializing param=vision_model.encoder.layers.9.layer_norm2.bias]  
Loading weights:  93%|█████████▎| 437/472 [00:00<00:00, 4568.37it/s, Materializing param=vision_model.encoder.layers.9.layer_norm2.bias]
Loading weights:  93%|█████████▎| 438/472 [00:00<00:00, 4576.33it/s, Materializing param=vision_model.encoder.layers.9.layer_norm2.weight]
Loading weights:  93%|█████████▎| 438/472 [00:00<00:00, 4574.85it/s, Materializing param=vision_model.encoder.layers.9.layer_norm2.weight]
Loading weights:  93%|█████████▎| 439/472 [00:00<00:00, 4582.71it/s, Materializing param=vision_model.encoder.layers.9.mlp.fc1.bias]      
Loading weights:  93%|█████████▎| 439/472 [00:00<00:00, 4581.22it/s, Materializing param=vision_model.encoder.layers.9.mlp.fc1.bias]
Loading weights:  93%|█████████▎| 440/472 [00:00<00:00, 4589.08it/s, Materializing param=vision_model.encoder.layers.9.mlp.fc1.weight]
Loading weights:  93%|█████████▎| 440/472 [00:00<00:00, 4587.60it/s, Materializing param=vision_model.encoder.layers.9.mlp.fc1.weight]
Loading weights:  93%|█████████▎| 441/472 [00:00<00:00, 4595.68it/s, Materializing param=vision_model.encoder.layers.9.mlp.fc2.bias]  
Loading weights:  93%|█████████▎| 441/472 [00:00<00:00, 4594.23it/s, Materializing param=vision_model.encoder.layers.9.mlp.fc2.bias]
Loading weights:  94%|█████████▎| 442/472 [00:00<00:00, 4602.35it/s, Materializing param=vision_model.encoder.layers.9.mlp.fc2.weight]
Loading weights:  94%|█████████▎| 442/472 [00:00<00:00, 4600.92it/s, Materializing param=vision_model.encoder.layers.9.mlp.fc2.weight]
Loading weights:  94%|█████████▍| 443/472 [00:00<00:00, 4609.02it/s, Materializing param=vision_model.encoder.layers.9.self_attn.projection.bias]
Loading weights:  94%|█████████▍| 443/472 [00:00<00:00, 4607.58it/s, Materializing param=vision_model.encoder.layers.9.self_attn.projection.bias]
Loading weights:  94%|█████████▍| 444/472 [00:00<00:00, 4615.55it/s, Materializing param=vision_model.encoder.layers.9.self_attn.projection.weight]
Loading weights:  94%|█████████▍| 444/472 [00:00<00:00, 4614.00it/s, Materializing param=vision_model.encoder.layers.9.self_attn.projection.weight]
Loading weights:  94%|█████████▍| 445/472 [00:00<00:00, 4622.08it/s, Materializing param=vision_model.encoder.layers.9.self_attn.qkv.bias]         
Loading weights:  94%|█████████▍| 445/472 [00:00<00:00, 4620.41it/s, Materializing param=vision_model.encoder.layers.9.self_attn.qkv.bias]
Loading weights:  94%|█████████▍| 446/472 [00:00<00:00, 4628.25it/s, Materializing param=vision_model.encoder.layers.9.self_attn.qkv.weight]
Loading weights:  94%|█████████▍| 446/472 [00:00<00:00, 4626.67it/s, Materializing param=vision_model.encoder.layers.9.self_attn.qkv.weight]
Loading weights:  95%|█████████▍| 447/472 [00:00<00:00, 4634.40it/s, Materializing param=vision_model.encoder.layers.10.layer_norm1.bias]   
Loading weights:  95%|█████████▍| 447/472 [00:00<00:00, 4632.86it/s, Materializing param=vision_model.encoder.layers.10.layer_norm1.bias]
Loading weights:  95%|█████████▍| 448/472 [00:00<00:00, 4640.47it/s, Materializing param=vision_model.encoder.layers.10.layer_norm1.weight]
Loading weights:  95%|█████████▍| 448/472 [00:00<00:00, 4638.94it/s, Materializing param=vision_model.encoder.layers.10.layer_norm1.weight]
Loading weights:  95%|█████████▌| 449/472 [00:00<00:00, 4646.75it/s, Materializing param=vision_model.encoder.layers.10.layer_norm2.bias]  
Loading weights:  95%|█████████▌| 449/472 [00:00<00:00, 4645.20it/s, Materializing param=vision_model.encoder.layers.10.layer_norm2.bias]
Loading weights:  95%|█████████▌| 450/472 [00:00<00:00, 4652.90it/s, Materializing param=vision_model.encoder.layers.10.layer_norm2.weight]
Loading weights:  95%|█████████▌| 450/472 [00:00<00:00, 4651.36it/s, Materializing param=vision_model.encoder.layers.10.layer_norm2.weight]
Loading weights:  96%|█████████▌| 451/472 [00:00<00:00, 4659.14it/s, Materializing param=vision_model.encoder.layers.10.mlp.fc1.bias]      
Loading weights:  96%|█████████▌| 451/472 [00:00<00:00, 4657.47it/s, Materializing param=vision_model.encoder.layers.10.mlp.fc1.bias]
Loading weights:  96%|█████████▌| 452/472 [00:00<00:00, 4665.25it/s, Materializing param=vision_model.encoder.layers.10.mlp.fc1.weight]
Loading weights:  96%|█████████▌| 452/472 [00:00<00:00, 4663.61it/s, Materializing param=vision_model.encoder.layers.10.mlp.fc1.weight]
Loading weights:  96%|█████████▌| 453/472 [00:00<00:00, 4671.42it/s, Materializing param=vision_model.encoder.layers.10.mlp.fc2.bias]  
Loading weights:  96%|█████████▌| 453/472 [00:00<00:00, 4669.78it/s, Materializing param=vision_model.encoder.layers.10.mlp.fc2.bias]
Loading weights:  96%|█████████▌| 454/472 [00:00<00:00, 4677.49it/s, Materializing param=vision_model.encoder.layers.10.mlp.fc2.weight]
Loading weights:  96%|█████████▌| 454/472 [00:00<00:00, 4675.99it/s, Materializing param=vision_model.encoder.layers.10.mlp.fc2.weight]
Loading weights:  96%|█████████▋| 455/472 [00:00<00:00, 4683.68it/s, Materializing param=vision_model.encoder.layers.10.self_attn.projection.bias]
Loading weights:  96%|█████████▋| 455/472 [00:00<00:00, 4682.10it/s, Materializing param=vision_model.encoder.layers.10.self_attn.projection.bias]
Loading weights:  97%|█████████▋| 456/472 [00:00<00:00, 4689.97it/s, Materializing param=vision_model.encoder.layers.10.self_attn.projection.weight]
Loading weights:  97%|█████████▋| 456/472 [00:00<00:00, 4688.48it/s, Materializing param=vision_model.encoder.layers.10.self_attn.projection.weight]
Loading weights:  97%|█████████▋| 457/472 [00:00<00:00, 4696.19it/s, Materializing param=vision_model.encoder.layers.10.self_attn.qkv.bias]         
Loading weights:  97%|█████████▋| 457/472 [00:00<00:00, 4694.69it/s, Materializing param=vision_model.encoder.layers.10.self_attn.qkv.bias]
Loading weights:  97%|█████████▋| 458/472 [00:00<00:00, 4702.50it/s, Materializing param=vision_model.encoder.layers.10.self_attn.qkv.weight]
Loading weights:  97%|█████████▋| 458/472 [00:00<00:00, 4700.91it/s, Materializing param=vision_model.encoder.layers.10.self_attn.qkv.weight]
Loading weights:  97%|█████████▋| 459/472 [00:00<00:00, 4707.55it/s, Materializing param=vision_model.encoder.layers.11.layer_norm1.bias]    
Loading weights:  97%|█████████▋| 459/472 [00:00<00:00, 4705.63it/s, Materializing param=vision_model.encoder.layers.11.layer_norm1.bias]
Loading weights:  97%|█████████▋| 460/472 [00:00<00:00, 4712.10it/s, Materializing param=vision_model.encoder.layers.11.layer_norm1.weight]
Loading weights:  97%|█████████▋| 460/472 [00:00<00:00, 4709.37it/s, Materializing param=vision_model.encoder.layers.11.layer_norm1.weight]
Loading weights:  98%|█████████▊| 461/472 [00:00<00:00, 4716.03it/s, Materializing param=vision_model.encoder.layers.11.layer_norm2.bias]  
Loading weights:  98%|█████████▊| 461/472 [00:00<00:00, 4714.29it/s, Materializing param=vision_model.encoder.layers.11.layer_norm2.bias]
Loading weights:  98%|█████████▊| 462/472 [00:00<00:00, 4721.52it/s, Materializing param=vision_model.encoder.layers.11.layer_norm2.weight]
Loading weights:  98%|█████████▊| 462/472 [00:00<00:00, 4719.93it/s, Materializing param=vision_model.encoder.layers.11.layer_norm2.weight]
Loading weights:  98%|█████████▊| 463/472 [00:00<00:00, 4727.59it/s, Materializing param=vision_model.encoder.layers.11.mlp.fc1.bias]      
Loading weights:  98%|█████████▊| 463/472 [00:00<00:00, 4726.10it/s, Materializing param=vision_model.encoder.layers.11.mlp.fc1.bias]
Loading weights:  98%|█████████▊| 464/472 [00:00<00:00, 4733.79it/s, Materializing param=vision_model.encoder.layers.11.mlp.fc1.weight]
Loading weights:  98%|█████████▊| 464/472 [00:00<00:00, 4732.25it/s, Materializing param=vision_model.encoder.layers.11.mlp.fc1.weight]
Loading weights:  99%|█████████▊| 465/472 [00:00<00:00, 4739.94it/s, Materializing param=vision_model.encoder.layers.11.mlp.fc2.bias]  
Loading weights:  99%|█████████▊| 465/472 [00:00<00:00, 4738.38it/s, Materializing param=vision_model.encoder.layers.11.mlp.fc2.bias]
Loading weights:  99%|█████████▊| 466/472 [00:00<00:00, 4746.10it/s, Materializing param=vision_model.encoder.layers.11.mlp.fc2.weight]
Loading weights:  99%|█████████▊| 466/472 [00:00<00:00, 4744.70it/s, Materializing param=vision_model.encoder.layers.11.mlp.fc2.weight]
Loading weights:  99%|█████████▉| 467/472 [00:00<00:00, 4752.42it/s, Materializing param=vision_model.encoder.layers.11.self_attn.projection.bias]
Loading weights:  99%|█████████▉| 467/472 [00:00<00:00, 4750.87it/s, Materializing param=vision_model.encoder.layers.11.self_attn.projection.bias]
Loading weights:  99%|█████████▉| 468/472 [00:00<00:00, 4757.75it/s, Materializing param=vision_model.encoder.layers.11.self_attn.projection.weight]
Loading weights:  99%|█████████▉| 468/472 [00:00<00:00, 4756.21it/s, Materializing param=vision_model.encoder.layers.11.self_attn.projection.weight]
Loading weights:  99%|█████████▉| 469/472 [00:00<00:00, 4763.75it/s, Materializing param=vision_model.encoder.layers.11.self_attn.qkv.bias]         
Loading weights:  99%|█████████▉| 469/472 [00:00<00:00, 4762.20it/s, Materializing param=vision_model.encoder.layers.11.self_attn.qkv.bias]
Loading weights: 100%|█████████▉| 470/472 [00:00<00:00, 4769.84it/s, Materializing param=vision_model.encoder.layers.11.self_attn.qkv.weight]
Loading weights: 100%|█████████▉| 470/472 [00:00<00:00, 4768.34it/s, Materializing param=vision_model.encoder.layers.11.self_attn.qkv.weight]
Loading weights: 100%|█████████▉| 471/472 [00:00<00:00, 4775.97it/s, Materializing param=vision_model.post_layernorm.bias]                   
Loading weights: 100%|█████████▉| 471/472 [00:00<00:00, 4774.47it/s, Materializing param=vision_model.post_layernorm.bias]
Loading weights: 100%|██████████| 472/472 [00:00<00:00, 4782.37it/s, Materializing param=vision_model.post_layernorm.weight]
Loading weights: 100%|██████████| 472/472 [00:00<00:00, 4780.87it/s, Materializing param=vision_model.post_layernorm.weight]
Loading weights: 100%|██████████| 472/472 [00:00<00:00, 4777.01it/s, Materializing param=vision_model.post_layernorm.weight]
The tied weights mapping and config for this model specifies to tie text_decoder.bert.embeddings.word_embeddings.weight to text_decoder.cls.predictions.decoder.weight, but both are present in the checkpoints, so we will NOT tie them. You should update the config with `tie_word_embeddings=False` to silence this warning
BlipForConditionalGeneration LOAD REPORT from: Salesforce/blip-image-captioning-base
Key                                       | Status     |  | 
------------------------------------------+------------+--+-
text_decoder.bert.embeddings.position_ids | UNEXPECTED |  | 

Notes:
- UNEXPECTED	:can be ignored when loading from different task/architecture; not ok if you expect identical arch.

Loading weights:   0%|          | 0/788 [00:00<?, ?it/s]
Loading weights:   0%|          | 1/788 [00:00<00:00, 50533.78it/s, Materializing param=text_decoder.bert.embeddings.LayerNorm.bias]
Loading weights:   0%|          | 1/788 [00:00<00:00, 12520.31it/s, Materializing param=text_decoder.bert.embeddings.LayerNorm.bias]
Loading weights:   0%|          | 2/788 [00:00<00:00, 10538.45it/s, Materializing param=text_decoder.bert.embeddings.LayerNorm.weight]
Loading weights:   0%|          | 2/788 [00:00<00:00, 7489.83it/s, Materializing param=text_decoder.bert.embeddings.LayerNorm.weight] 
Loading weights:   0%|          | 3/788 [00:00<00:00, 4800.81it/s, Materializing param=text_decoder.bert.embeddings.position_embeddings.weight]
Loading weights:   0%|          | 3/788 [00:00<00:00, 2521.63it/s, Materializing param=text_decoder.bert.embeddings.position_embeddings.weight]
Loading weights:   1%|          | 4/788 [00:00<00:00, 2954.26it/s, Materializing param=text_decoder.bert.embeddings.word_embeddings.weight]    
Loading weights:   1%|          | 4/788 [00:00<00:00, 2293.85it/s, Materializing param=text_decoder.bert.embeddings.word_embeddings.weight]
Loading weights:   1%|          | 5/788 [00:00<00:00, 1992.16it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.output.LayerNorm.bias]
Loading weights:   1%|          | 5/788 [00:00<00:00, 1576.92it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.output.LayerNorm.bias]
Loading weights:   1%|          | 6/788 [00:00<00:00, 1646.65it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.output.LayerNorm.weight]
Loading weights:   1%|          | 6/788 [00:00<00:00, 1615.99it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.output.LayerNorm.weight]
Loading weights:   1%|          | 7/788 [00:00<00:00, 1837.30it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.output.dense.bias]      
Loading weights:   1%|          | 7/788 [00:00<00:00, 1678.68it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.output.dense.bias]
Loading weights:   1%|          | 8/788 [00:00<00:00, 1744.08it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.output.dense.weight]
Loading weights:   1%|          | 8/788 [00:00<00:00, 1729.08it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.output.dense.weight]
Loading weights:   1%|          | 9/788 [00:00<00:00, 1733.50it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.self.key.bias]      
Loading weights:   1%|          | 9/788 [00:00<00:00, 1648.70it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.self.key.bias]
Loading weights:   1%|▏         | 10/788 [00:00<00:00, 1686.08it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.self.key.weight]
Loading weights:   1%|▏         | 10/788 [00:00<00:00, 1570.14it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.self.key.weight]
Loading weights:   1%|▏         | 11/788 [00:00<00:00, 1340.54it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.self.query.bias]
Loading weights:   1%|▏         | 11/788 [00:00<00:00, 1252.03it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.self.query.bias]
Loading weights:   2%|▏         | 12/788 [00:00<00:00, 1298.61it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.self.query.weight]
Loading weights:   2%|▏         | 12/788 [00:00<00:00, 1285.64it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.self.query.weight]
Loading weights:   2%|▏         | 13/788 [00:00<00:00, 1366.43it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.self.value.bias]  
Loading weights:   2%|▏         | 13/788 [00:00<00:00, 1359.00it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.self.value.bias]
Loading weights:   2%|▏         | 14/788 [00:00<00:00, 1338.72it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.self.value.weight]
Loading weights:   2%|▏         | 14/788 [00:00<00:00, 1329.93it/s, Materializing param=text_decoder.bert.encoder.layer.0.attention.self.value.weight]
Loading weights:   2%|▏         | 15/788 [00:00<00:00, 1346.66it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.output.LayerNorm.bias]
Loading weights:   2%|▏         | 15/788 [00:00<00:00, 1297.72it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.output.LayerNorm.bias]
Loading weights:   2%|▏         | 16/788 [00:00<00:00, 1357.00it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.output.LayerNorm.weight]
Loading weights:   2%|▏         | 16/788 [00:00<00:00, 1316.45it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.output.LayerNorm.weight]
Loading weights:   2%|▏         | 17/788 [00:00<00:00, 1311.54it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.output.dense.bias]      
Loading weights:   2%|▏         | 17/788 [00:00<00:00, 1304.08it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.output.dense.bias]
Loading weights:   2%|▏         | 18/788 [00:00<00:00, 1340.20it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.output.dense.weight]
Loading weights:   2%|▏         | 18/788 [00:00<00:00, 1307.29it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.output.dense.weight]
Loading weights:   2%|▏         | 19/788 [00:00<00:00, 1273.56it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.self.key.bias]      
Loading weights:   2%|▏         | 19/788 [00:00<00:00, 1245.18it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.self.key.bias]
Loading weights:   3%|▎         | 20/788 [00:00<00:00, 1277.23it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.self.key.weight]
Loading weights:   3%|▎         | 20/788 [00:00<00:00, 1270.50it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.self.key.weight]
Loading weights:   3%|▎         | 21/788 [00:00<00:00, 1305.40it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.self.query.bias]
Loading weights:   3%|▎         | 21/788 [00:00<00:00, 1295.26it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.self.query.bias]
Loading weights:   3%|▎         | 22/788 [00:00<00:00, 1321.49it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.self.query.weight]
Loading weights:   3%|▎         | 22/788 [00:00<00:00, 1268.31it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.self.query.weight]
Loading weights:   3%|▎         | 23/788 [00:00<00:00, 1298.93it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.self.value.bias]  
Loading weights:   3%|▎         | 23/788 [00:00<00:00, 1226.36it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.self.value.bias]
Loading weights:   3%|▎         | 24/788 [00:00<00:00, 1266.16it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.self.value.weight]
Loading weights:   3%|▎         | 24/788 [00:00<00:00, 1207.07it/s, Materializing param=text_decoder.bert.encoder.layer.0.crossattention.self.value.weight]
Loading weights:   3%|▎         | 25/788 [00:00<00:00, 1236.78it/s, Materializing param=text_decoder.bert.encoder.layer.0.intermediate.dense.bias]         
Loading weights:   3%|▎         | 25/788 [00:00<00:00, 1232.81it/s, Materializing param=text_decoder.bert.encoder.layer.0.intermediate.dense.bias]
Loading weights:   3%|▎         | 26/788 [00:00<00:00, 1251.99it/s, Materializing param=text_decoder.bert.encoder.layer.0.intermediate.dense.weight]
Loading weights:   3%|▎         | 26/788 [00:00<00:00, 1220.22it/s, Materializing param=text_decoder.bert.encoder.layer.0.intermediate.dense.weight]
Loading weights:   3%|▎         | 27/788 [00:00<00:00, 1258.81it/s, Materializing param=text_decoder.bert.encoder.layer.0.output.LayerNorm.bias]    
Loading weights:   3%|▎         | 27/788 [00:00<00:00, 1256.59it/s, Materializing param=text_decoder.bert.encoder.layer.0.output.LayerNorm.bias]
Loading weights:   4%|▎         | 28/788 [00:00<00:00, 1279.13it/s, Materializing param=text_decoder.bert.encoder.layer.0.output.LayerNorm.weight]
Loading weights:   4%|▎         | 28/788 [00:00<00:00, 1260.08it/s, Materializing param=text_decoder.bert.encoder.layer.0.output.LayerNorm.weight]
Loading weights:   4%|▎         | 29/788 [00:00<00:00, 1278.56it/s, Materializing param=text_decoder.bert.encoder.layer.0.output.dense.bias]      
Loading weights:   4%|▎         | 29/788 [00:00<00:00, 1258.74it/s, Materializing param=text_decoder.bert.encoder.layer.0.output.dense.bias]
Loading weights:   4%|▍         | 30/788 [00:00<00:00, 1230.74it/s, Materializing param=text_decoder.bert.encoder.layer.0.output.dense.weight]
Loading weights:   4%|▍         | 30/788 [00:00<00:00, 1212.76it/s, Materializing param=text_decoder.bert.encoder.layer.0.output.dense.weight]
Loading weights:   4%|▍         | 31/788 [00:00<00:00, 1224.97it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.output.LayerNorm.bias]
Loading weights:   4%|▍         | 31/788 [00:00<00:00, 1206.19it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.output.LayerNorm.bias]
Loading weights:   4%|▍         | 32/788 [00:00<00:00, 1226.77it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.output.LayerNorm.weight]
Loading weights:   4%|▍         | 32/788 [00:00<00:00, 1222.64it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.output.LayerNorm.weight]
Loading weights:   4%|▍         | 33/788 [00:00<00:00, 1198.35it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.output.dense.bias]      
Loading weights:   4%|▍         | 33/788 [00:00<00:00, 1180.94it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.output.dense.bias]
Loading weights:   4%|▍         | 34/788 [00:00<00:00, 1204.02it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.output.dense.weight]
Loading weights:   4%|▍         | 34/788 [00:00<00:00, 1176.89it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.output.dense.weight]
Loading weights:   4%|▍         | 35/788 [00:00<00:00, 1195.98it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.self.key.bias]      
Loading weights:   4%|▍         | 35/788 [00:00<00:00, 1186.13it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.self.key.bias]
Loading weights:   5%|▍         | 36/788 [00:00<00:00, 1175.97it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.self.key.weight]
Loading weights:   5%|▍         | 36/788 [00:00<00:00, 1174.10it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.self.key.weight]
Loading weights:   5%|▍         | 37/788 [00:00<00:00, 1191.94it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.self.query.bias]
Loading weights:   5%|▍         | 37/788 [00:00<00:00, 1190.44it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.self.query.bias]
Loading weights:   5%|▍         | 38/788 [00:00<00:00, 1208.93it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.self.query.weight]
Loading weights:   5%|▍         | 38/788 [00:00<00:00, 1207.23it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.self.query.weight]
Loading weights:   5%|▍         | 39/788 [00:00<00:00, 1194.02it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.self.value.bias]  
Loading weights:   5%|▍         | 39/788 [00:00<00:00, 1189.93it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.self.value.bias]
Loading weights:   5%|▌         | 40/788 [00:00<00:00, 1204.75it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.self.value.weight]
Loading weights:   5%|▌         | 40/788 [00:00<00:00, 1201.96it/s, Materializing param=text_decoder.bert.encoder.layer.1.attention.self.value.weight]
Loading weights:   5%|▌         | 41/788 [00:00<00:00, 1210.80it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.output.LayerNorm.bias]
Loading weights:   5%|▌         | 41/788 [00:00<00:00, 1174.93it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.output.LayerNorm.bias]
Loading weights:   5%|▌         | 42/788 [00:00<00:00, 1199.35it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.output.LayerNorm.weight]
Loading weights:   5%|▌         | 42/788 [00:00<00:00, 1196.83it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.output.LayerNorm.weight]
Loading weights:   5%|▌         | 43/788 [00:00<00:00, 1209.37it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.output.dense.bias]      
Loading weights:   5%|▌         | 43/788 [00:00<00:00, 1198.41it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.output.dense.bias]
Loading weights:   6%|▌         | 44/788 [00:00<00:00, 1196.14it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.output.dense.weight]
Loading weights:   6%|▌         | 44/788 [00:00<00:00, 1186.37it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.output.dense.weight]
Loading weights:   6%|▌         | 45/788 [00:00<00:00, 1205.92it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.self.key.bias]      
Loading weights:   6%|▌         | 45/788 [00:00<00:00, 1176.51it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.self.key.bias]
Loading weights:   6%|▌         | 46/788 [00:00<00:00, 1190.36it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.self.key.weight]
Loading weights:   6%|▌         | 46/788 [00:00<00:00, 1187.47it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.self.key.weight]
Loading weights:   6%|▌         | 47/788 [00:00<00:00, 1184.87it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.self.query.bias]
Loading weights:   6%|▌         | 47/788 [00:00<00:00, 1164.04it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.self.query.bias]
Loading weights:   6%|▌         | 48/788 [00:00<00:00, 1185.13it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.self.query.weight]
Loading weights:   6%|▌         | 48/788 [00:00<00:00, 1176.56it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.self.query.weight]
Loading weights:   6%|▌         | 49/788 [00:00<00:00, 1190.65it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.self.value.bias]  
Loading weights:   6%|▌         | 49/788 [00:00<00:00, 1179.88it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.self.value.bias]
Loading weights:   6%|▋         | 50/788 [00:00<00:00, 1190.00it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.self.value.weight]
Loading weights:   6%|▋         | 50/788 [00:00<00:00, 1188.78it/s, Materializing param=text_decoder.bert.encoder.layer.1.crossattention.self.value.weight]
Loading weights:   6%|▋         | 51/788 [00:00<00:00, 1195.61it/s, Materializing param=text_decoder.bert.encoder.layer.1.intermediate.dense.bias]         
Loading weights:   6%|▋         | 51/788 [00:00<00:00, 1186.77it/s, Materializing param=text_decoder.bert.encoder.layer.1.intermediate.dense.bias]
Loading weights:   7%|▋         | 52/788 [00:00<00:00, 1202.32it/s, Materializing param=text_decoder.bert.encoder.layer.1.intermediate.dense.weight]
Loading weights:   7%|▋         | 52/788 [00:00<00:00, 1199.62it/s, Materializing param=text_decoder.bert.encoder.layer.1.intermediate.dense.weight]
Loading weights:   7%|▋         | 53/788 [00:00<00:00, 1209.36it/s, Materializing param=text_decoder.bert.encoder.layer.1.output.LayerNorm.bias]    
Loading weights:   7%|▋         | 53/788 [00:00<00:00, 1200.89it/s, Materializing param=text_decoder.bert.encoder.layer.1.output.LayerNorm.bias]
Loading weights:   7%|▋         | 54/788 [00:00<00:00, 1215.40it/s, Materializing param=text_decoder.bert.encoder.layer.1.output.LayerNorm.weight]
Loading weights:   7%|▋         | 54/788 [00:00<00:00, 1200.21it/s, Materializing param=text_decoder.bert.encoder.layer.1.output.LayerNorm.weight]
Loading weights:   7%|▋         | 55/788 [00:00<00:00, 1204.50it/s, Materializing param=text_decoder.bert.encoder.layer.1.output.dense.bias]      
Loading weights:   7%|▋         | 55/788 [00:00<00:00, 1203.35it/s, Materializing param=text_decoder.bert.encoder.layer.1.output.dense.bias]
Loading weights:   7%|▋         | 56/788 [00:00<00:00, 1213.83it/s, Materializing param=text_decoder.bert.encoder.layer.1.output.dense.weight]
Loading weights:   7%|▋         | 56/788 [00:00<00:00, 1211.81it/s, Materializing param=text_decoder.bert.encoder.layer.1.output.dense.weight]
Loading weights:   7%|▋         | 57/788 [00:00<00:00, 1210.86it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.output.LayerNorm.bias]
Loading weights:   7%|▋         | 57/788 [00:00<00:00, 1208.76it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.output.LayerNorm.bias]
Loading weights:   7%|▋         | 58/788 [00:00<00:00, 1226.89it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.output.LayerNorm.weight]
Loading weights:   7%|▋         | 58/788 [00:00<00:00, 1225.00it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.output.LayerNorm.weight]
Loading weights:   7%|▋         | 59/788 [00:00<00:00, 1228.68it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.output.dense.bias]      
Loading weights:   7%|▋         | 59/788 [00:00<00:00, 1226.67it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.output.dense.bias]
Loading weights:   8%|▊         | 60/788 [00:00<00:00, 1240.50it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.output.dense.weight]
Loading weights:   8%|▊         | 60/788 [00:00<00:00, 1233.30it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.output.dense.weight]
Loading weights:   8%|▊         | 61/788 [00:00<00:00, 1232.13it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.self.key.bias]      
Loading weights:   8%|▊         | 61/788 [00:00<00:00, 1230.61it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.self.key.bias]
Loading weights:   8%|▊         | 62/788 [00:00<00:00, 1236.96it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.self.key.weight]
Loading weights:   8%|▊         | 62/788 [00:00<00:00, 1220.26it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.self.key.weight]
Loading weights:   8%|▊         | 63/788 [00:00<00:00, 1229.73it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.self.query.bias]
Loading weights:   8%|▊         | 63/788 [00:00<00:00, 1224.35it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.self.query.bias]
Loading weights:   8%|▊         | 64/788 [00:00<00:00, 1228.39it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.self.query.weight]
Loading weights:   8%|▊         | 64/788 [00:00<00:00, 1216.25it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.self.query.weight]
Loading weights:   8%|▊         | 65/788 [00:00<00:00, 1221.65it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.self.value.bias]  
Loading weights:   8%|▊         | 65/788 [00:00<00:00, 1213.41it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.self.value.bias]
Loading weights:   8%|▊         | 66/788 [00:00<00:00, 1224.72it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.self.value.weight]
Loading weights:   8%|▊         | 66/788 [00:00<00:00, 1214.58it/s, Materializing param=text_decoder.bert.encoder.layer.2.attention.self.value.weight]
Loading weights:   9%|▊         | 67/788 [00:00<00:00, 1209.67it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.output.LayerNorm.bias]
Loading weights:   9%|▊         | 67/788 [00:00<00:00, 1207.78it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.output.LayerNorm.bias]
Loading weights:   9%|▊         | 68/788 [00:00<00:00, 1211.33it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.output.LayerNorm.weight]
Loading weights:   9%|▊         | 68/788 [00:00<00:00, 1210.35it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.output.LayerNorm.weight]
Loading weights:   9%|▉         | 69/788 [00:00<00:00, 1211.04it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.output.dense.bias]      
Loading weights:   9%|▉         | 69/788 [00:00<00:00, 1205.92it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.output.dense.bias]
Loading weights:   9%|▉         | 70/788 [00:00<00:00, 1202.32it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.output.dense.weight]
Loading weights:   9%|▉         | 70/788 [00:00<00:00, 1196.50it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.output.dense.weight]
Loading weights:   9%|▉         | 71/788 [00:00<00:00, 1199.05it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.self.key.bias]      
Loading weights:   9%|▉         | 71/788 [00:00<00:00, 1190.40it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.self.key.bias]
Loading weights:   9%|▉         | 72/788 [00:00<00:00, 1199.01it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.self.key.weight]
Loading weights:   9%|▉         | 72/788 [00:00<00:00, 1188.38it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.self.key.weight]
Loading weights:   9%|▉         | 73/788 [00:00<00:00, 1195.37it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.self.query.bias]
Loading weights:   9%|▉         | 73/788 [00:00<00:00, 1189.08it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.self.query.bias]
Loading weights:   9%|▉         | 74/788 [00:00<00:00, 1194.07it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.self.query.weight]
Loading weights:   9%|▉         | 74/788 [00:00<00:00, 1188.21it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.self.query.weight]
Loading weights:  10%|▉         | 75/788 [00:00<00:00, 1195.87it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.self.value.bias]  
Loading weights:  10%|▉         | 75/788 [00:00<00:00, 1193.68it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.self.value.bias]
Loading weights:  10%|▉         | 76/788 [00:00<00:00, 1189.68it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.self.value.weight]
Loading weights:  10%|▉         | 76/788 [00:00<00:00, 1183.71it/s, Materializing param=text_decoder.bert.encoder.layer.2.crossattention.self.value.weight]
Loading weights:  10%|▉         | 77/788 [00:00<00:00, 1192.56it/s, Materializing param=text_decoder.bert.encoder.layer.2.intermediate.dense.bias]         
Loading weights:  10%|▉         | 77/788 [00:00<00:00, 1191.31it/s, Materializing param=text_decoder.bert.encoder.layer.2.intermediate.dense.bias]
Loading weights:  10%|▉         | 78/788 [00:00<00:00, 1201.35it/s, Materializing param=text_decoder.bert.encoder.layer.2.intermediate.dense.weight]
Loading weights:  10%|▉         | 78/788 [00:00<00:00, 1182.40it/s, Materializing param=text_decoder.bert.encoder.layer.2.intermediate.dense.weight]
Loading weights:  10%|█         | 79/788 [00:00<00:00, 1186.13it/s, Materializing param=text_decoder.bert.encoder.layer.2.output.LayerNorm.bias]    
Loading weights:  10%|█         | 79/788 [00:00<00:00, 1180.09it/s, Materializing param=text_decoder.bert.encoder.layer.2.output.LayerNorm.bias]
Loading weights:  10%|█         | 80/788 [00:00<00:00, 1192.25it/s, Materializing param=text_decoder.bert.encoder.layer.2.output.LayerNorm.weight]
Loading weights:  10%|█         | 80/788 [00:00<00:00, 1176.11it/s, Materializing param=text_decoder.bert.encoder.layer.2.output.LayerNorm.weight]
Loading weights:  10%|█         | 81/788 [00:00<00:00, 1179.49it/s, Materializing param=text_decoder.bert.encoder.layer.2.output.dense.bias]      
Loading weights:  10%|█         | 81/788 [00:00<00:00, 1178.19it/s, Materializing param=text_decoder.bert.encoder.layer.2.output.dense.bias]
Loading weights:  10%|█         | 82/788 [00:00<00:00, 1175.78it/s, Materializing param=text_decoder.bert.encoder.layer.2.output.dense.weight]
Loading weights:  10%|█         | 82/788 [00:00<00:00, 1169.53it/s, Materializing param=text_decoder.bert.encoder.layer.2.output.dense.weight]
Loading weights:  11%|█         | 83/788 [00:00<00:00, 1176.64it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.output.LayerNorm.bias]
Loading weights:  11%|█         | 83/788 [00:00<00:00, 1171.39it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.output.LayerNorm.bias]
Loading weights:  11%|█         | 84/788 [00:00<00:00, 1180.34it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.output.LayerNorm.weight]
Loading weights:  11%|█         | 84/788 [00:00<00:00, 1176.67it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.output.LayerNorm.weight]
Loading weights:  11%|█         | 85/788 [00:00<00:00, 1188.33it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.output.dense.bias]      
Loading weights:  11%|█         | 85/788 [00:00<00:00, 1174.75it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.output.dense.bias]
Loading weights:  11%|█         | 86/788 [00:00<00:00, 1176.91it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.output.dense.weight]
Loading weights:  11%|█         | 86/788 [00:00<00:00, 1175.97it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.output.dense.weight]
Loading weights:  11%|█         | 87/788 [00:00<00:00, 1183.31it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.self.key.bias]      
Loading weights:  11%|█         | 87/788 [00:00<00:00, 1182.57it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.self.key.bias]
Loading weights:  11%|█         | 88/788 [00:00<00:00, 1179.21it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.self.key.weight]
Loading weights:  11%|█         | 88/788 [00:00<00:00, 1178.60it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.self.key.weight]
Loading weights:  11%|█▏        | 89/788 [00:00<00:00, 1185.06it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.self.query.bias]
Loading weights:  11%|█▏        | 89/788 [00:00<00:00, 1181.50it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.self.query.bias]
Loading weights:  11%|█▏        | 90/788 [00:00<00:00, 1188.91it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.self.query.weight]
Loading weights:  11%|█▏        | 90/788 [00:00<00:00, 1183.51it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.self.query.weight]
Loading weights:  12%|█▏        | 91/788 [00:00<00:00, 1192.08it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.self.value.bias]  
Loading weights:  12%|█▏        | 91/788 [00:00<00:00, 1186.63it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.self.value.bias]
Loading weights:  12%|█▏        | 92/788 [00:00<00:00, 1192.76it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.self.value.weight]
Loading weights:  12%|█▏        | 92/788 [00:00<00:00, 1192.11it/s, Materializing param=text_decoder.bert.encoder.layer.3.attention.self.value.weight]
Loading weights:  12%|█▏        | 93/788 [00:00<00:00, 1202.75it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.output.LayerNorm.bias]
Loading weights:  12%|█▏        | 93/788 [00:00<00:00, 1202.13it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.output.LayerNorm.bias]
Loading weights:  12%|█▏        | 94/788 [00:00<00:00, 1207.15it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.output.LayerNorm.weight]
Loading weights:  12%|█▏        | 94/788 [00:00<00:00, 1201.45it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.output.LayerNorm.weight]
Loading weights:  12%|█▏        | 95/788 [00:00<00:00, 1203.17it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.output.dense.bias]      
Loading weights:  12%|█▏        | 95/788 [00:00<00:00, 1193.89it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.output.dense.bias]
Loading weights:  12%|█▏        | 96/788 [00:00<00:00, 1196.58it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.output.dense.weight]
Loading weights:  12%|█▏        | 96/788 [00:00<00:00, 1195.12it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.output.dense.weight]
Loading weights:  12%|█▏        | 97/788 [00:00<00:00, 1192.95it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.self.key.bias]      
Loading weights:  12%|█▏        | 97/788 [00:00<00:00, 1192.43it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.self.key.bias]
Loading weights:  12%|█▏        | 98/788 [00:00<00:00, 1196.63it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.self.key.weight]
Loading weights:  12%|█▏        | 98/788 [00:00<00:00, 1176.80it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.self.key.weight]
Loading weights:  13%|█▎        | 99/788 [00:00<00:00, 1174.31it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.self.query.bias]
Loading weights:  13%|█▎        | 99/788 [00:00<00:00, 1173.73it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.self.query.bias]
Loading weights:  13%|█▎        | 100/788 [00:00<00:00, 1177.98it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.self.query.weight]
Loading weights:  13%|█▎        | 100/788 [00:00<00:00, 1168.70it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.self.query.weight]
Loading weights:  13%|█▎        | 101/788 [00:00<00:00, 1174.61it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.self.value.bias]  
Loading weights:  13%|█▎        | 101/788 [00:00<00:00, 1172.83it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.self.value.bias]
Loading weights:  13%|█▎        | 102/788 [00:00<00:00, 1182.31it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.self.value.weight]
Loading weights:  13%|█▎        | 102/788 [00:00<00:00, 1176.56it/s, Materializing param=text_decoder.bert.encoder.layer.3.crossattention.self.value.weight]
Loading weights:  13%|█▎        | 103/788 [00:00<00:00, 1181.37it/s, Materializing param=text_decoder.bert.encoder.layer.3.intermediate.dense.bias]         
Loading weights:  13%|█▎        | 103/788 [00:00<00:00, 1171.62it/s, Materializing param=text_decoder.bert.encoder.layer.3.intermediate.dense.bias]
Loading weights:  13%|█▎        | 104/788 [00:00<00:00, 1179.97it/s, Materializing param=text_decoder.bert.encoder.layer.3.intermediate.dense.weight]
Loading weights:  13%|█▎        | 104/788 [00:00<00:00, 1178.67it/s, Materializing param=text_decoder.bert.encoder.layer.3.intermediate.dense.weight]
Loading weights:  13%|█▎        | 105/788 [00:00<00:00, 1183.89it/s, Materializing param=text_decoder.bert.encoder.layer.3.output.LayerNorm.bias]    
Loading weights:  13%|█▎        | 105/788 [00:00<00:00, 1182.61it/s, Materializing param=text_decoder.bert.encoder.layer.3.output.LayerNorm.bias]
Loading weights:  13%|█▎        | 106/788 [00:00<00:00, 1179.75it/s, Materializing param=text_decoder.bert.encoder.layer.3.output.LayerNorm.weight]
Loading weights:  13%|█▎        | 106/788 [00:00<00:00, 1163.98it/s, Materializing param=text_decoder.bert.encoder.layer.3.output.LayerNorm.weight]
Loading weights:  14%|█▎        | 107/788 [00:00<00:00, 1172.94it/s, Materializing param=text_decoder.bert.encoder.layer.3.output.dense.bias]      
Loading weights:  14%|█▎        | 107/788 [00:00<00:00, 1163.82it/s, Materializing param=text_decoder.bert.encoder.layer.3.output.dense.bias]
Loading weights:  14%|█▎        | 108/788 [00:00<00:00, 1173.13it/s, Materializing param=text_decoder.bert.encoder.layer.3.output.dense.weight]
Loading weights:  14%|█▎        | 108/788 [00:00<00:00, 1163.88it/s, Materializing param=text_decoder.bert.encoder.layer.3.output.dense.weight]
Loading weights:  14%|█▍        | 109/788 [00:00<00:00, 1171.97it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.output.LayerNorm.bias]
Loading weights:  14%|█▍        | 109/788 [00:00<00:00, 1167.96it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.output.LayerNorm.bias]
Loading weights:  14%|█▍        | 110/788 [00:00<00:00, 1169.98it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.output.LayerNorm.weight]
Loading weights:  14%|█▍        | 110/788 [00:00<00:00, 1169.35it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.output.LayerNorm.weight]
Loading weights:  14%|█▍        | 111/788 [00:00<00:00, 1174.89it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.output.dense.bias]      
Loading weights:  14%|█▍        | 111/788 [00:00<00:00, 1170.65it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.output.dense.bias]
Loading weights:  14%|█▍        | 112/788 [00:00<00:00, 1169.40it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.output.dense.weight]
Loading weights:  14%|█▍        | 112/788 [00:00<00:00, 1168.58it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.output.dense.weight]
Loading weights:  14%|█▍        | 113/788 [00:00<00:00, 1177.10it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.self.key.bias]      
Loading weights:  14%|█▍        | 113/788 [00:00<00:00, 1176.30it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.self.key.bias]
Loading weights:  14%|█▍        | 114/788 [00:00<00:00, 1176.73it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.self.key.weight]
Loading weights:  14%|█▍        | 114/788 [00:00<00:00, 1173.04it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.self.key.weight]
Loading weights:  15%|█▍        | 115/788 [00:00<00:00, 1172.56it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.self.query.bias]
Loading weights:  15%|█▍        | 115/788 [00:00<00:00, 1171.59it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.self.query.bias]
Loading weights:  15%|█▍        | 116/788 [00:00<00:00, 1169.41it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.self.query.weight]
Loading weights:  15%|█▍        | 116/788 [00:00<00:00, 1168.39it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.self.query.weight]
Loading weights:  15%|█▍        | 117/788 [00:00<00:00, 1171.51it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.self.value.bias]  
Loading weights:  15%|█▍        | 117/788 [00:00<00:00, 1167.76it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.self.value.bias]
Loading weights:  15%|█▍        | 118/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.self.value.bias]
Loading weights:  15%|█▍        | 118/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.self.value.weight]
Loading weights:  15%|█▍        | 118/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.4.attention.self.value.weight]
Loading weights:  15%|█▌        | 119/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.output.LayerNorm.bias]
Loading weights:  15%|█▌        | 119/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.output.LayerNorm.bias]
Loading weights:  15%|█▌        | 120/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.output.LayerNorm.weight]
Loading weights:  15%|█▌        | 120/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.output.LayerNorm.weight]
Loading weights:  15%|█▌        | 121/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.output.dense.bias]      
Loading weights:  15%|█▌        | 121/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.output.dense.bias]
Loading weights:  15%|█▌        | 122/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.output.dense.weight]
Loading weights:  15%|█▌        | 122/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.output.dense.weight]
Loading weights:  16%|█▌        | 123/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.self.key.bias]      
Loading weights:  16%|█▌        | 123/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.self.key.bias]
Loading weights:  16%|█▌        | 124/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.self.key.weight]
Loading weights:  16%|█▌        | 124/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.self.key.weight]
Loading weights:  16%|█▌        | 125/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.self.query.bias]
Loading weights:  16%|█▌        | 125/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.self.query.bias]
Loading weights:  16%|█▌        | 126/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.self.query.weight]
Loading weights:  16%|█▌        | 126/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.self.query.weight]
Loading weights:  16%|█▌        | 127/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.self.value.bias]  
Loading weights:  16%|█▌        | 127/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.self.value.bias]
Loading weights:  16%|█▌        | 128/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.self.value.weight]
Loading weights:  16%|█▌        | 128/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.4.crossattention.self.value.weight]
Loading weights:  16%|█▋        | 129/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.4.intermediate.dense.bias]         
Loading weights:  16%|█▋        | 129/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.4.intermediate.dense.bias]
Loading weights:  16%|█▋        | 130/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.4.intermediate.dense.weight]
Loading weights:  16%|█▋        | 130/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.4.intermediate.dense.weight]
Loading weights:  17%|█▋        | 131/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.4.output.LayerNorm.bias]    
Loading weights:  17%|█▋        | 131/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.4.output.LayerNorm.bias]
Loading weights:  17%|█▋        | 132/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.4.output.LayerNorm.weight]
Loading weights:  17%|█▋        | 132/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.4.output.LayerNorm.weight]
Loading weights:  17%|█▋        | 133/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.4.output.dense.bias]      
Loading weights:  17%|█▋        | 133/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.4.output.dense.bias]
Loading weights:  17%|█▋        | 134/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.4.output.dense.weight]
Loading weights:  17%|█▋        | 134/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.4.output.dense.weight]
Loading weights:  17%|█▋        | 135/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.output.LayerNorm.bias]
Loading weights:  17%|█▋        | 135/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.output.LayerNorm.bias]
Loading weights:  17%|█▋        | 136/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.output.LayerNorm.weight]
Loading weights:  17%|█▋        | 136/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.output.LayerNorm.weight]
Loading weights:  17%|█▋        | 137/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.output.dense.bias]      
Loading weights:  17%|█▋        | 137/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.output.dense.bias]
Loading weights:  18%|█▊        | 138/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.output.dense.weight]
Loading weights:  18%|█▊        | 138/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.output.dense.weight]
Loading weights:  18%|█▊        | 139/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.self.key.bias]      
Loading weights:  18%|█▊        | 139/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.self.key.bias]
Loading weights:  18%|█▊        | 140/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.self.key.weight]
Loading weights:  18%|█▊        | 140/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.self.key.weight]
Loading weights:  18%|█▊        | 141/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.self.query.bias]
Loading weights:  18%|█▊        | 141/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.self.query.bias]
Loading weights:  18%|█▊        | 142/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.self.query.weight]
Loading weights:  18%|█▊        | 142/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.self.query.weight]
Loading weights:  18%|█▊        | 143/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.self.value.bias]  
Loading weights:  18%|█▊        | 143/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.self.value.bias]
Loading weights:  18%|█▊        | 144/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.self.value.weight]
Loading weights:  18%|█▊        | 144/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.attention.self.value.weight]
Loading weights:  18%|█▊        | 145/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.output.LayerNorm.bias]
Loading weights:  18%|█▊        | 145/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.output.LayerNorm.bias]
Loading weights:  19%|█▊        | 146/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.output.LayerNorm.weight]
Loading weights:  19%|█▊        | 146/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.output.LayerNorm.weight]
Loading weights:  19%|█▊        | 147/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.output.dense.bias]      
Loading weights:  19%|█▊        | 147/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.output.dense.bias]
Loading weights:  19%|█▉        | 148/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.output.dense.weight]
Loading weights:  19%|█▉        | 148/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.output.dense.weight]
Loading weights:  19%|█▉        | 149/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.self.key.bias]      
Loading weights:  19%|█▉        | 149/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.self.key.bias]
Loading weights:  19%|█▉        | 150/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.self.key.weight]
Loading weights:  19%|█▉        | 150/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.self.key.weight]
Loading weights:  19%|█▉        | 151/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.self.query.bias]
Loading weights:  19%|█▉        | 151/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.self.query.bias]
Loading weights:  19%|█▉        | 152/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.self.query.weight]
Loading weights:  19%|█▉        | 152/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.self.query.weight]
Loading weights:  19%|█▉        | 153/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.self.value.bias]  
Loading weights:  19%|█▉        | 153/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.self.value.bias]
Loading weights:  20%|█▉        | 154/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.self.value.weight]
Loading weights:  20%|█▉        | 154/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.crossattention.self.value.weight]
Loading weights:  20%|█▉        | 155/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.intermediate.dense.bias]         
Loading weights:  20%|█▉        | 155/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.intermediate.dense.bias]
Loading weights:  20%|█▉        | 156/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.intermediate.dense.weight]
Loading weights:  20%|█▉        | 156/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.intermediate.dense.weight]
Loading weights:  20%|█▉        | 157/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.output.LayerNorm.bias]    
Loading weights:  20%|█▉        | 157/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.output.LayerNorm.bias]
Loading weights:  20%|██        | 158/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.output.LayerNorm.weight]
Loading weights:  20%|██        | 158/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.output.LayerNorm.weight]
Loading weights:  20%|██        | 159/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.output.dense.bias]      
Loading weights:  20%|██        | 159/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.output.dense.bias]
Loading weights:  20%|██        | 160/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.output.dense.weight]
Loading weights:  20%|██        | 160/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.5.output.dense.weight]
Loading weights:  20%|██        | 161/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.output.LayerNorm.bias]
Loading weights:  20%|██        | 161/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.output.LayerNorm.bias]
Loading weights:  21%|██        | 162/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.output.LayerNorm.weight]
Loading weights:  21%|██        | 162/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.output.LayerNorm.weight]
Loading weights:  21%|██        | 163/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.output.dense.bias]      
Loading weights:  21%|██        | 163/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.output.dense.bias]
Loading weights:  21%|██        | 164/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.output.dense.weight]
Loading weights:  21%|██        | 164/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.output.dense.weight]
Loading weights:  21%|██        | 165/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.self.key.bias]      
Loading weights:  21%|██        | 165/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.self.key.bias]
Loading weights:  21%|██        | 166/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.self.key.weight]
Loading weights:  21%|██        | 166/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.self.key.weight]
Loading weights:  21%|██        | 167/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.self.query.bias]
Loading weights:  21%|██        | 167/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.self.query.bias]
Loading weights:  21%|██▏       | 168/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.self.query.weight]
Loading weights:  21%|██▏       | 168/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.self.query.weight]
Loading weights:  21%|██▏       | 169/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.self.value.bias]  
Loading weights:  21%|██▏       | 169/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.self.value.bias]
Loading weights:  22%|██▏       | 170/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.self.value.weight]
Loading weights:  22%|██▏       | 170/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.attention.self.value.weight]
Loading weights:  22%|██▏       | 171/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.output.LayerNorm.bias]
Loading weights:  22%|██▏       | 171/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.output.LayerNorm.bias]
Loading weights:  22%|██▏       | 172/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.output.LayerNorm.weight]
Loading weights:  22%|██▏       | 172/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.output.LayerNorm.weight]
Loading weights:  22%|██▏       | 173/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.output.dense.bias]      
Loading weights:  22%|██▏       | 173/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.output.dense.bias]
Loading weights:  22%|██▏       | 174/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.output.dense.weight]
Loading weights:  22%|██▏       | 174/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.output.dense.weight]
Loading weights:  22%|██▏       | 175/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.self.key.bias]      
Loading weights:  22%|██▏       | 175/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.self.key.bias]
Loading weights:  22%|██▏       | 176/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.self.key.weight]
Loading weights:  22%|██▏       | 176/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.self.key.weight]
Loading weights:  22%|██▏       | 177/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.self.query.bias]
Loading weights:  22%|██▏       | 177/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.self.query.bias]
Loading weights:  23%|██▎       | 178/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.self.query.weight]
Loading weights:  23%|██▎       | 178/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.self.query.weight]
Loading weights:  23%|██▎       | 179/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.self.value.bias]  
Loading weights:  23%|██▎       | 179/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.self.value.bias]
Loading weights:  23%|██▎       | 180/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.self.value.weight]
Loading weights:  23%|██▎       | 180/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.crossattention.self.value.weight]
Loading weights:  23%|██▎       | 181/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.intermediate.dense.bias]         
Loading weights:  23%|██▎       | 181/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.intermediate.dense.bias]
Loading weights:  23%|██▎       | 182/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.intermediate.dense.weight]
Loading weights:  23%|██▎       | 182/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.intermediate.dense.weight]
Loading weights:  23%|██▎       | 183/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.output.LayerNorm.bias]    
Loading weights:  23%|██▎       | 183/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.output.LayerNorm.bias]
Loading weights:  23%|██▎       | 184/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.output.LayerNorm.weight]
Loading weights:  23%|██▎       | 184/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.output.LayerNorm.weight]
Loading weights:  23%|██▎       | 185/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.output.dense.bias]      
Loading weights:  23%|██▎       | 185/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.output.dense.bias]
Loading weights:  24%|██▎       | 186/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.output.dense.weight]
Loading weights:  24%|██▎       | 186/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.6.output.dense.weight]
Loading weights:  24%|██▎       | 187/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.output.LayerNorm.bias]
Loading weights:  24%|██▎       | 187/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.output.LayerNorm.bias]
Loading weights:  24%|██▍       | 188/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.output.LayerNorm.weight]
Loading weights:  24%|██▍       | 188/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.output.LayerNorm.weight]
Loading weights:  24%|██▍       | 189/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.output.dense.bias]      
Loading weights:  24%|██▍       | 189/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.output.dense.bias]
Loading weights:  24%|██▍       | 190/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.output.dense.weight]
Loading weights:  24%|██▍       | 190/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.output.dense.weight]
Loading weights:  24%|██▍       | 191/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.self.key.bias]      
Loading weights:  24%|██▍       | 191/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.self.key.bias]
Loading weights:  24%|██▍       | 192/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.self.key.weight]
Loading weights:  24%|██▍       | 192/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.self.key.weight]
Loading weights:  24%|██▍       | 193/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.self.query.bias]
Loading weights:  24%|██▍       | 193/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.self.query.bias]
Loading weights:  25%|██▍       | 194/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.self.query.weight]
Loading weights:  25%|██▍       | 194/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.self.query.weight]
Loading weights:  25%|██▍       | 195/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.self.value.bias]  
Loading weights:  25%|██▍       | 195/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.self.value.bias]
Loading weights:  25%|██▍       | 196/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.self.value.weight]
Loading weights:  25%|██▍       | 196/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.attention.self.value.weight]
Loading weights:  25%|██▌       | 197/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.output.LayerNorm.bias]
Loading weights:  25%|██▌       | 197/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.output.LayerNorm.bias]
Loading weights:  25%|██▌       | 198/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.output.LayerNorm.weight]
Loading weights:  25%|██▌       | 198/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.output.LayerNorm.weight]
Loading weights:  25%|██▌       | 199/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.output.dense.bias]      
Loading weights:  25%|██▌       | 199/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.output.dense.bias]
Loading weights:  25%|██▌       | 200/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.output.dense.weight]
Loading weights:  25%|██▌       | 200/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.output.dense.weight]
Loading weights:  26%|██▌       | 201/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.self.key.bias]      
Loading weights:  26%|██▌       | 201/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.self.key.bias]
Loading weights:  26%|██▌       | 202/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.self.key.weight]
Loading weights:  26%|██▌       | 202/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.self.key.weight]
Loading weights:  26%|██▌       | 203/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.self.query.bias]
Loading weights:  26%|██▌       | 203/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.self.query.bias]
Loading weights:  26%|██▌       | 204/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.self.query.weight]
Loading weights:  26%|██▌       | 204/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.self.query.weight]
Loading weights:  26%|██▌       | 205/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.self.value.bias]  
Loading weights:  26%|██▌       | 205/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.self.value.bias]
Loading weights:  26%|██▌       | 206/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.self.value.weight]
Loading weights:  26%|██▌       | 206/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.crossattention.self.value.weight]
Loading weights:  26%|██▋       | 207/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.intermediate.dense.bias]         
Loading weights:  26%|██▋       | 207/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.intermediate.dense.bias]
Loading weights:  26%|██▋       | 208/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.intermediate.dense.weight]
Loading weights:  26%|██▋       | 208/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.intermediate.dense.weight]
Loading weights:  27%|██▋       | 209/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.output.LayerNorm.bias]    
Loading weights:  27%|██▋       | 209/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.output.LayerNorm.bias]
Loading weights:  27%|██▋       | 210/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.output.LayerNorm.weight]
Loading weights:  27%|██▋       | 210/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.output.LayerNorm.weight]
Loading weights:  27%|██▋       | 211/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.output.dense.bias]      
Loading weights:  27%|██▋       | 211/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.output.dense.bias]
Loading weights:  27%|██▋       | 212/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.output.dense.weight]
Loading weights:  27%|██▋       | 212/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.7.output.dense.weight]
Loading weights:  27%|██▋       | 213/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.output.LayerNorm.bias]
Loading weights:  27%|██▋       | 213/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.output.LayerNorm.bias]
Loading weights:  27%|██▋       | 214/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.output.LayerNorm.weight]
Loading weights:  27%|██▋       | 214/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.output.LayerNorm.weight]
Loading weights:  27%|██▋       | 215/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.output.dense.bias]      
Loading weights:  27%|██▋       | 215/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.output.dense.bias]
Loading weights:  27%|██▋       | 216/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.output.dense.weight]
Loading weights:  27%|██▋       | 216/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.output.dense.weight]
Loading weights:  28%|██▊       | 217/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.self.key.bias]      
Loading weights:  28%|██▊       | 217/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.self.key.bias]
Loading weights:  28%|██▊       | 218/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.self.key.weight]
Loading weights:  28%|██▊       | 218/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.self.key.weight]
Loading weights:  28%|██▊       | 219/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.self.query.bias]
Loading weights:  28%|██▊       | 219/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.self.query.bias]
Loading weights:  28%|██▊       | 220/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.self.query.weight]
Loading weights:  28%|██▊       | 220/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.self.query.weight]
Loading weights:  28%|██▊       | 221/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.self.value.bias]  
Loading weights:  28%|██▊       | 221/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.self.value.bias]
Loading weights:  28%|██▊       | 222/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.self.value.weight]
Loading weights:  28%|██▊       | 222/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.attention.self.value.weight]
Loading weights:  28%|██▊       | 223/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.output.LayerNorm.bias]
Loading weights:  28%|██▊       | 223/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.output.LayerNorm.bias]
Loading weights:  28%|██▊       | 224/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.output.LayerNorm.weight]
Loading weights:  28%|██▊       | 224/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.output.LayerNorm.weight]
Loading weights:  29%|██▊       | 225/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.output.dense.bias]      
Loading weights:  29%|██▊       | 225/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.output.dense.bias]
Loading weights:  29%|██▊       | 226/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.output.dense.weight]
Loading weights:  29%|██▊       | 226/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.output.dense.weight]
Loading weights:  29%|██▉       | 227/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.self.key.bias]      
Loading weights:  29%|██▉       | 227/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.self.key.bias]
Loading weights:  29%|██▉       | 228/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.self.key.weight]
Loading weights:  29%|██▉       | 228/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.self.key.weight]
Loading weights:  29%|██▉       | 229/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.self.query.bias]
Loading weights:  29%|██▉       | 229/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.self.query.bias]
Loading weights:  29%|██▉       | 230/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.self.query.weight]
Loading weights:  29%|██▉       | 230/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.self.query.weight]
Loading weights:  29%|██▉       | 231/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.self.value.bias]  
Loading weights:  29%|██▉       | 231/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.self.value.bias]
Loading weights:  29%|██▉       | 232/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.self.value.weight]
Loading weights:  29%|██▉       | 232/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.crossattention.self.value.weight]
Loading weights:  30%|██▉       | 233/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.intermediate.dense.bias]         
Loading weights:  30%|██▉       | 233/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.intermediate.dense.bias]
Loading weights:  30%|██▉       | 234/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.intermediate.dense.weight]
Loading weights:  30%|██▉       | 234/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.intermediate.dense.weight]
Loading weights:  30%|██▉       | 235/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.output.LayerNorm.bias]    
Loading weights:  30%|██▉       | 235/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.output.LayerNorm.bias]
Loading weights:  30%|██▉       | 236/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.output.LayerNorm.weight]
Loading weights:  30%|██▉       | 236/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.output.LayerNorm.weight]
Loading weights:  30%|███       | 237/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.output.dense.bias]      
Loading weights:  30%|███       | 237/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.output.dense.bias]
Loading weights:  30%|███       | 238/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.output.dense.weight]
Loading weights:  30%|███       | 238/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.8.output.dense.weight]
Loading weights:  30%|███       | 239/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.output.LayerNorm.bias]
Loading weights:  30%|███       | 239/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.output.LayerNorm.bias]
Loading weights:  30%|███       | 240/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.output.LayerNorm.weight]
Loading weights:  30%|███       | 240/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.output.LayerNorm.weight]
Loading weights:  31%|███       | 241/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.output.dense.bias]      
Loading weights:  31%|███       | 241/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.output.dense.bias]
Loading weights:  31%|███       | 242/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.output.dense.weight]
Loading weights:  31%|███       | 242/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.output.dense.weight]
Loading weights:  31%|███       | 243/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.self.key.bias]      
Loading weights:  31%|███       | 243/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.self.key.bias]
Loading weights:  31%|███       | 244/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.self.key.weight]
Loading weights:  31%|███       | 244/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.self.key.weight]
Loading weights:  31%|███       | 245/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.self.query.bias]
Loading weights:  31%|███       | 245/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.self.query.bias]
Loading weights:  31%|███       | 246/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.self.query.weight]
Loading weights:  31%|███       | 246/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.self.query.weight]
Loading weights:  31%|███▏      | 247/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.self.value.bias]  
Loading weights:  31%|███▏      | 247/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.self.value.bias]
Loading weights:  31%|███▏      | 248/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.self.value.weight]
Loading weights:  31%|███▏      | 248/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.attention.self.value.weight]
Loading weights:  32%|███▏      | 249/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.output.LayerNorm.bias]
Loading weights:  32%|███▏      | 249/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.output.LayerNorm.bias]
Loading weights:  32%|███▏      | 250/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.output.LayerNorm.weight]
Loading weights:  32%|███▏      | 250/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.output.LayerNorm.weight]
Loading weights:  32%|███▏      | 251/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.output.dense.bias]      
Loading weights:  32%|███▏      | 251/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.output.dense.bias]
Loading weights:  32%|███▏      | 252/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.output.dense.weight]
Loading weights:  32%|███▏      | 252/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.output.dense.weight]
Loading weights:  32%|███▏      | 253/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.self.key.bias]      
Loading weights:  32%|███▏      | 253/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.self.key.bias]
Loading weights:  32%|███▏      | 254/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.self.key.weight]
Loading weights:  32%|███▏      | 254/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.self.key.weight]
Loading weights:  32%|███▏      | 255/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.self.query.bias]
Loading weights:  32%|███▏      | 255/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.self.query.bias]
Loading weights:  32%|███▏      | 256/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.self.query.weight]
Loading weights:  32%|███▏      | 256/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.self.query.weight]
Loading weights:  33%|███▎      | 257/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.self.value.bias]  
Loading weights:  33%|███▎      | 257/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.self.value.bias]
Loading weights:  33%|███▎      | 258/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.self.value.weight]
Loading weights:  33%|███▎      | 258/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.crossattention.self.value.weight]
Loading weights:  33%|███▎      | 259/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.intermediate.dense.bias]         
Loading weights:  33%|███▎      | 259/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.intermediate.dense.bias]
Loading weights:  33%|███▎      | 260/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.intermediate.dense.weight]
Loading weights:  33%|███▎      | 260/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.intermediate.dense.weight]
Loading weights:  33%|███▎      | 261/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.output.LayerNorm.bias]    
Loading weights:  33%|███▎      | 261/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.output.LayerNorm.bias]
Loading weights:  33%|███▎      | 262/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.output.LayerNorm.weight]
Loading weights:  33%|███▎      | 262/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.output.LayerNorm.weight]
Loading weights:  33%|███▎      | 263/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.output.dense.bias]      
Loading weights:  33%|███▎      | 263/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.output.dense.bias]
Loading weights:  34%|███▎      | 264/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.output.dense.weight]
Loading weights:  34%|███▎      | 264/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.9.output.dense.weight]
Loading weights:  34%|███▎      | 265/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.output.LayerNorm.bias]
Loading weights:  34%|███▎      | 265/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.output.LayerNorm.bias]
Loading weights:  34%|███▍      | 266/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.output.LayerNorm.weight]
Loading weights:  34%|███▍      | 266/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.output.LayerNorm.weight]
Loading weights:  34%|███▍      | 267/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.output.dense.bias]      
Loading weights:  34%|███▍      | 267/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.output.dense.bias]
Loading weights:  34%|███▍      | 268/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.output.dense.weight]
Loading weights:  34%|███▍      | 268/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.output.dense.weight]
Loading weights:  34%|███▍      | 269/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.self.key.bias]      
Loading weights:  34%|███▍      | 269/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.self.key.bias]
Loading weights:  34%|███▍      | 270/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.self.key.weight]
Loading weights:  34%|███▍      | 270/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.self.key.weight]
Loading weights:  34%|███▍      | 271/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.self.query.bias]
Loading weights:  34%|███▍      | 271/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.self.query.bias]
Loading weights:  35%|███▍      | 272/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.self.query.weight]
Loading weights:  35%|███▍      | 272/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.self.query.weight]
Loading weights:  35%|███▍      | 273/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.self.value.bias]  
Loading weights:  35%|███▍      | 273/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.self.value.bias]
Loading weights:  35%|███▍      | 274/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.self.value.weight]
Loading weights:  35%|███▍      | 274/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.attention.self.value.weight]
Loading weights:  35%|███▍      | 275/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.output.LayerNorm.bias]
Loading weights:  35%|███▍      | 275/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.output.LayerNorm.bias]
Loading weights:  35%|███▌      | 276/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.output.LayerNorm.weight]
Loading weights:  35%|███▌      | 276/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.output.LayerNorm.weight]
Loading weights:  35%|███▌      | 277/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.output.dense.bias]      
Loading weights:  35%|███▌      | 277/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.output.dense.bias]
Loading weights:  35%|███▌      | 278/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.output.dense.weight]
Loading weights:  35%|███▌      | 278/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.output.dense.weight]
Loading weights:  35%|███▌      | 279/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.self.key.bias]      
Loading weights:  35%|███▌      | 279/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.self.key.bias]
Loading weights:  36%|███▌      | 280/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.self.key.weight]
Loading weights:  36%|███▌      | 280/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.self.key.weight]
Loading weights:  36%|███▌      | 281/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.self.query.bias]
Loading weights:  36%|███▌      | 281/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.self.query.bias]
Loading weights:  36%|███▌      | 282/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.self.query.weight]
Loading weights:  36%|███▌      | 282/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.self.query.weight]
Loading weights:  36%|███▌      | 283/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.self.value.bias]  
Loading weights:  36%|███▌      | 283/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.self.value.bias]
Loading weights:  36%|███▌      | 284/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.self.value.weight]
Loading weights:  36%|███▌      | 284/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.crossattention.self.value.weight]
Loading weights:  36%|███▌      | 285/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.intermediate.dense.bias]         
Loading weights:  36%|███▌      | 285/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.intermediate.dense.bias]
Loading weights:  36%|███▋      | 286/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.intermediate.dense.weight]
Loading weights:  36%|███▋      | 286/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.intermediate.dense.weight]
Loading weights:  36%|███▋      | 287/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.output.LayerNorm.bias]    
Loading weights:  36%|███▋      | 287/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.output.LayerNorm.bias]
Loading weights:  37%|███▋      | 288/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.output.LayerNorm.weight]
Loading weights:  37%|███▋      | 288/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.output.LayerNorm.weight]
Loading weights:  37%|███▋      | 289/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.output.dense.bias]      
Loading weights:  37%|███▋      | 289/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.output.dense.bias]
Loading weights:  37%|███▋      | 290/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.output.dense.weight]
Loading weights:  37%|███▋      | 290/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.10.output.dense.weight]
Loading weights:  37%|███▋      | 291/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.output.LayerNorm.bias]
Loading weights:  37%|███▋      | 291/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.output.LayerNorm.bias]
Loading weights:  37%|███▋      | 292/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.output.LayerNorm.weight]
Loading weights:  37%|███▋      | 292/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.output.LayerNorm.weight]
Loading weights:  37%|███▋      | 293/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.output.dense.bias]      
Loading weights:  37%|███▋      | 293/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.output.dense.bias]
Loading weights:  37%|███▋      | 294/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.output.dense.weight]
Loading weights:  37%|███▋      | 294/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.output.dense.weight]
Loading weights:  37%|███▋      | 295/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.self.key.bias]      
Loading weights:  37%|███▋      | 295/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.self.key.bias]
Loading weights:  38%|███▊      | 296/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.self.key.weight]
Loading weights:  38%|███▊      | 296/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.self.key.weight]
Loading weights:  38%|███▊      | 297/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.self.query.bias]
Loading weights:  38%|███▊      | 297/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.self.query.bias]
Loading weights:  38%|███▊      | 298/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.self.query.weight]
Loading weights:  38%|███▊      | 298/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.self.query.weight]
Loading weights:  38%|███▊      | 299/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.self.value.bias]  
Loading weights:  38%|███▊      | 299/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.self.value.bias]
Loading weights:  38%|███▊      | 300/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.self.value.weight]
Loading weights:  38%|███▊      | 300/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.attention.self.value.weight]
Loading weights:  38%|███▊      | 301/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.output.LayerNorm.bias]
Loading weights:  38%|███▊      | 301/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.output.LayerNorm.bias]
Loading weights:  38%|███▊      | 302/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.output.LayerNorm.weight]
Loading weights:  38%|███▊      | 302/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.output.LayerNorm.weight]
Loading weights:  38%|███▊      | 303/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.output.dense.bias]      
Loading weights:  38%|███▊      | 303/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.output.dense.bias]
Loading weights:  39%|███▊      | 304/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.output.dense.weight]
Loading weights:  39%|███▊      | 304/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.output.dense.weight]
Loading weights:  39%|███▊      | 305/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.self.key.bias]      
Loading weights:  39%|███▊      | 305/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.self.key.bias]
Loading weights:  39%|███▉      | 306/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.self.key.weight]
Loading weights:  39%|███▉      | 306/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.self.key.weight]
Loading weights:  39%|███▉      | 307/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.self.query.bias]
Loading weights:  39%|███▉      | 307/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.self.query.bias]
Loading weights:  39%|███▉      | 308/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.self.query.weight]
Loading weights:  39%|███▉      | 308/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.self.query.weight]
Loading weights:  39%|███▉      | 309/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.self.value.bias]  
Loading weights:  39%|███▉      | 309/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.self.value.bias]
Loading weights:  39%|███▉      | 310/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.self.value.weight]
Loading weights:  39%|███▉      | 310/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.crossattention.self.value.weight]
Loading weights:  39%|███▉      | 311/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.intermediate.dense.bias]         
Loading weights:  39%|███▉      | 311/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.intermediate.dense.bias]
Loading weights:  40%|███▉      | 312/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.intermediate.dense.weight]
Loading weights:  40%|███▉      | 312/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.intermediate.dense.weight]
Loading weights:  40%|███▉      | 313/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.output.LayerNorm.bias]    
Loading weights:  40%|███▉      | 313/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.output.LayerNorm.bias]
Loading weights:  40%|███▉      | 314/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.output.LayerNorm.weight]
Loading weights:  40%|███▉      | 314/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.output.LayerNorm.weight]
Loading weights:  40%|███▉      | 315/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.output.dense.bias]      
Loading weights:  40%|███▉      | 315/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.output.dense.bias]
Loading weights:  40%|████      | 316/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.output.dense.weight]
Loading weights:  40%|████      | 316/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.bert.encoder.layer.11.output.dense.weight]
Loading weights:  40%|████      | 317/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.cls.predictions.bias]                     
Loading weights:  40%|████      | 317/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.cls.predictions.bias]
Loading weights:  40%|████      | 318/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.cls.predictions.decoder.weight]
Loading weights:  40%|████      | 318/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.cls.predictions.decoder.weight]
Loading weights:  40%|████      | 319/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.cls.predictions.transform.LayerNorm.bias]
Loading weights:  40%|████      | 319/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.cls.predictions.transform.LayerNorm.bias]
Loading weights:  41%|████      | 320/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.cls.predictions.transform.LayerNorm.weight]
Loading weights:  41%|████      | 320/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.cls.predictions.transform.LayerNorm.weight]
Loading weights:  41%|████      | 321/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.cls.predictions.transform.dense.bias]      
Loading weights:  41%|████      | 321/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.cls.predictions.transform.dense.bias]
Loading weights:  41%|████      | 322/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.cls.predictions.transform.dense.weight]
Loading weights:  41%|████      | 322/788 [00:00<00:00, 1169.03it/s, Materializing param=text_decoder.cls.predictions.transform.dense.weight]
Loading weights:  41%|████      | 323/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.embeddings.LayerNorm.bias]             
Loading weights:  41%|████      | 323/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.embeddings.LayerNorm.bias]
Loading weights:  41%|████      | 324/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.embeddings.LayerNorm.weight]
Loading weights:  41%|████      | 324/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.embeddings.LayerNorm.weight]
Loading weights:  41%|████      | 325/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.embeddings.position_embeddings.weight]
Loading weights:  41%|████      | 325/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.embeddings.position_embeddings.weight]
Loading weights:  41%|████▏     | 326/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.embeddings.word_embeddings.weight]    
Loading weights:  41%|████▏     | 326/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.embeddings.word_embeddings.weight]
Loading weights:  41%|████▏     | 327/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.attention.output.LayerNorm.bias]
Loading weights:  41%|████▏     | 327/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.attention.output.LayerNorm.bias]
Loading weights:  42%|████▏     | 328/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.attention.output.LayerNorm.weight]
Loading weights:  42%|████▏     | 328/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.attention.output.LayerNorm.weight]
Loading weights:  42%|████▏     | 329/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.attention.output.dense.bias]      
Loading weights:  42%|████▏     | 329/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.attention.output.dense.bias]
Loading weights:  42%|████▏     | 330/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.attention.output.dense.weight]
Loading weights:  42%|████▏     | 330/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.attention.output.dense.weight]
Loading weights:  42%|████▏     | 331/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.attention.self.key.bias]      
Loading weights:  42%|████▏     | 331/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.attention.self.key.bias]
Loading weights:  42%|████▏     | 332/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.attention.self.key.weight]
Loading weights:  42%|████▏     | 332/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.attention.self.key.weight]
Loading weights:  42%|████▏     | 333/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.attention.self.query.bias]
Loading weights:  42%|████▏     | 333/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.attention.self.query.bias]
Loading weights:  42%|████▏     | 334/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.attention.self.query.weight]
Loading weights:  42%|████▏     | 334/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.attention.self.query.weight]
Loading weights:  43%|████▎     | 335/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.attention.self.value.bias]  
Loading weights:  43%|████▎     | 335/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.attention.self.value.bias]
Loading weights:  43%|████▎     | 336/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.attention.self.value.weight]
Loading weights:  43%|████▎     | 336/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.attention.self.value.weight]
Loading weights:  43%|████▎     | 337/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.crossattention.output.LayerNorm.bias]
Loading weights:  43%|████▎     | 337/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.crossattention.output.LayerNorm.bias]
Loading weights:  43%|████▎     | 338/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.crossattention.output.LayerNorm.weight]
Loading weights:  43%|████▎     | 338/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.crossattention.output.LayerNorm.weight]
Loading weights:  43%|████▎     | 339/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.crossattention.output.dense.bias]      
Loading weights:  43%|████▎     | 339/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.crossattention.output.dense.bias]
Loading weights:  43%|████▎     | 340/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.crossattention.output.dense.weight]
Loading weights:  43%|████▎     | 340/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.crossattention.output.dense.weight]
Loading weights:  43%|████▎     | 341/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.crossattention.self.key.bias]      
Loading weights:  43%|████▎     | 341/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.crossattention.self.key.bias]
Loading weights:  43%|████▎     | 342/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.crossattention.self.key.weight]
Loading weights:  43%|████▎     | 342/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.crossattention.self.key.weight]
Loading weights:  44%|████▎     | 343/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.crossattention.self.query.bias]
Loading weights:  44%|████▎     | 343/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.crossattention.self.query.bias]
Loading weights:  44%|████▎     | 344/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.crossattention.self.query.weight]
Loading weights:  44%|████▎     | 344/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.crossattention.self.query.weight]
Loading weights:  44%|████▍     | 345/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.crossattention.self.value.bias]  
Loading weights:  44%|████▍     | 345/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.crossattention.self.value.bias]
Loading weights:  44%|████▍     | 346/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.crossattention.self.value.weight]
Loading weights:  44%|████▍     | 346/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.crossattention.self.value.weight]
Loading weights:  44%|████▍     | 347/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.intermediate.dense.bias]         
Loading weights:  44%|████▍     | 347/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.intermediate.dense.bias]
Loading weights:  44%|████▍     | 348/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.intermediate.dense.weight]
Loading weights:  44%|████▍     | 348/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.intermediate.dense.weight]
Loading weights:  44%|████▍     | 349/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.output.LayerNorm.bias]    
Loading weights:  44%|████▍     | 349/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.output.LayerNorm.bias]
Loading weights:  44%|████▍     | 350/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.output.LayerNorm.weight]
Loading weights:  44%|████▍     | 350/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.output.LayerNorm.weight]
Loading weights:  45%|████▍     | 351/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.output.dense.bias]      
Loading weights:  45%|████▍     | 351/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.output.dense.bias]
Loading weights:  45%|████▍     | 352/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.output.dense.weight]
Loading weights:  45%|████▍     | 352/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.0.output.dense.weight]
Loading weights:  45%|████▍     | 353/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.attention.output.LayerNorm.bias]
Loading weights:  45%|████▍     | 353/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.attention.output.LayerNorm.bias]
Loading weights:  45%|████▍     | 354/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.attention.output.LayerNorm.weight]
Loading weights:  45%|████▍     | 354/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.attention.output.LayerNorm.weight]
Loading weights:  45%|████▌     | 355/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.attention.output.dense.bias]      
Loading weights:  45%|████▌     | 355/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.attention.output.dense.bias]
Loading weights:  45%|████▌     | 356/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.attention.output.dense.weight]
Loading weights:  45%|████▌     | 356/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.attention.output.dense.weight]
Loading weights:  45%|████▌     | 357/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.attention.self.key.bias]      
Loading weights:  45%|████▌     | 357/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.attention.self.key.bias]
Loading weights:  45%|████▌     | 358/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.attention.self.key.weight]
Loading weights:  45%|████▌     | 358/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.attention.self.key.weight]
Loading weights:  46%|████▌     | 359/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.attention.self.query.bias]
Loading weights:  46%|████▌     | 359/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.attention.self.query.bias]
Loading weights:  46%|████▌     | 360/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.attention.self.query.weight]
Loading weights:  46%|████▌     | 360/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.attention.self.query.weight]
Loading weights:  46%|████▌     | 361/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.attention.self.value.bias]  
Loading weights:  46%|████▌     | 361/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.attention.self.value.bias]
Loading weights:  46%|████▌     | 362/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.attention.self.value.weight]
Loading weights:  46%|████▌     | 362/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.attention.self.value.weight]
Loading weights:  46%|████▌     | 363/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.crossattention.output.LayerNorm.bias]
Loading weights:  46%|████▌     | 363/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.crossattention.output.LayerNorm.bias]
Loading weights:  46%|████▌     | 364/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.crossattention.output.LayerNorm.weight]
Loading weights:  46%|████▌     | 364/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.crossattention.output.LayerNorm.weight]
Loading weights:  46%|████▋     | 365/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.crossattention.output.dense.bias]      
Loading weights:  46%|████▋     | 365/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.crossattention.output.dense.bias]
Loading weights:  46%|████▋     | 366/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.crossattention.output.dense.weight]
Loading weights:  46%|████▋     | 366/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.crossattention.output.dense.weight]
Loading weights:  47%|████▋     | 367/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.crossattention.self.key.bias]      
Loading weights:  47%|████▋     | 367/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.crossattention.self.key.bias]
Loading weights:  47%|████▋     | 368/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.crossattention.self.key.weight]
Loading weights:  47%|████▋     | 368/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.crossattention.self.key.weight]
Loading weights:  47%|████▋     | 369/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.crossattention.self.query.bias]
Loading weights:  47%|████▋     | 369/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.crossattention.self.query.bias]
Loading weights:  47%|████▋     | 370/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.crossattention.self.query.weight]
Loading weights:  47%|████▋     | 370/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.crossattention.self.query.weight]
Loading weights:  47%|████▋     | 371/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.crossattention.self.value.bias]  
Loading weights:  47%|████▋     | 371/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.crossattention.self.value.bias]
Loading weights:  47%|████▋     | 372/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.crossattention.self.value.weight]
Loading weights:  47%|████▋     | 372/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.crossattention.self.value.weight]
Loading weights:  47%|████▋     | 373/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.intermediate.dense.bias]         
Loading weights:  47%|████▋     | 373/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.intermediate.dense.bias]
Loading weights:  47%|████▋     | 374/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.intermediate.dense.weight]
Loading weights:  47%|████▋     | 374/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.intermediate.dense.weight]
Loading weights:  48%|████▊     | 375/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.output.LayerNorm.bias]    
Loading weights:  48%|████▊     | 375/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.output.LayerNorm.bias]
Loading weights:  48%|████▊     | 376/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.output.LayerNorm.weight]
Loading weights:  48%|████▊     | 376/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.output.LayerNorm.weight]
Loading weights:  48%|████▊     | 377/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.output.dense.bias]      
Loading weights:  48%|████▊     | 377/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.output.dense.bias]
Loading weights:  48%|████▊     | 378/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.output.dense.weight]
Loading weights:  48%|████▊     | 378/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.1.output.dense.weight]
Loading weights:  48%|████▊     | 379/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.attention.output.LayerNorm.bias]
Loading weights:  48%|████▊     | 379/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.attention.output.LayerNorm.bias]
Loading weights:  48%|████▊     | 380/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.attention.output.LayerNorm.weight]
Loading weights:  48%|████▊     | 380/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.attention.output.LayerNorm.weight]
Loading weights:  48%|████▊     | 381/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.attention.output.dense.bias]      
Loading weights:  48%|████▊     | 381/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.attention.output.dense.bias]
Loading weights:  48%|████▊     | 382/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.attention.output.dense.weight]
Loading weights:  48%|████▊     | 382/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.attention.output.dense.weight]
Loading weights:  49%|████▊     | 383/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.attention.self.key.bias]      
Loading weights:  49%|████▊     | 383/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.attention.self.key.bias]
Loading weights:  49%|████▊     | 384/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.attention.self.key.weight]
Loading weights:  49%|████▊     | 384/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.attention.self.key.weight]
Loading weights:  49%|████▉     | 385/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.attention.self.query.bias]
Loading weights:  49%|████▉     | 385/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.attention.self.query.bias]
Loading weights:  49%|████▉     | 386/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.attention.self.query.weight]
Loading weights:  49%|████▉     | 386/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.attention.self.query.weight]
Loading weights:  49%|████▉     | 387/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.attention.self.value.bias]  
Loading weights:  49%|████▉     | 387/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.attention.self.value.bias]
Loading weights:  49%|████▉     | 388/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.attention.self.value.weight]
Loading weights:  49%|████▉     | 388/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.attention.self.value.weight]
Loading weights:  49%|████▉     | 389/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.crossattention.output.LayerNorm.bias]
Loading weights:  49%|████▉     | 389/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.crossattention.output.LayerNorm.bias]
Loading weights:  49%|████▉     | 390/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.crossattention.output.LayerNorm.weight]
Loading weights:  49%|████▉     | 390/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.crossattention.output.LayerNorm.weight]
Loading weights:  50%|████▉     | 391/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.crossattention.output.dense.bias]      
Loading weights:  50%|████▉     | 391/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.crossattention.output.dense.bias]
Loading weights:  50%|████▉     | 392/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.crossattention.output.dense.weight]
Loading weights:  50%|████▉     | 392/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.crossattention.output.dense.weight]
Loading weights:  50%|████▉     | 393/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.crossattention.self.key.bias]      
Loading weights:  50%|████▉     | 393/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.crossattention.self.key.bias]
Loading weights:  50%|█████     | 394/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.crossattention.self.key.weight]
Loading weights:  50%|█████     | 394/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.crossattention.self.key.weight]
Loading weights:  50%|█████     | 395/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.crossattention.self.query.bias]
Loading weights:  50%|█████     | 395/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.crossattention.self.query.bias]
Loading weights:  50%|█████     | 396/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.crossattention.self.query.weight]
Loading weights:  50%|█████     | 396/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.crossattention.self.query.weight]
Loading weights:  50%|█████     | 397/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.crossattention.self.value.bias]  
Loading weights:  50%|█████     | 397/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.crossattention.self.value.bias]
Loading weights:  51%|█████     | 398/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.crossattention.self.value.weight]
Loading weights:  51%|█████     | 398/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.crossattention.self.value.weight]
Loading weights:  51%|█████     | 399/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.intermediate.dense.bias]         
Loading weights:  51%|█████     | 399/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.intermediate.dense.bias]
Loading weights:  51%|█████     | 400/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.intermediate.dense.weight]
Loading weights:  51%|█████     | 400/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.intermediate.dense.weight]
Loading weights:  51%|█████     | 401/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.output.LayerNorm.bias]    
Loading weights:  51%|█████     | 401/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.output.LayerNorm.bias]
Loading weights:  51%|█████     | 402/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.output.LayerNorm.weight]
Loading weights:  51%|█████     | 402/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.output.LayerNorm.weight]
Loading weights:  51%|█████     | 403/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.output.dense.bias]      
Loading weights:  51%|█████     | 403/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.output.dense.bias]
Loading weights:  51%|█████▏    | 404/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.output.dense.weight]
Loading weights:  51%|█████▏    | 404/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.2.output.dense.weight]
Loading weights:  51%|█████▏    | 405/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.attention.output.LayerNorm.bias]
Loading weights:  51%|█████▏    | 405/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.attention.output.LayerNorm.bias]
Loading weights:  52%|█████▏    | 406/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.attention.output.LayerNorm.weight]
Loading weights:  52%|█████▏    | 406/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.attention.output.LayerNorm.weight]
Loading weights:  52%|█████▏    | 407/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.attention.output.dense.bias]      
Loading weights:  52%|█████▏    | 407/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.attention.output.dense.bias]
Loading weights:  52%|█████▏    | 408/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.attention.output.dense.weight]
Loading weights:  52%|█████▏    | 408/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.attention.output.dense.weight]
Loading weights:  52%|█████▏    | 409/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.attention.self.key.bias]      
Loading weights:  52%|█████▏    | 409/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.attention.self.key.bias]
Loading weights:  52%|█████▏    | 410/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.attention.self.key.weight]
Loading weights:  52%|█████▏    | 410/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.attention.self.key.weight]
Loading weights:  52%|█████▏    | 411/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.attention.self.query.bias]
Loading weights:  52%|█████▏    | 411/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.attention.self.query.bias]
Loading weights:  52%|█████▏    | 412/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.attention.self.query.weight]
Loading weights:  52%|█████▏    | 412/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.attention.self.query.weight]
Loading weights:  52%|█████▏    | 413/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.attention.self.value.bias]  
Loading weights:  52%|█████▏    | 413/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.attention.self.value.bias]
Loading weights:  53%|█████▎    | 414/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.attention.self.value.weight]
Loading weights:  53%|█████▎    | 414/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.attention.self.value.weight]
Loading weights:  53%|█████▎    | 415/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.crossattention.output.LayerNorm.bias]
Loading weights:  53%|█████▎    | 415/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.crossattention.output.LayerNorm.bias]
Loading weights:  53%|█████▎    | 416/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.crossattention.output.LayerNorm.weight]
Loading weights:  53%|█████▎    | 416/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.crossattention.output.LayerNorm.weight]
Loading weights:  53%|█████▎    | 417/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.crossattention.output.dense.bias]      
Loading weights:  53%|█████▎    | 417/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.crossattention.output.dense.bias]
Loading weights:  53%|█████▎    | 418/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.crossattention.output.dense.weight]
Loading weights:  53%|█████▎    | 418/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.crossattention.output.dense.weight]
Loading weights:  53%|█████▎    | 419/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.crossattention.self.key.bias]      
Loading weights:  53%|█████▎    | 419/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.crossattention.self.key.bias]
Loading weights:  53%|█████▎    | 420/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.crossattention.self.key.weight]
Loading weights:  53%|█████▎    | 420/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.crossattention.self.key.weight]
Loading weights:  53%|█████▎    | 421/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.crossattention.self.query.bias]
Loading weights:  53%|█████▎    | 421/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.crossattention.self.query.bias]
Loading weights:  54%|█████▎    | 422/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.crossattention.self.query.weight]
Loading weights:  54%|█████▎    | 422/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.crossattention.self.query.weight]
Loading weights:  54%|█████▎    | 423/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.crossattention.self.value.bias]  
Loading weights:  54%|█████▎    | 423/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.crossattention.self.value.bias]
Loading weights:  54%|█████▍    | 424/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.crossattention.self.value.weight]
Loading weights:  54%|█████▍    | 424/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.crossattention.self.value.weight]
Loading weights:  54%|█████▍    | 425/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.intermediate.dense.bias]         
Loading weights:  54%|█████▍    | 425/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.intermediate.dense.bias]
Loading weights:  54%|█████▍    | 426/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.intermediate.dense.weight]
Loading weights:  54%|█████▍    | 426/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.intermediate.dense.weight]
Loading weights:  54%|█████▍    | 427/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.output.LayerNorm.bias]    
Loading weights:  54%|█████▍    | 427/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.output.LayerNorm.bias]
Loading weights:  54%|█████▍    | 428/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.output.LayerNorm.weight]
Loading weights:  54%|█████▍    | 428/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.output.LayerNorm.weight]
Loading weights:  54%|█████▍    | 429/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.output.dense.bias]      
Loading weights:  54%|█████▍    | 429/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.output.dense.bias]
Loading weights:  55%|█████▍    | 430/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.output.dense.weight]
Loading weights:  55%|█████▍    | 430/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.3.output.dense.weight]
Loading weights:  55%|█████▍    | 431/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.attention.output.LayerNorm.bias]
Loading weights:  55%|█████▍    | 431/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.attention.output.LayerNorm.bias]
Loading weights:  55%|█████▍    | 432/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.attention.output.LayerNorm.weight]
Loading weights:  55%|█████▍    | 432/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.attention.output.LayerNorm.weight]
Loading weights:  55%|█████▍    | 433/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.attention.output.dense.bias]      
Loading weights:  55%|█████▍    | 433/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.attention.output.dense.bias]
Loading weights:  55%|█████▌    | 434/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.attention.output.dense.weight]
Loading weights:  55%|█████▌    | 434/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.attention.output.dense.weight]
Loading weights:  55%|█████▌    | 435/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.attention.self.key.bias]      
Loading weights:  55%|█████▌    | 435/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.attention.self.key.bias]
Loading weights:  55%|█████▌    | 436/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.attention.self.key.weight]
Loading weights:  55%|█████▌    | 436/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.attention.self.key.weight]
Loading weights:  55%|█████▌    | 437/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.attention.self.query.bias]
Loading weights:  55%|█████▌    | 437/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.attention.self.query.bias]
Loading weights:  56%|█████▌    | 438/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.attention.self.query.weight]
Loading weights:  56%|█████▌    | 438/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.attention.self.query.weight]
Loading weights:  56%|█████▌    | 439/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.attention.self.value.bias]  
Loading weights:  56%|█████▌    | 439/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.attention.self.value.bias]
Loading weights:  56%|█████▌    | 440/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.attention.self.value.weight]
Loading weights:  56%|█████▌    | 440/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.attention.self.value.weight]
Loading weights:  56%|█████▌    | 441/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.crossattention.output.LayerNorm.bias]
Loading weights:  56%|█████▌    | 441/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.crossattention.output.LayerNorm.bias]
Loading weights:  56%|█████▌    | 442/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.crossattention.output.LayerNorm.weight]
Loading weights:  56%|█████▌    | 442/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.crossattention.output.LayerNorm.weight]
Loading weights:  56%|█████▌    | 443/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.crossattention.output.dense.bias]      
Loading weights:  56%|█████▌    | 443/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.crossattention.output.dense.bias]
Loading weights:  56%|█████▋    | 444/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.crossattention.output.dense.weight]
Loading weights:  56%|█████▋    | 444/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.crossattention.output.dense.weight]
Loading weights:  56%|█████▋    | 445/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.crossattention.self.key.bias]      
Loading weights:  56%|█████▋    | 445/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.crossattention.self.key.bias]
Loading weights:  57%|█████▋    | 446/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.crossattention.self.key.weight]
Loading weights:  57%|█████▋    | 446/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.crossattention.self.key.weight]
Loading weights:  57%|█████▋    | 447/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.crossattention.self.query.bias]
Loading weights:  57%|█████▋    | 447/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.crossattention.self.query.bias]
Loading weights:  57%|█████▋    | 448/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.crossattention.self.query.weight]
Loading weights:  57%|█████▋    | 448/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.crossattention.self.query.weight]
Loading weights:  57%|█████▋    | 449/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.crossattention.self.value.bias]  
Loading weights:  57%|█████▋    | 449/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.crossattention.self.value.bias]
Loading weights:  57%|█████▋    | 450/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.crossattention.self.value.weight]
Loading weights:  57%|█████▋    | 450/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.crossattention.self.value.weight]
Loading weights:  57%|█████▋    | 451/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.intermediate.dense.bias]         
Loading weights:  57%|█████▋    | 451/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.intermediate.dense.bias]
Loading weights:  57%|█████▋    | 452/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.intermediate.dense.weight]
Loading weights:  57%|█████▋    | 452/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.intermediate.dense.weight]
Loading weights:  57%|█████▋    | 453/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.output.LayerNorm.bias]    
Loading weights:  57%|█████▋    | 453/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.output.LayerNorm.bias]
Loading weights:  58%|█████▊    | 454/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.output.LayerNorm.weight]
Loading weights:  58%|█████▊    | 454/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.output.LayerNorm.weight]
Loading weights:  58%|█████▊    | 455/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.output.dense.bias]      
Loading weights:  58%|█████▊    | 455/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.output.dense.bias]
Loading weights:  58%|█████▊    | 456/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.output.dense.weight]
Loading weights:  58%|█████▊    | 456/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.4.output.dense.weight]
Loading weights:  58%|█████▊    | 457/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.attention.output.LayerNorm.bias]
Loading weights:  58%|█████▊    | 457/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.attention.output.LayerNorm.bias]
Loading weights:  58%|█████▊    | 458/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.attention.output.LayerNorm.weight]
Loading weights:  58%|█████▊    | 458/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.attention.output.LayerNorm.weight]
Loading weights:  58%|█████▊    | 459/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.attention.output.dense.bias]      
Loading weights:  58%|█████▊    | 459/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.attention.output.dense.bias]
Loading weights:  58%|█████▊    | 460/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.attention.output.dense.weight]
Loading weights:  58%|█████▊    | 460/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.attention.output.dense.weight]
Loading weights:  59%|█████▊    | 461/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.attention.self.key.bias]      
Loading weights:  59%|█████▊    | 461/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.attention.self.key.bias]
Loading weights:  59%|█████▊    | 462/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.attention.self.key.weight]
Loading weights:  59%|█████▊    | 462/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.attention.self.key.weight]
Loading weights:  59%|█████▉    | 463/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.attention.self.query.bias]
Loading weights:  59%|█████▉    | 463/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.attention.self.query.bias]
Loading weights:  59%|█████▉    | 464/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.attention.self.query.weight]
Loading weights:  59%|█████▉    | 464/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.attention.self.query.weight]
Loading weights:  59%|█████▉    | 465/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.attention.self.value.bias]  
Loading weights:  59%|█████▉    | 465/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.attention.self.value.bias]
Loading weights:  59%|█████▉    | 466/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.attention.self.value.weight]
Loading weights:  59%|█████▉    | 466/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.attention.self.value.weight]
Loading weights:  59%|█████▉    | 467/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.crossattention.output.LayerNorm.bias]
Loading weights:  59%|█████▉    | 467/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.crossattention.output.LayerNorm.bias]
Loading weights:  59%|█████▉    | 468/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.crossattention.output.LayerNorm.weight]
Loading weights:  59%|█████▉    | 468/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.crossattention.output.LayerNorm.weight]
Loading weights:  60%|█████▉    | 469/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.crossattention.output.dense.bias]      
Loading weights:  60%|█████▉    | 469/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.crossattention.output.dense.bias]
Loading weights:  60%|█████▉    | 470/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.crossattention.output.dense.weight]
Loading weights:  60%|█████▉    | 470/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.crossattention.output.dense.weight]
Loading weights:  60%|█████▉    | 471/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.crossattention.self.key.bias]      
Loading weights:  60%|█████▉    | 471/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.crossattention.self.key.bias]
Loading weights:  60%|█████▉    | 472/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.crossattention.self.key.weight]
Loading weights:  60%|█████▉    | 472/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.crossattention.self.key.weight]
Loading weights:  60%|██████    | 473/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.crossattention.self.query.bias]
Loading weights:  60%|██████    | 473/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.crossattention.self.query.bias]
Loading weights:  60%|██████    | 474/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.crossattention.self.query.weight]
Loading weights:  60%|██████    | 474/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.crossattention.self.query.weight]
Loading weights:  60%|██████    | 475/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.crossattention.self.value.bias]  
Loading weights:  60%|██████    | 475/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.crossattention.self.value.bias]
Loading weights:  60%|██████    | 476/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.crossattention.self.value.weight]
Loading weights:  60%|██████    | 476/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.crossattention.self.value.weight]
Loading weights:  61%|██████    | 477/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.intermediate.dense.bias]         
Loading weights:  61%|██████    | 477/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.intermediate.dense.bias]
Loading weights:  61%|██████    | 478/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.intermediate.dense.weight]
Loading weights:  61%|██████    | 478/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.intermediate.dense.weight]
Loading weights:  61%|██████    | 479/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.output.LayerNorm.bias]    
Loading weights:  61%|██████    | 479/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.output.LayerNorm.bias]
Loading weights:  61%|██████    | 480/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.output.LayerNorm.weight]
Loading weights:  61%|██████    | 480/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.output.LayerNorm.weight]
Loading weights:  61%|██████    | 481/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.output.dense.bias]      
Loading weights:  61%|██████    | 481/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.output.dense.bias]
Loading weights:  61%|██████    | 482/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.output.dense.weight]
Loading weights:  61%|██████    | 482/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.5.output.dense.weight]
Loading weights:  61%|██████▏   | 483/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.attention.output.LayerNorm.bias]
Loading weights:  61%|██████▏   | 483/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.attention.output.LayerNorm.bias]
Loading weights:  61%|██████▏   | 484/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.attention.output.LayerNorm.weight]
Loading weights:  61%|██████▏   | 484/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.attention.output.LayerNorm.weight]
Loading weights:  62%|██████▏   | 485/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.attention.output.dense.bias]      
Loading weights:  62%|██████▏   | 485/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.attention.output.dense.bias]
Loading weights:  62%|██████▏   | 486/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.attention.output.dense.weight]
Loading weights:  62%|██████▏   | 486/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.attention.output.dense.weight]
Loading weights:  62%|██████▏   | 487/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.attention.self.key.bias]      
Loading weights:  62%|██████▏   | 487/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.attention.self.key.bias]
Loading weights:  62%|██████▏   | 488/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.attention.self.key.weight]
Loading weights:  62%|██████▏   | 488/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.attention.self.key.weight]
Loading weights:  62%|██████▏   | 489/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.attention.self.query.bias]
Loading weights:  62%|██████▏   | 489/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.attention.self.query.bias]
Loading weights:  62%|██████▏   | 490/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.attention.self.query.weight]
Loading weights:  62%|██████▏   | 490/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.attention.self.query.weight]
Loading weights:  62%|██████▏   | 491/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.attention.self.value.bias]  
Loading weights:  62%|██████▏   | 491/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.attention.self.value.bias]
Loading weights:  62%|██████▏   | 492/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.attention.self.value.weight]
Loading weights:  62%|██████▏   | 492/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.attention.self.value.weight]
Loading weights:  63%|██████▎   | 493/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.crossattention.output.LayerNorm.bias]
Loading weights:  63%|██████▎   | 493/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.crossattention.output.LayerNorm.bias]
Loading weights:  63%|██████▎   | 494/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.crossattention.output.LayerNorm.weight]
Loading weights:  63%|██████▎   | 494/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.crossattention.output.LayerNorm.weight]
Loading weights:  63%|██████▎   | 495/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.crossattention.output.dense.bias]      
Loading weights:  63%|██████▎   | 495/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.crossattention.output.dense.bias]
Loading weights:  63%|██████▎   | 496/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.crossattention.output.dense.weight]
Loading weights:  63%|██████▎   | 496/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.crossattention.output.dense.weight]
Loading weights:  63%|██████▎   | 497/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.crossattention.self.key.bias]      
Loading weights:  63%|██████▎   | 497/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.crossattention.self.key.bias]
Loading weights:  63%|██████▎   | 498/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.crossattention.self.key.weight]
Loading weights:  63%|██████▎   | 498/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.crossattention.self.key.weight]
Loading weights:  63%|██████▎   | 499/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.crossattention.self.query.bias]
Loading weights:  63%|██████▎   | 499/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.crossattention.self.query.bias]
Loading weights:  63%|██████▎   | 500/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.crossattention.self.query.weight]
Loading weights:  63%|██████▎   | 500/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.crossattention.self.query.weight]
Loading weights:  64%|██████▎   | 501/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.crossattention.self.value.bias]  
Loading weights:  64%|██████▎   | 501/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.crossattention.self.value.bias]
Loading weights:  64%|██████▎   | 502/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.crossattention.self.value.weight]
Loading weights:  64%|██████▎   | 502/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.crossattention.self.value.weight]
Loading weights:  64%|██████▍   | 503/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.intermediate.dense.bias]         
Loading weights:  64%|██████▍   | 503/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.intermediate.dense.bias]
Loading weights:  64%|██████▍   | 504/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.intermediate.dense.weight]
Loading weights:  64%|██████▍   | 504/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.intermediate.dense.weight]
Loading weights:  64%|██████▍   | 505/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.output.LayerNorm.bias]    
Loading weights:  64%|██████▍   | 505/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.output.LayerNorm.bias]
Loading weights:  64%|██████▍   | 506/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.output.LayerNorm.weight]
Loading weights:  64%|██████▍   | 506/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.output.LayerNorm.weight]
Loading weights:  64%|██████▍   | 507/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.output.dense.bias]      
Loading weights:  64%|██████▍   | 507/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.output.dense.bias]
Loading weights:  64%|██████▍   | 508/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.output.dense.weight]
Loading weights:  64%|██████▍   | 508/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.6.output.dense.weight]
Loading weights:  65%|██████▍   | 509/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.attention.output.LayerNorm.bias]
Loading weights:  65%|██████▍   | 509/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.attention.output.LayerNorm.bias]
Loading weights:  65%|██████▍   | 510/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.attention.output.LayerNorm.weight]
Loading weights:  65%|██████▍   | 510/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.attention.output.LayerNorm.weight]
Loading weights:  65%|██████▍   | 511/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.attention.output.dense.bias]      
Loading weights:  65%|██████▍   | 511/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.attention.output.dense.bias]
Loading weights:  65%|██████▍   | 512/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.attention.output.dense.weight]
Loading weights:  65%|██████▍   | 512/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.attention.output.dense.weight]
Loading weights:  65%|██████▌   | 513/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.attention.self.key.bias]      
Loading weights:  65%|██████▌   | 513/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.attention.self.key.bias]
Loading weights:  65%|██████▌   | 514/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.attention.self.key.weight]
Loading weights:  65%|██████▌   | 514/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.attention.self.key.weight]
Loading weights:  65%|██████▌   | 515/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.attention.self.query.bias]
Loading weights:  65%|██████▌   | 515/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.attention.self.query.bias]
Loading weights:  65%|██████▌   | 516/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.attention.self.query.weight]
Loading weights:  65%|██████▌   | 516/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.attention.self.query.weight]
Loading weights:  66%|██████▌   | 517/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.attention.self.value.bias]  
Loading weights:  66%|██████▌   | 517/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.attention.self.value.bias]
Loading weights:  66%|██████▌   | 518/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.attention.self.value.weight]
Loading weights:  66%|██████▌   | 518/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.attention.self.value.weight]
Loading weights:  66%|██████▌   | 519/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.crossattention.output.LayerNorm.bias]
Loading weights:  66%|██████▌   | 519/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.crossattention.output.LayerNorm.bias]
Loading weights:  66%|██████▌   | 520/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.crossattention.output.LayerNorm.weight]
Loading weights:  66%|██████▌   | 520/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.crossattention.output.LayerNorm.weight]
Loading weights:  66%|██████▌   | 521/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.crossattention.output.dense.bias]      
Loading weights:  66%|██████▌   | 521/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.crossattention.output.dense.bias]
Loading weights:  66%|██████▌   | 522/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.crossattention.output.dense.weight]
Loading weights:  66%|██████▌   | 522/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.crossattention.output.dense.weight]
Loading weights:  66%|██████▋   | 523/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.crossattention.self.key.bias]      
Loading weights:  66%|██████▋   | 523/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.crossattention.self.key.bias]
Loading weights:  66%|██████▋   | 524/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.crossattention.self.key.weight]
Loading weights:  66%|██████▋   | 524/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.crossattention.self.key.weight]
Loading weights:  67%|██████▋   | 525/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.crossattention.self.query.bias]
Loading weights:  67%|██████▋   | 525/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.crossattention.self.query.bias]
Loading weights:  67%|██████▋   | 526/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.crossattention.self.query.weight]
Loading weights:  67%|██████▋   | 526/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.crossattention.self.query.weight]
Loading weights:  67%|██████▋   | 527/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.crossattention.self.value.bias]  
Loading weights:  67%|██████▋   | 527/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.crossattention.self.value.bias]
Loading weights:  67%|██████▋   | 528/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.crossattention.self.value.weight]
Loading weights:  67%|██████▋   | 528/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.crossattention.self.value.weight]
Loading weights:  67%|██████▋   | 529/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.intermediate.dense.bias]         
Loading weights:  67%|██████▋   | 529/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.intermediate.dense.bias]
Loading weights:  67%|██████▋   | 530/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.intermediate.dense.weight]
Loading weights:  67%|██████▋   | 530/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.intermediate.dense.weight]
Loading weights:  67%|██████▋   | 531/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.output.LayerNorm.bias]    
Loading weights:  67%|██████▋   | 531/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.output.LayerNorm.bias]
Loading weights:  68%|██████▊   | 532/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.output.LayerNorm.weight]
Loading weights:  68%|██████▊   | 532/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.output.LayerNorm.weight]
Loading weights:  68%|██████▊   | 533/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.output.dense.bias]      
Loading weights:  68%|██████▊   | 533/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.output.dense.bias]
Loading weights:  68%|██████▊   | 534/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.output.dense.weight]
Loading weights:  68%|██████▊   | 534/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.7.output.dense.weight]
Loading weights:  68%|██████▊   | 535/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.attention.output.LayerNorm.bias]
Loading weights:  68%|██████▊   | 535/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.attention.output.LayerNorm.bias]
Loading weights:  68%|██████▊   | 536/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.attention.output.LayerNorm.weight]
Loading weights:  68%|██████▊   | 536/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.attention.output.LayerNorm.weight]
Loading weights:  68%|██████▊   | 537/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.attention.output.dense.bias]      
Loading weights:  68%|██████▊   | 537/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.attention.output.dense.bias]
Loading weights:  68%|██████▊   | 538/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.attention.output.dense.weight]
Loading weights:  68%|██████▊   | 538/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.attention.output.dense.weight]
Loading weights:  68%|██████▊   | 539/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.attention.self.key.bias]      
Loading weights:  68%|██████▊   | 539/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.attention.self.key.bias]
Loading weights:  69%|██████▊   | 540/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.attention.self.key.weight]
Loading weights:  69%|██████▊   | 540/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.attention.self.key.weight]
Loading weights:  69%|██████▊   | 541/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.attention.self.query.bias]
Loading weights:  69%|██████▊   | 541/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.attention.self.query.bias]
Loading weights:  69%|██████▉   | 542/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.attention.self.query.weight]
Loading weights:  69%|██████▉   | 542/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.attention.self.query.weight]
Loading weights:  69%|██████▉   | 543/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.attention.self.value.bias]  
Loading weights:  69%|██████▉   | 543/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.attention.self.value.bias]
Loading weights:  69%|██████▉   | 544/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.attention.self.value.weight]
Loading weights:  69%|██████▉   | 544/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.attention.self.value.weight]
Loading weights:  69%|██████▉   | 545/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.crossattention.output.LayerNorm.bias]
Loading weights:  69%|██████▉   | 545/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.crossattention.output.LayerNorm.bias]
Loading weights:  69%|██████▉   | 546/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.crossattention.output.LayerNorm.weight]
Loading weights:  69%|██████▉   | 546/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.crossattention.output.LayerNorm.weight]
Loading weights:  69%|██████▉   | 547/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.crossattention.output.dense.bias]      
Loading weights:  69%|██████▉   | 547/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.crossattention.output.dense.bias]
Loading weights:  70%|██████▉   | 548/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.crossattention.output.dense.weight]
Loading weights:  70%|██████▉   | 548/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.crossattention.output.dense.weight]
Loading weights:  70%|██████▉   | 549/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.crossattention.self.key.bias]      
Loading weights:  70%|██████▉   | 549/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.crossattention.self.key.bias]
Loading weights:  70%|██████▉   | 550/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.crossattention.self.key.weight]
Loading weights:  70%|██████▉   | 550/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.crossattention.self.key.weight]
Loading weights:  70%|██████▉   | 551/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.crossattention.self.query.bias]
Loading weights:  70%|██████▉   | 551/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.crossattention.self.query.bias]
Loading weights:  70%|███████   | 552/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.crossattention.self.query.weight]
Loading weights:  70%|███████   | 552/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.crossattention.self.query.weight]
Loading weights:  70%|███████   | 553/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.crossattention.self.value.bias]  
Loading weights:  70%|███████   | 553/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.crossattention.self.value.bias]
Loading weights:  70%|███████   | 554/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.crossattention.self.value.weight]
Loading weights:  70%|███████   | 554/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.crossattention.self.value.weight]
Loading weights:  70%|███████   | 555/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.intermediate.dense.bias]         
Loading weights:  70%|███████   | 555/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.intermediate.dense.bias]
Loading weights:  71%|███████   | 556/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.intermediate.dense.weight]
Loading weights:  71%|███████   | 556/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.intermediate.dense.weight]
Loading weights:  71%|███████   | 557/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.output.LayerNorm.bias]    
Loading weights:  71%|███████   | 557/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.output.LayerNorm.bias]
Loading weights:  71%|███████   | 558/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.output.LayerNorm.weight]
Loading weights:  71%|███████   | 558/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.output.LayerNorm.weight]
Loading weights:  71%|███████   | 559/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.output.dense.bias]      
Loading weights:  71%|███████   | 559/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.output.dense.bias]
Loading weights:  71%|███████   | 560/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.output.dense.weight]
Loading weights:  71%|███████   | 560/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.8.output.dense.weight]
Loading weights:  71%|███████   | 561/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.attention.output.LayerNorm.bias]
Loading weights:  71%|███████   | 561/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.attention.output.LayerNorm.bias]
Loading weights:  71%|███████▏  | 562/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.attention.output.LayerNorm.weight]
Loading weights:  71%|███████▏  | 562/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.attention.output.LayerNorm.weight]
Loading weights:  71%|███████▏  | 563/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.attention.output.dense.bias]      
Loading weights:  71%|███████▏  | 563/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.attention.output.dense.bias]
Loading weights:  72%|███████▏  | 564/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.attention.output.dense.weight]
Loading weights:  72%|███████▏  | 564/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.attention.output.dense.weight]
Loading weights:  72%|███████▏  | 565/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.attention.self.key.bias]      
Loading weights:  72%|███████▏  | 565/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.attention.self.key.bias]
Loading weights:  72%|███████▏  | 566/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.attention.self.key.weight]
Loading weights:  72%|███████▏  | 566/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.attention.self.key.weight]
Loading weights:  72%|███████▏  | 567/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.attention.self.query.bias]
Loading weights:  72%|███████▏  | 567/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.attention.self.query.bias]
Loading weights:  72%|███████▏  | 568/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.attention.self.query.weight]
Loading weights:  72%|███████▏  | 568/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.attention.self.query.weight]
Loading weights:  72%|███████▏  | 569/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.attention.self.value.bias]  
Loading weights:  72%|███████▏  | 569/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.attention.self.value.bias]
Loading weights:  72%|███████▏  | 570/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.attention.self.value.weight]
Loading weights:  72%|███████▏  | 570/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.attention.self.value.weight]
Loading weights:  72%|███████▏  | 571/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.crossattention.output.LayerNorm.bias]
Loading weights:  72%|███████▏  | 571/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.crossattention.output.LayerNorm.bias]
Loading weights:  73%|███████▎  | 572/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.crossattention.output.LayerNorm.weight]
Loading weights:  73%|███████▎  | 572/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.crossattention.output.LayerNorm.weight]
Loading weights:  73%|███████▎  | 573/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.crossattention.output.dense.bias]      
Loading weights:  73%|███████▎  | 573/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.crossattention.output.dense.bias]
Loading weights:  73%|███████▎  | 574/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.crossattention.output.dense.weight]
Loading weights:  73%|███████▎  | 574/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.crossattention.output.dense.weight]
Loading weights:  73%|███████▎  | 575/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.crossattention.self.key.bias]      
Loading weights:  73%|███████▎  | 575/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.crossattention.self.key.bias]
Loading weights:  73%|███████▎  | 576/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.crossattention.self.key.weight]
Loading weights:  73%|███████▎  | 576/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.crossattention.self.key.weight]
Loading weights:  73%|███████▎  | 577/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.crossattention.self.query.bias]
Loading weights:  73%|███████▎  | 577/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.crossattention.self.query.bias]
Loading weights:  73%|███████▎  | 578/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.crossattention.self.query.weight]
Loading weights:  73%|███████▎  | 578/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.crossattention.self.query.weight]
Loading weights:  73%|███████▎  | 579/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.crossattention.self.value.bias]  
Loading weights:  73%|███████▎  | 579/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.crossattention.self.value.bias]
Loading weights:  74%|███████▎  | 580/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.crossattention.self.value.weight]
Loading weights:  74%|███████▎  | 580/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.crossattention.self.value.weight]
Loading weights:  74%|███████▎  | 581/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.intermediate.dense.bias]         
Loading weights:  74%|███████▎  | 581/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.intermediate.dense.bias]
Loading weights:  74%|███████▍  | 582/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.intermediate.dense.weight]
Loading weights:  74%|███████▍  | 582/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.intermediate.dense.weight]
Loading weights:  74%|███████▍  | 583/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.output.LayerNorm.bias]    
Loading weights:  74%|███████▍  | 583/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.output.LayerNorm.bias]
Loading weights:  74%|███████▍  | 584/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.output.LayerNorm.weight]
Loading weights:  74%|███████▍  | 584/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.output.LayerNorm.weight]
Loading weights:  74%|███████▍  | 585/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.output.dense.bias]      
Loading weights:  74%|███████▍  | 585/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.output.dense.bias]
Loading weights:  74%|███████▍  | 586/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.output.dense.weight]
Loading weights:  74%|███████▍  | 586/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.9.output.dense.weight]
Loading weights:  74%|███████▍  | 587/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.attention.output.LayerNorm.bias]
Loading weights:  74%|███████▍  | 587/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.attention.output.LayerNorm.bias]
Loading weights:  75%|███████▍  | 588/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.attention.output.LayerNorm.weight]
Loading weights:  75%|███████▍  | 588/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.attention.output.LayerNorm.weight]
Loading weights:  75%|███████▍  | 589/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.attention.output.dense.bias]      
Loading weights:  75%|███████▍  | 589/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.attention.output.dense.bias]
Loading weights:  75%|███████▍  | 590/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.attention.output.dense.weight]
Loading weights:  75%|███████▍  | 590/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.attention.output.dense.weight]
Loading weights:  75%|███████▌  | 591/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.attention.self.key.bias]      
Loading weights:  75%|███████▌  | 591/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.attention.self.key.bias]
Loading weights:  75%|███████▌  | 592/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.attention.self.key.weight]
Loading weights:  75%|███████▌  | 592/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.attention.self.key.weight]
Loading weights:  75%|███████▌  | 593/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.attention.self.query.bias]
Loading weights:  75%|███████▌  | 593/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.attention.self.query.bias]
Loading weights:  75%|███████▌  | 594/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.attention.self.query.weight]
Loading weights:  75%|███████▌  | 594/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.attention.self.query.weight]
Loading weights:  76%|███████▌  | 595/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.attention.self.value.bias]  
Loading weights:  76%|███████▌  | 595/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.attention.self.value.bias]
Loading weights:  76%|███████▌  | 596/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.attention.self.value.weight]
Loading weights:  76%|███████▌  | 596/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.attention.self.value.weight]
Loading weights:  76%|███████▌  | 597/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.crossattention.output.LayerNorm.bias]
Loading weights:  76%|███████▌  | 597/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.crossattention.output.LayerNorm.bias]
Loading weights:  76%|███████▌  | 598/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.crossattention.output.LayerNorm.weight]
Loading weights:  76%|███████▌  | 598/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.crossattention.output.LayerNorm.weight]
Loading weights:  76%|███████▌  | 599/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.crossattention.output.dense.bias]      
Loading weights:  76%|███████▌  | 599/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.crossattention.output.dense.bias]
Loading weights:  76%|███████▌  | 600/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.crossattention.output.dense.weight]
Loading weights:  76%|███████▌  | 600/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.crossattention.output.dense.weight]
Loading weights:  76%|███████▋  | 601/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.crossattention.self.key.bias]      
Loading weights:  76%|███████▋  | 601/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.crossattention.self.key.bias]
Loading weights:  76%|███████▋  | 602/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.crossattention.self.key.weight]
Loading weights:  76%|███████▋  | 602/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.crossattention.self.key.weight]
Loading weights:  77%|███████▋  | 603/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.crossattention.self.query.bias]
Loading weights:  77%|███████▋  | 603/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.crossattention.self.query.bias]
Loading weights:  77%|███████▋  | 604/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.crossattention.self.query.weight]
Loading weights:  77%|███████▋  | 604/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.crossattention.self.query.weight]
Loading weights:  77%|███████▋  | 605/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.crossattention.self.value.bias]  
Loading weights:  77%|███████▋  | 605/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.crossattention.self.value.bias]
Loading weights:  77%|███████▋  | 606/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.crossattention.self.value.weight]
Loading weights:  77%|███████▋  | 606/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.crossattention.self.value.weight]
Loading weights:  77%|███████▋  | 607/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.intermediate.dense.bias]         
Loading weights:  77%|███████▋  | 607/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.intermediate.dense.bias]
Loading weights:  77%|███████▋  | 608/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.intermediate.dense.weight]
Loading weights:  77%|███████▋  | 608/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.intermediate.dense.weight]
Loading weights:  77%|███████▋  | 609/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.output.LayerNorm.bias]    
Loading weights:  77%|███████▋  | 609/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.output.LayerNorm.bias]
Loading weights:  77%|███████▋  | 610/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.output.LayerNorm.weight]
Loading weights:  77%|███████▋  | 610/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.output.LayerNorm.weight]
Loading weights:  78%|███████▊  | 611/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.output.dense.bias]      
Loading weights:  78%|███████▊  | 611/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.output.dense.bias]
Loading weights:  78%|███████▊  | 612/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.output.dense.weight]
Loading weights:  78%|███████▊  | 612/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.10.output.dense.weight]
Loading weights:  78%|███████▊  | 613/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.attention.output.LayerNorm.bias]
Loading weights:  78%|███████▊  | 613/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.attention.output.LayerNorm.bias]
Loading weights:  78%|███████▊  | 614/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.attention.output.LayerNorm.weight]
Loading weights:  78%|███████▊  | 614/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.attention.output.LayerNorm.weight]
Loading weights:  78%|███████▊  | 615/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.attention.output.dense.bias]      
Loading weights:  78%|███████▊  | 615/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.attention.output.dense.bias]
Loading weights:  78%|███████▊  | 616/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.attention.output.dense.weight]
Loading weights:  78%|███████▊  | 616/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.attention.output.dense.weight]
Loading weights:  78%|███████▊  | 617/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.attention.self.key.bias]      
Loading weights:  78%|███████▊  | 617/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.attention.self.key.bias]
Loading weights:  78%|███████▊  | 618/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.attention.self.key.weight]
Loading weights:  78%|███████▊  | 618/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.attention.self.key.weight]
Loading weights:  79%|███████▊  | 619/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.attention.self.query.bias]
Loading weights:  79%|███████▊  | 619/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.attention.self.query.bias]
Loading weights:  79%|███████▊  | 620/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.attention.self.query.weight]
Loading weights:  79%|███████▊  | 620/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.attention.self.query.weight]
Loading weights:  79%|███████▉  | 621/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.attention.self.value.bias]  
Loading weights:  79%|███████▉  | 621/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.attention.self.value.bias]
Loading weights:  79%|███████▉  | 622/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.attention.self.value.weight]
Loading weights:  79%|███████▉  | 622/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.attention.self.value.weight]
Loading weights:  79%|███████▉  | 623/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.crossattention.output.LayerNorm.bias]
Loading weights:  79%|███████▉  | 623/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.crossattention.output.LayerNorm.bias]
Loading weights:  79%|███████▉  | 624/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.crossattention.output.LayerNorm.weight]
Loading weights:  79%|███████▉  | 624/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.crossattention.output.LayerNorm.weight]
Loading weights:  79%|███████▉  | 625/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.crossattention.output.dense.bias]      
Loading weights:  79%|███████▉  | 625/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.crossattention.output.dense.bias]
Loading weights:  79%|███████▉  | 626/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.crossattention.output.dense.weight]
Loading weights:  79%|███████▉  | 626/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.crossattention.output.dense.weight]
Loading weights:  80%|███████▉  | 627/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.crossattention.self.key.bias]      
Loading weights:  80%|███████▉  | 627/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.crossattention.self.key.bias]
Loading weights:  80%|███████▉  | 628/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.crossattention.self.key.weight]
Loading weights:  80%|███████▉  | 628/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.crossattention.self.key.weight]
Loading weights:  80%|███████▉  | 629/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.crossattention.self.query.bias]
Loading weights:  80%|███████▉  | 629/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.crossattention.self.query.bias]
Loading weights:  80%|███████▉  | 630/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.crossattention.self.query.weight]
Loading weights:  80%|███████▉  | 630/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.crossattention.self.query.weight]
Loading weights:  80%|████████  | 631/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.crossattention.self.value.bias]  
Loading weights:  80%|████████  | 631/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.crossattention.self.value.bias]
Loading weights:  80%|████████  | 632/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.crossattention.self.value.weight]
Loading weights:  80%|████████  | 632/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.crossattention.self.value.weight]
Loading weights:  80%|████████  | 633/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.intermediate.dense.bias]         
Loading weights:  80%|████████  | 633/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.intermediate.dense.bias]
Loading weights:  80%|████████  | 634/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.intermediate.dense.weight]
Loading weights:  80%|████████  | 634/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.intermediate.dense.weight]
Loading weights:  81%|████████  | 635/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.output.LayerNorm.bias]    
Loading weights:  81%|████████  | 635/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.output.LayerNorm.bias]
Loading weights:  81%|████████  | 636/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.output.LayerNorm.weight]
Loading weights:  81%|████████  | 636/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.output.LayerNorm.weight]
Loading weights:  81%|████████  | 637/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.output.dense.bias]      
Loading weights:  81%|████████  | 637/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.output.dense.bias]
Loading weights:  81%|████████  | 638/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.output.dense.weight]
Loading weights:  81%|████████  | 638/788 [00:00<00:00, 1169.03it/s, Materializing param=text_encoder.encoder.layer.11.output.dense.weight]
Loading weights:  81%|████████  | 639/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.embeddings.class_embedding]          
Loading weights:  81%|████████  | 639/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.embeddings.class_embedding]
Loading weights:  81%|████████  | 640/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.embeddings.patch_embedding.bias]
Loading weights:  81%|████████  | 640/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.embeddings.patch_embedding.bias]
Loading weights:  81%|████████▏ | 641/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.embeddings.patch_embedding.weight]
Loading weights:  81%|████████▏ | 641/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.embeddings.patch_embedding.weight]
Loading weights:  81%|████████▏ | 642/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.embeddings.position_embedding]    
Loading weights:  81%|████████▏ | 642/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.embeddings.position_embedding]
Loading weights:  82%|████████▏ | 643/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.0.layer_norm1.bias]
Loading weights:  82%|████████▏ | 643/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.0.layer_norm1.bias]
Loading weights:  82%|████████▏ | 644/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.0.layer_norm1.weight]
Loading weights:  82%|████████▏ | 644/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.0.layer_norm1.weight]
Loading weights:  82%|████████▏ | 645/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.0.layer_norm2.bias]  
Loading weights:  82%|████████▏ | 645/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.0.layer_norm2.bias]
Loading weights:  82%|████████▏ | 646/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.0.layer_norm2.weight]
Loading weights:  82%|████████▏ | 646/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.0.layer_norm2.weight]
Loading weights:  82%|████████▏ | 647/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.0.mlp.fc1.bias]      
Loading weights:  82%|████████▏ | 647/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.0.mlp.fc1.bias]
Loading weights:  82%|████████▏ | 648/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.0.mlp.fc1.weight]
Loading weights:  82%|████████▏ | 648/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.0.mlp.fc1.weight]
Loading weights:  82%|████████▏ | 649/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.0.mlp.fc2.bias]  
Loading weights:  82%|████████▏ | 649/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.0.mlp.fc2.bias]
Loading weights:  82%|████████▏ | 650/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.0.mlp.fc2.weight]
Loading weights:  82%|████████▏ | 650/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.0.mlp.fc2.weight]
Loading weights:  83%|████████▎ | 651/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.0.self_attn.projection.bias]
Loading weights:  83%|████████▎ | 651/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.0.self_attn.projection.bias]
Loading weights:  83%|████████▎ | 652/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.0.self_attn.projection.weight]
Loading weights:  83%|████████▎ | 652/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.0.self_attn.projection.weight]
Loading weights:  83%|████████▎ | 653/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.0.self_attn.qkv.bias]         
Loading weights:  83%|████████▎ | 653/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.0.self_attn.qkv.bias]
Loading weights:  83%|████████▎ | 654/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.0.self_attn.qkv.weight]
Loading weights:  83%|████████▎ | 654/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.0.self_attn.qkv.weight]
Loading weights:  83%|████████▎ | 655/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.1.layer_norm1.bias]    
Loading weights:  83%|████████▎ | 655/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.1.layer_norm1.bias]
Loading weights:  83%|████████▎ | 656/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.1.layer_norm1.weight]
Loading weights:  83%|████████▎ | 656/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.1.layer_norm1.weight]
Loading weights:  83%|████████▎ | 657/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.1.layer_norm2.bias]  
Loading weights:  83%|████████▎ | 657/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.1.layer_norm2.bias]
Loading weights:  84%|████████▎ | 658/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.1.layer_norm2.weight]
Loading weights:  84%|████████▎ | 658/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.1.layer_norm2.weight]
Loading weights:  84%|████████▎ | 659/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.1.mlp.fc1.bias]      
Loading weights:  84%|████████▎ | 659/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.1.mlp.fc1.bias]
Loading weights:  84%|████████▍ | 660/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.1.mlp.fc1.weight]
Loading weights:  84%|████████▍ | 660/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.1.mlp.fc1.weight]
Loading weights:  84%|████████▍ | 661/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.1.mlp.fc2.bias]  
Loading weights:  84%|████████▍ | 661/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.1.mlp.fc2.bias]
Loading weights:  84%|████████▍ | 662/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.1.mlp.fc2.weight]
Loading weights:  84%|████████▍ | 662/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.1.mlp.fc2.weight]
Loading weights:  84%|████████▍ | 663/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.1.self_attn.projection.bias]
Loading weights:  84%|████████▍ | 663/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.1.self_attn.projection.bias]
Loading weights:  84%|████████▍ | 664/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.1.self_attn.projection.weight]
Loading weights:  84%|████████▍ | 664/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.1.self_attn.projection.weight]
Loading weights:  84%|████████▍ | 665/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.1.self_attn.qkv.bias]         
Loading weights:  84%|████████▍ | 665/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.1.self_attn.qkv.bias]
Loading weights:  85%|████████▍ | 666/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.1.self_attn.qkv.weight]
Loading weights:  85%|████████▍ | 666/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.1.self_attn.qkv.weight]
Loading weights:  85%|████████▍ | 667/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.2.layer_norm1.bias]    
Loading weights:  85%|████████▍ | 667/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.2.layer_norm1.bias]
Loading weights:  85%|████████▍ | 668/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.2.layer_norm1.weight]
Loading weights:  85%|████████▍ | 668/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.2.layer_norm1.weight]
Loading weights:  85%|████████▍ | 669/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.2.layer_norm2.bias]  
Loading weights:  85%|████████▍ | 669/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.2.layer_norm2.bias]
Loading weights:  85%|████████▌ | 670/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.2.layer_norm2.weight]
Loading weights:  85%|████████▌ | 670/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.2.layer_norm2.weight]
Loading weights:  85%|████████▌ | 671/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.2.mlp.fc1.bias]      
Loading weights:  85%|████████▌ | 671/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.2.mlp.fc1.bias]
Loading weights:  85%|████████▌ | 672/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.2.mlp.fc1.weight]
Loading weights:  85%|████████▌ | 672/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.2.mlp.fc1.weight]
Loading weights:  85%|████████▌ | 673/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.2.mlp.fc2.bias]  
Loading weights:  85%|████████▌ | 673/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.2.mlp.fc2.bias]
Loading weights:  86%|████████▌ | 674/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.2.mlp.fc2.weight]
Loading weights:  86%|████████▌ | 674/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.2.mlp.fc2.weight]
Loading weights:  86%|████████▌ | 675/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.2.self_attn.projection.bias]
Loading weights:  86%|████████▌ | 675/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.2.self_attn.projection.bias]
Loading weights:  86%|████████▌ | 676/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.2.self_attn.projection.weight]
Loading weights:  86%|████████▌ | 676/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.2.self_attn.projection.weight]
Loading weights:  86%|████████▌ | 677/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.2.self_attn.qkv.bias]         
Loading weights:  86%|████████▌ | 677/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.2.self_attn.qkv.bias]
Loading weights:  86%|████████▌ | 678/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.2.self_attn.qkv.weight]
Loading weights:  86%|████████▌ | 678/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.2.self_attn.qkv.weight]
Loading weights:  86%|████████▌ | 679/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.3.layer_norm1.bias]    
Loading weights:  86%|████████▌ | 679/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.3.layer_norm1.bias]
Loading weights:  86%|████████▋ | 680/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.3.layer_norm1.weight]
Loading weights:  86%|████████▋ | 680/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.3.layer_norm1.weight]
Loading weights:  86%|████████▋ | 681/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.3.layer_norm2.bias]  
Loading weights:  86%|████████▋ | 681/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.3.layer_norm2.bias]
Loading weights:  87%|████████▋ | 682/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.3.layer_norm2.weight]
Loading weights:  87%|████████▋ | 682/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.3.layer_norm2.weight]
Loading weights:  87%|████████▋ | 683/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.3.mlp.fc1.bias]      
Loading weights:  87%|████████▋ | 683/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.3.mlp.fc1.bias]
Loading weights:  87%|████████▋ | 684/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.3.mlp.fc1.weight]
Loading weights:  87%|████████▋ | 684/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.3.mlp.fc1.weight]
Loading weights:  87%|████████▋ | 685/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.3.mlp.fc2.bias]  
Loading weights:  87%|████████▋ | 685/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.3.mlp.fc2.bias]
Loading weights:  87%|████████▋ | 686/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.3.mlp.fc2.weight]
Loading weights:  87%|████████▋ | 686/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.3.mlp.fc2.weight]
Loading weights:  87%|████████▋ | 687/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.3.self_attn.projection.bias]
Loading weights:  87%|████████▋ | 687/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.3.self_attn.projection.bias]
Loading weights:  87%|████████▋ | 688/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.3.self_attn.projection.weight]
Loading weights:  87%|████████▋ | 688/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.3.self_attn.projection.weight]
Loading weights:  87%|████████▋ | 689/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.3.self_attn.qkv.bias]         
Loading weights:  87%|████████▋ | 689/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.3.self_attn.qkv.bias]
Loading weights:  88%|████████▊ | 690/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.3.self_attn.qkv.weight]
Loading weights:  88%|████████▊ | 690/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.3.self_attn.qkv.weight]
Loading weights:  88%|████████▊ | 691/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.4.layer_norm1.bias]    
Loading weights:  88%|████████▊ | 691/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.4.layer_norm1.bias]
Loading weights:  88%|████████▊ | 692/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.4.layer_norm1.weight]
Loading weights:  88%|████████▊ | 692/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.4.layer_norm1.weight]
Loading weights:  88%|████████▊ | 693/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.4.layer_norm2.bias]  
Loading weights:  88%|████████▊ | 693/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.4.layer_norm2.bias]
Loading weights:  88%|████████▊ | 694/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.4.layer_norm2.weight]
Loading weights:  88%|████████▊ | 694/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.4.layer_norm2.weight]
Loading weights:  88%|████████▊ | 695/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.4.mlp.fc1.bias]      
Loading weights:  88%|████████▊ | 695/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.4.mlp.fc1.bias]
Loading weights:  88%|████████▊ | 696/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.4.mlp.fc1.weight]
Loading weights:  88%|████████▊ | 696/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.4.mlp.fc1.weight]
Loading weights:  88%|████████▊ | 697/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.4.mlp.fc2.bias]  
Loading weights:  88%|████████▊ | 697/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.4.mlp.fc2.bias]
Loading weights:  89%|████████▊ | 698/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.4.mlp.fc2.weight]
Loading weights:  89%|████████▊ | 698/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.4.mlp.fc2.weight]
Loading weights:  89%|████████▊ | 699/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.4.self_attn.projection.bias]
Loading weights:  89%|████████▊ | 699/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.4.self_attn.projection.bias]
Loading weights:  89%|████████▉ | 700/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.4.self_attn.projection.weight]
Loading weights:  89%|████████▉ | 700/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.4.self_attn.projection.weight]
Loading weights:  89%|████████▉ | 701/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.4.self_attn.qkv.bias]         
Loading weights:  89%|████████▉ | 701/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.4.self_attn.qkv.bias]
Loading weights:  89%|████████▉ | 702/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.4.self_attn.qkv.weight]
Loading weights:  89%|████████▉ | 702/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.4.self_attn.qkv.weight]
Loading weights:  89%|████████▉ | 703/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.5.layer_norm1.bias]    
Loading weights:  89%|████████▉ | 703/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.5.layer_norm1.bias]
Loading weights:  89%|████████▉ | 704/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.5.layer_norm1.weight]
Loading weights:  89%|████████▉ | 704/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.5.layer_norm1.weight]
Loading weights:  89%|████████▉ | 705/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.5.layer_norm2.bias]  
Loading weights:  89%|████████▉ | 705/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.5.layer_norm2.bias]
Loading weights:  90%|████████▉ | 706/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.5.layer_norm2.weight]
Loading weights:  90%|████████▉ | 706/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.5.layer_norm2.weight]
Loading weights:  90%|████████▉ | 707/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.5.mlp.fc1.bias]      
Loading weights:  90%|████████▉ | 707/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.5.mlp.fc1.bias]
Loading weights:  90%|████████▉ | 708/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.5.mlp.fc1.weight]
Loading weights:  90%|████████▉ | 708/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.5.mlp.fc1.weight]
Loading weights:  90%|████████▉ | 709/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.5.mlp.fc2.bias]  
Loading weights:  90%|████████▉ | 709/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.5.mlp.fc2.bias]
Loading weights:  90%|█████████ | 710/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.5.mlp.fc2.weight]
Loading weights:  90%|█████████ | 710/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.5.mlp.fc2.weight]
Loading weights:  90%|█████████ | 711/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.5.self_attn.projection.bias]
Loading weights:  90%|█████████ | 711/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.5.self_attn.projection.bias]
Loading weights:  90%|█████████ | 712/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.5.self_attn.projection.weight]
Loading weights:  90%|█████████ | 712/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.5.self_attn.projection.weight]
Loading weights:  90%|█████████ | 713/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.5.self_attn.qkv.bias]         
Loading weights:  90%|█████████ | 713/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.5.self_attn.qkv.bias]
Loading weights:  91%|█████████ | 714/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.5.self_attn.qkv.weight]
Loading weights:  91%|█████████ | 714/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.5.self_attn.qkv.weight]
Loading weights:  91%|█████████ | 715/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.6.layer_norm1.bias]    
Loading weights:  91%|█████████ | 715/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.6.layer_norm1.bias]
Loading weights:  91%|█████████ | 716/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.6.layer_norm1.weight]
Loading weights:  91%|█████████ | 716/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.6.layer_norm1.weight]
Loading weights:  91%|█████████ | 717/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.6.layer_norm2.bias]  
Loading weights:  91%|█████████ | 717/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.6.layer_norm2.bias]
Loading weights:  91%|█████████ | 718/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.6.layer_norm2.weight]
Loading weights:  91%|█████████ | 718/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.6.layer_norm2.weight]
Loading weights:  91%|█████████ | 719/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.6.mlp.fc1.bias]      
Loading weights:  91%|█████████ | 719/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.6.mlp.fc1.bias]
Loading weights:  91%|█████████▏| 720/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.6.mlp.fc1.weight]
Loading weights:  91%|█████████▏| 720/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.6.mlp.fc1.weight]
Loading weights:  91%|█████████▏| 721/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.6.mlp.fc2.bias]  
Loading weights:  91%|█████████▏| 721/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.6.mlp.fc2.bias]
Loading weights:  92%|█████████▏| 722/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.6.mlp.fc2.weight]
Loading weights:  92%|█████████▏| 722/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.6.mlp.fc2.weight]
Loading weights:  92%|█████████▏| 723/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.6.self_attn.projection.bias]
Loading weights:  92%|█████████▏| 723/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.6.self_attn.projection.bias]
Loading weights:  92%|█████████▏| 724/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.6.self_attn.projection.weight]
Loading weights:  92%|█████████▏| 724/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.6.self_attn.projection.weight]
Loading weights:  92%|█████████▏| 725/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.6.self_attn.qkv.bias]         
Loading weights:  92%|█████████▏| 725/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.6.self_attn.qkv.bias]
Loading weights:  92%|█████████▏| 726/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.6.self_attn.qkv.weight]
Loading weights:  92%|█████████▏| 726/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.6.self_attn.qkv.weight]
Loading weights:  92%|█████████▏| 727/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.7.layer_norm1.bias]    
Loading weights:  92%|█████████▏| 727/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.7.layer_norm1.bias]
Loading weights:  92%|█████████▏| 728/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.7.layer_norm1.weight]
Loading weights:  92%|█████████▏| 728/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.7.layer_norm1.weight]
Loading weights:  93%|█████████▎| 729/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.7.layer_norm2.bias]  
Loading weights:  93%|█████████▎| 729/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.7.layer_norm2.bias]
Loading weights:  93%|█████████▎| 730/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.7.layer_norm2.weight]
Loading weights:  93%|█████████▎| 730/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.7.layer_norm2.weight]
Loading weights:  93%|█████████▎| 731/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.7.mlp.fc1.bias]      
Loading weights:  93%|█████████▎| 731/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.7.mlp.fc1.bias]
Loading weights:  93%|█████████▎| 732/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.7.mlp.fc1.weight]
Loading weights:  93%|█████████▎| 732/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.7.mlp.fc1.weight]
Loading weights:  93%|█████████▎| 733/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.7.mlp.fc2.bias]  
Loading weights:  93%|█████████▎| 733/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.7.mlp.fc2.bias]
Loading weights:  93%|█████████▎| 734/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.7.mlp.fc2.weight]
Loading weights:  93%|█████████▎| 734/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.7.mlp.fc2.weight]
Loading weights:  93%|█████████▎| 735/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.7.self_attn.projection.bias]
Loading weights:  93%|█████████▎| 735/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.7.self_attn.projection.bias]
Loading weights:  93%|█████████▎| 736/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.7.self_attn.projection.weight]
Loading weights:  93%|█████████▎| 736/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.7.self_attn.projection.weight]
Loading weights:  94%|█████████▎| 737/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.7.self_attn.qkv.bias]         
Loading weights:  94%|█████████▎| 737/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.7.self_attn.qkv.bias]
Loading weights:  94%|█████████▎| 738/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.7.self_attn.qkv.weight]
Loading weights:  94%|█████████▎| 738/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.7.self_attn.qkv.weight]
Loading weights:  94%|█████████▍| 739/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.8.layer_norm1.bias]    
Loading weights:  94%|█████████▍| 739/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.8.layer_norm1.bias]
Loading weights:  94%|█████████▍| 740/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.8.layer_norm1.weight]
Loading weights:  94%|█████████▍| 740/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.8.layer_norm1.weight]
Loading weights:  94%|█████████▍| 741/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.8.layer_norm2.bias]  
Loading weights:  94%|█████████▍| 741/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.8.layer_norm2.bias]
Loading weights:  94%|█████████▍| 742/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.8.layer_norm2.weight]
Loading weights:  94%|█████████▍| 742/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.8.layer_norm2.weight]
Loading weights:  94%|█████████▍| 743/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.8.mlp.fc1.bias]      
Loading weights:  94%|█████████▍| 743/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.8.mlp.fc1.bias]
Loading weights:  94%|█████████▍| 744/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.8.mlp.fc1.weight]
Loading weights:  94%|█████████▍| 744/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.8.mlp.fc1.weight]
Loading weights:  95%|█████████▍| 745/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.8.mlp.fc2.bias]  
Loading weights:  95%|█████████▍| 745/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.8.mlp.fc2.bias]
Loading weights:  95%|█████████▍| 746/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.8.mlp.fc2.weight]
Loading weights:  95%|█████████▍| 746/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.8.mlp.fc2.weight]
Loading weights:  95%|█████████▍| 747/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.8.self_attn.projection.bias]
Loading weights:  95%|█████████▍| 747/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.8.self_attn.projection.bias]
Loading weights:  95%|█████████▍| 748/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.8.self_attn.projection.weight]
Loading weights:  95%|█████████▍| 748/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.8.self_attn.projection.weight]
Loading weights:  95%|█████████▌| 749/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.8.self_attn.qkv.bias]         
Loading weights:  95%|█████████▌| 749/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.8.self_attn.qkv.bias]
Loading weights:  95%|█████████▌| 750/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.8.self_attn.qkv.weight]
Loading weights:  95%|█████████▌| 750/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.8.self_attn.qkv.weight]
Loading weights:  95%|█████████▌| 751/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.9.layer_norm1.bias]    
Loading weights:  95%|█████████▌| 751/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.9.layer_norm1.bias]
Loading weights:  95%|█████████▌| 752/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.9.layer_norm1.weight]
Loading weights:  95%|█████████▌| 752/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.9.layer_norm1.weight]
Loading weights:  96%|█████████▌| 753/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.9.layer_norm2.bias]  
Loading weights:  96%|█████████▌| 753/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.9.layer_norm2.bias]
Loading weights:  96%|█████████▌| 754/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.9.layer_norm2.weight]
Loading weights:  96%|█████████▌| 754/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.9.layer_norm2.weight]
Loading weights:  96%|█████████▌| 755/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.9.mlp.fc1.bias]      
Loading weights:  96%|█████████▌| 755/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.9.mlp.fc1.bias]
Loading weights:  96%|█████████▌| 756/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.9.mlp.fc1.weight]
Loading weights:  96%|█████████▌| 756/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.9.mlp.fc1.weight]
Loading weights:  96%|█████████▌| 757/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.9.mlp.fc2.bias]  
Loading weights:  96%|█████████▌| 757/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.9.mlp.fc2.bias]
Loading weights:  96%|█████████▌| 758/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.9.mlp.fc2.weight]
Loading weights:  96%|█████████▌| 758/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.9.mlp.fc2.weight]
Loading weights:  96%|█████████▋| 759/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.9.self_attn.projection.bias]
Loading weights:  96%|█████████▋| 759/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.9.self_attn.projection.bias]
Loading weights:  96%|█████████▋| 760/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.9.self_attn.projection.weight]
Loading weights:  96%|█████████▋| 760/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.9.self_attn.projection.weight]
Loading weights:  97%|█████████▋| 761/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.9.self_attn.qkv.bias]         
Loading weights:  97%|█████████▋| 761/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.9.self_attn.qkv.bias]
Loading weights:  97%|█████████▋| 762/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.9.self_attn.qkv.weight]
Loading weights:  97%|█████████▋| 762/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.9.self_attn.qkv.weight]
Loading weights:  97%|█████████▋| 763/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.10.layer_norm1.bias]   
Loading weights:  97%|█████████▋| 763/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.10.layer_norm1.bias]
Loading weights:  97%|█████████▋| 764/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.10.layer_norm1.weight]
Loading weights:  97%|█████████▋| 764/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.10.layer_norm1.weight]
Loading weights:  97%|█████████▋| 765/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.10.layer_norm2.bias]  
Loading weights:  97%|█████████▋| 765/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.10.layer_norm2.bias]
Loading weights:  97%|█████████▋| 766/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.10.layer_norm2.weight]
Loading weights:  97%|█████████▋| 766/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.10.layer_norm2.weight]
Loading weights:  97%|█████████▋| 767/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.10.mlp.fc1.bias]      
Loading weights:  97%|█████████▋| 767/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.10.mlp.fc1.bias]
Loading weights:  97%|█████████▋| 768/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.10.mlp.fc1.weight]
Loading weights:  97%|█████████▋| 768/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.10.mlp.fc1.weight]
Loading weights:  98%|█████████▊| 769/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.10.mlp.fc2.bias]  
Loading weights:  98%|█████████▊| 769/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.10.mlp.fc2.bias]
Loading weights:  98%|█████████▊| 770/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.10.mlp.fc2.weight]
Loading weights:  98%|█████████▊| 770/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.10.mlp.fc2.weight]
Loading weights:  98%|█████████▊| 771/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.10.self_attn.projection.bias]
Loading weights:  98%|█████████▊| 771/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.10.self_attn.projection.bias]
Loading weights:  98%|█████████▊| 772/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.10.self_attn.projection.weight]
Loading weights:  98%|█████████▊| 772/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.10.self_attn.projection.weight]
Loading weights:  98%|█████████▊| 773/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.10.self_attn.qkv.bias]         
Loading weights:  98%|█████████▊| 773/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.10.self_attn.qkv.bias]
Loading weights:  98%|█████████▊| 774/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.10.self_attn.qkv.weight]
Loading weights:  98%|█████████▊| 774/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.10.self_attn.qkv.weight]
Loading weights:  98%|█████████▊| 775/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.11.layer_norm1.bias]    
Loading weights:  98%|█████████▊| 775/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.11.layer_norm1.bias]
Loading weights:  98%|█████████▊| 776/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.11.layer_norm1.weight]
Loading weights:  98%|█████████▊| 776/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.11.layer_norm1.weight]
Loading weights:  99%|█████████▊| 777/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.11.layer_norm2.bias]  
Loading weights:  99%|█████████▊| 777/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.11.layer_norm2.bias]
Loading weights:  99%|█████████▊| 778/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.11.layer_norm2.weight]
Loading weights:  99%|█████████▊| 778/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.11.layer_norm2.weight]
Loading weights:  99%|█████████▉| 779/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.11.mlp.fc1.bias]      
Loading weights:  99%|█████████▉| 779/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.11.mlp.fc1.bias]
Loading weights:  99%|█████████▉| 780/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.11.mlp.fc1.weight]
Loading weights:  99%|█████████▉| 780/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.11.mlp.fc1.weight]
Loading weights:  99%|█████████▉| 781/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.11.mlp.fc2.bias]  
Loading weights:  99%|█████████▉| 781/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.11.mlp.fc2.bias]
Loading weights:  99%|█████████▉| 782/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.11.mlp.fc2.weight]
Loading weights:  99%|█████████▉| 782/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.11.mlp.fc2.weight]
Loading weights:  99%|█████████▉| 783/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.11.self_attn.projection.bias]
Loading weights:  99%|█████████▉| 783/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.11.self_attn.projection.bias]
Loading weights:  99%|█████████▉| 784/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.11.self_attn.projection.weight]
Loading weights:  99%|█████████▉| 784/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.11.self_attn.projection.weight]
Loading weights: 100%|█████████▉| 785/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.11.self_attn.qkv.bias]         
Loading weights: 100%|█████████▉| 785/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.11.self_attn.qkv.bias]
Loading weights: 100%|█████████▉| 786/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.11.self_attn.qkv.weight]
Loading weights: 100%|█████████▉| 786/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.encoder.layers.11.self_attn.qkv.weight]
Loading weights: 100%|█████████▉| 787/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.post_layernorm.bias]                   
Loading weights: 100%|█████████▉| 787/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.post_layernorm.bias]
Loading weights: 100%|██████████| 788/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.post_layernorm.weight]
Loading weights: 100%|██████████| 788/788 [00:00<00:00, 1169.03it/s, Materializing param=vision_model.post_layernorm.weight]
Loading weights: 100%|██████████| 788/788 [00:00<00:00, 4890.80it/s, Materializing param=vision_model.post_layernorm.weight]
The tied weights mapping and config for this model specifies to tie text_decoder.bert.embeddings.word_embeddings.weight to text_decoder.cls.predictions.decoder.weight, but both are present in the checkpoints, so we will NOT tie them. You should update the config with `tie_word_embeddings=False` to silence this warning
BlipForQuestionAnswering LOAD REPORT from: Salesforce/blip-vqa-base
Key                                       | Status     |  | 
------------------------------------------+------------+--+-
text_encoder.embeddings.position_ids      | UNEXPECTED |  | 
text_decoder.bert.embeddings.position_ids | UNEXPECTED |  | 

Notes:
- UNEXPECTED	:can be ignored when loading from different task/architecture; not ok if you expect identical arch.
  ✓ BLIP (captioning + VQA)
Loading real instruction data from data/coco/annotations/captions_train2017.json...
✅ Loaded 5000 instruction samples from COCO

🚀 Starting Instruction Tuning (Reasoning Module)...

Epoch 1/4:   0%|          | 0/1250 [00:00<?, ?it/s]We strongly recommend passing in an `attention_mask` since your input_ids may be padded. See https://huggingface.co/docs/transformers/troubleshooting#incorrect-output-when-padding-tokens-arent-masked.

Epoch 1/4:   0%|          | 0/1250 [00:03<?, ?it/s, loss=9.85]
Epoch 1/4:   0%|          | 1/1250 [00:03<1:07:52,  3.26s/it, loss=9.85]
Epoch 1/4:   0%|          | 1/1250 [00:04<1:07:52,  3.26s/it, loss=9.23]
Epoch 1/4:   0%|          | 2/1250 [00:04<47:58,  2.31s/it, loss=9.23]  
Epoch 1/4:   0%|          | 2/1250 [00:06<47:58,  2.31s/it, loss=8.59]
Epoch 1/4:   0%|          | 3/1250 [00:06<38:20,  1.84s/it, loss=8.59]
Epoch 1/4:   0%|          | 3/1250 [00:07<38:20,  1.84s/it, loss=8.22]
Epoch 1/4:   0%|          | 4/1250 [00:07<35:46,  1.72s/it, loss=8.22]
Epoch 1/4:   0%|          | 4/1250 [00:09<35:46,  1.72s/it, loss=8.16]
Epoch 1/4:   0%|          | 5/1250 [00:09<35:47,  1.72s/it, loss=8.16]
Epoch 1/4:   0%|          | 5/1250 [00:10<35:47,  1.72s/it, loss=7.87]
Epoch 1/4:   0%|          | 6/1250 [00:10<32:33,  1.57s/it, loss=7.87]
Epoch 1/4:   0%|          | 6/1250 [00:12<32:33,  1.57s/it, loss=7.52]
Epoch 1/4:   1%|          | 7/1250 [00:12<31:13,  1.51s/it, loss=7.52]
Epoch 1/4:   1%|          | 7/1250 [00:13<31:13,  1.51s/it, loss=7.51]
Epoch 1/4:   1%|          | 8/1250 [00:13<31:24,  1.52s/it, loss=7.51]
Epoch 1/4:   1%|          | 8/1250 [00:15<31:24,  1.52s/it, loss=7.8] 
Epoch 1/4:   1%|          | 9/1250 [00:15<32:28,  1.57s/it, loss=7.8]
Epoch 1/4:   1%|          | 9/1250 [00:16<32:28,  1.57s/it, loss=6.59]
Epoch 1/4:   1%|          | 10/1250 [00:16<29:38,  1.43s/it, loss=6.59]
Epoch 1/4:   1%|          | 10/1250 [00:18<29:38,  1.43s/it, loss=6.93]
Epoch 1/4:   1%|          | 11/1250 [00:18<30:48,  1.49s/it, loss=6.93]
Epoch 1/4:   1%|          | 11/1250 [00:19<30:48,  1.49s/it, loss=7.27]
Epoch 1/4:   1%|          | 12/1250 [00:19<30:41,  1.49s/it, loss=7.27]
Epoch 1/4:   1%|          | 12/1250 [00:20<30:41,  1.49s/it, loss=7.1] 
Epoch 1/4:   1%|          | 13/1250 [00:20<29:37,  1.44s/it, loss=7.1]
Epoch 1/4:   1%|          | 13/1250 [00:22<29:37,  1.44s/it, loss=6.67]
Epoch 1/4:   1%|          | 14/1250 [00:22<29:14,  1.42s/it, loss=6.67]
Epoch 1/4:   1%|          | 14/1250 [00:23<29:14,  1.42s/it, loss=6.64]
Epoch 1/4:   1%|          | 15/1250 [00:23<30:51,  1.50s/it, loss=6.64]
Epoch 1/4:   1%|          | 15/1250 [00:25<30:51,  1.50s/it, loss=7.08]
Epoch 1/4:   1%|▏         | 16/1250 [00:25<30:50,  1.50s/it, loss=7.08]
Epoch 1/4:   1%|▏         | 16/1250 [00:27<30:50,  1.50s/it, loss=6.72]
Epoch 1/4:   1%|▏         | 17/1250 [00:27<33:25,  1.63s/it, loss=6.72]
Epoch 1/4:   1%|▏         | 17/1250 [00:28<33:25,  1.63s/it, loss=7.13]
Epoch 1/4:   1%|▏         | 18/1250 [00:28<33:16,  1.62s/it, loss=7.13]
Epoch 1/4:   1%|▏         | 18/1250 [00:30<33:16,  1.62s/it, loss=7.03]
Epoch 1/4:   2%|▏         | 19/1250 [00:30<32:37,  1.59s/it, loss=7.03]
Epoch 1/4:   2%|▏         | 19/1250 [00:32<32:37,  1.59s/it, loss=7.01]
Epoch 1/4:   2%|▏         | 20/1250 [00:32<32:23,  1.58s/it, loss=7.01]
Epoch 1/4:   2%|▏         | 20/1250 [00:33<32:23,  1.58s/it, loss=6.82]
Epoch 1/4:   2%|▏         | 21/1250 [00:33<32:18,  1.58s/it, loss=6.82]
Epoch 1/4:   2%|▏         | 21/1250 [00:35<32:18,  1.58s/it, loss=5.81]
Epoch 1/4:   2%|▏         | 22/1250 [00:35<31:39,  1.55s/it, loss=5.81]
Epoch 1/4:   2%|▏         | 22/1250 [00:36<31:39,  1.55s/it, loss=6.38]
Epoch 1/4:   2%|▏         | 23/1250 [00:36<31:02,  1.52s/it, loss=6.38]
Epoch 1/4:   2%|▏         | 23/1250 [00:38<31:02,  1.52s/it, loss=6.55]
Epoch 1/4:   2%|▏         | 24/1250 [00:38<31:11,  1.53s/it, loss=6.55]
Epoch 1/4:   2%|▏         | 24/1250 [00:39<31:11,  1.53s/it, loss=6.48]
Epoch 1/4:   2%|▏         | 25/1250 [00:39<31:01,  1.52s/it, loss=6.48]
Epoch 1/4:   2%|▏         | 25/1250 [00:41<31:01,  1.52s/it, loss=6.69]
Epoch 1/4:   2%|▏         | 26/1250 [00:41<30:39,  1.50s/it, loss=6.69]
Epoch 1/4:   2%|▏         | 26/1250 [00:42<30:39,  1.50s/it, loss=6.73]
Epoch 1/4:   2%|▏         | 27/1250 [00:42<30:25,  1.49s/it, loss=6.73]
Epoch 1/4:   2%|▏         | 27/1250 [00:44<30:25,  1.49s/it, loss=5.96]
Epoch 1/4:   2%|▏         | 28/1250 [00:44<30:32,  1.50s/it, loss=5.96]
Epoch 1/4:   2%|▏         | 28/1250 [00:45<30:32,  1.50s/it, loss=6.72]
Epoch 1/4:   2%|▏         | 29/1250 [00:45<32:10,  1.58s/it, loss=6.72]
Epoch 1/4:   2%|▏         | 29/1250 [00:47<32:10,  1.58s/it, loss=6.06]
Epoch 1/4:   2%|▏         | 30/1250 [00:47<31:19,  1.54s/it, loss=6.06]
Epoch 1/4:   2%|▏         | 30/1250 [00:48<31:19,  1.54s/it, loss=6.2] 
Epoch 1/4:   2%|▏         | 31/1250 [00:48<30:53,  1.52s/it, loss=6.2]
Epoch 1/4:   2%|▏         | 31/1250 [00:50<30:53,  1.52s/it, loss=6.2]
Epoch 1/4:   3%|▎         | 32/1250 [00:50<30:57,  1.53s/it, loss=6.2]
Epoch 1/4:   3%|▎         | 32/1250 [00:51<30:57,  1.53s/it, loss=5.98]
Epoch 1/4:   3%|▎         | 33/1250 [00:51<30:30,  1.50s/it, loss=5.98]
Epoch 1/4:   3%|▎         | 33/1250 [00:53<30:30,  1.50s/it, loss=6.13]
Epoch 1/4:   3%|▎         | 34/1250 [00:53<30:39,  1.51s/it, loss=6.13]
Epoch 1/4:   3%|▎         | 34/1250 [00:54<30:39,  1.51s/it, loss=6.32]
Epoch 1/4:   3%|▎         | 35/1250 [00:54<30:41,  1.52s/it, loss=6.32]
Epoch 1/4:   3%|▎         | 35/1250 [00:56<30:41,  1.52s/it, loss=6.29]
Epoch 1/4:   3%|▎         | 36/1250 [00:56<30:09,  1.49s/it, loss=6.29]
Epoch 1/4:   3%|▎         | 36/1250 [00:57<30:09,  1.49s/it, loss=5.94]
Epoch 1/4:   3%|▎         | 37/1250 [00:57<30:12,  1.49s/it, loss=5.94]
Epoch 1/4:   3%|▎         | 37/1250 [00:59<30:12,  1.49s/it, loss=5.67]
Epoch 1/4:   3%|▎         | 38/1250 [00:59<32:19,  1.60s/it, loss=5.67]
Epoch 1/4:   3%|▎         | 38/1250 [01:01<32:19,  1.60s/it, loss=5.8] 
Epoch 1/4:   3%|▎         | 39/1250 [01:01<32:35,  1.62s/it, loss=5.8]
Epoch 1/4:   3%|▎         | 39/1250 [01:02<32:35,  1.62s/it, loss=5.5]
Epoch 1/4:   3%|▎         | 40/1250 [01:02<32:14,  1.60s/it, loss=5.5]
Epoch 1/4:   3%|▎         | 40/1250 [01:04<32:14,  1.60s/it, loss=6.12]
Epoch 1/4:   3%|▎         | 41/1250 [01:04<35:13,  1.75s/it, loss=6.12]
Epoch 1/4:   3%|▎         | 41/1250 [01:06<35:13,  1.75s/it, loss=5.81]
Epoch 1/4:   3%|▎         | 42/1250 [01:06<35:45,  1.78s/it, loss=5.81]
Epoch 1/4:   3%|▎         | 42/1250 [01:08<35:45,  1.78s/it, loss=5.97]
Epoch 1/4:   3%|▎         | 43/1250 [01:08<34:29,  1.71s/it, loss=5.97]
Epoch 1/4:   3%|▎         | 43/1250 [01:09<34:29,  1.71s/it, loss=5.92]
Epoch 1/4:   4%|▎         | 44/1250 [01:09<34:26,  1.71s/it, loss=5.92]
Epoch 1/4:   4%|▎         | 44/1250 [01:11<34:26,  1.71s/it, loss=5.65]
Epoch 1/4:   4%|▎         | 45/1250 [01:11<33:57,  1.69s/it, loss=5.65]
Epoch 1/4:   4%|▎         | 45/1250 [01:13<33:57,  1.69s/it, loss=5.54]
Epoch 1/4:   4%|▎         | 46/1250 [01:13<35:00,  1.74s/it, loss=5.54]
Epoch 1/4:   4%|▎         | 46/1250 [01:14<35:00,  1.74s/it, loss=5.3] 
Epoch 1/4:   4%|▍         | 47/1250 [01:14<33:15,  1.66s/it, loss=5.3]
Epoch 1/4:   4%|▍         | 47/1250 [01:16<33:15,  1.66s/it, loss=5.89]
Epoch 1/4:   4%|▍         | 48/1250 [01:16<33:53,  1.69s/it, loss=5.89]
Epoch 1/4:   4%|▍         | 48/1250 [01:18<33:53,  1.69s/it, loss=5.24]
Epoch 1/4:   4%|▍         | 49/1250 [01:18<33:21,  1.67s/it, loss=5.24]
Epoch 1/4:   4%|▍         | 49/1250 [01:19<33:21,  1.67s/it, loss=5]   
Epoch 1/4:   4%|▍         | 50/1250 [01:19<32:34,  1.63s/it, loss=5]
Epoch 1/4:   4%|▍         | 50/1250 [01:21<32:34,  1.63s/it, loss=5.6]
Epoch 1/4:   4%|▍         | 51/1250 [01:21<34:28,  1.72s/it, loss=5.6]
Epoch 1/4:   4%|▍         | 51/1250 [01:23<34:28,  1.72s/it, loss=5.34]
Epoch 1/4:   4%|▍         | 52/1250 [01:23<32:23,  1.62s/it, loss=5.34]
Epoch 1/4:   4%|▍         | 52/1250 [01:24<32:23,  1.62s/it, loss=5.34]
Epoch 1/4:   4%|▍         | 53/1250 [01:24<32:01,  1.60s/it, loss=5.34]
Epoch 1/4:   4%|▍         | 53/1250 [01:26<32:01,  1.60s/it, loss=5.2] 
Epoch 1/4:   4%|▍         | 54/1250 [01:26<31:35,  1.59s/it, loss=5.2]
Epoch 1/4:   4%|▍         | 54/1250 [01:27<31:35,  1.59s/it, loss=5.41]
Epoch 1/4:   4%|▍         | 55/1250 [01:27<31:03,  1.56s/it, loss=5.41]
Epoch 1/4:   4%|▍         | 55/1250 [01:29<31:03,  1.56s/it, loss=5.34]
Epoch 1/4:   4%|▍         | 56/1250 [01:29<31:42,  1.59s/it, loss=5.34]
Epoch 1/4:   4%|▍         | 56/1250 [01:30<31:42,  1.59s/it, loss=5.54]
Epoch 1/4:   5%|▍         | 57/1250 [01:30<31:10,  1.57s/it, loss=5.54]
Epoch 1/4:   5%|▍         | 57/1250 [01:32<31:10,  1.57s/it, loss=5.24]
Epoch 1/4:   5%|▍         | 58/1250 [01:32<30:52,  1.55s/it, loss=5.24]
Epoch 1/4:   5%|▍         | 58/1250 [01:33<30:52,  1.55s/it, loss=5.55]
Epoch 1/4:   5%|▍         | 59/1250 [01:33<30:14,  1.52s/it, loss=5.55]
Epoch 1/4:   5%|▍         | 59/1250 [01:35<30:14,  1.52s/it, loss=4.88]
Epoch 1/4:   5%|▍         | 60/1250 [01:35<30:29,  1.54s/it, loss=4.88]
Epoch 1/4:   5%|▍         | 60/1250 [01:37<30:29,  1.54s/it, loss=5.5] 
Epoch 1/4:   5%|▍         | 61/1250 [01:37<30:10,  1.52s/it, loss=5.5]
Epoch 1/4:   5%|▍         | 61/1250 [01:38<30:10,  1.52s/it, loss=4.39]
Epoch 1/4:   5%|▍         | 62/1250 [01:38<30:00,  1.52s/it, loss=4.39]
Epoch 1/4:   5%|▍         | 62/1250 [01:40<30:00,  1.52s/it, loss=5.04]
Epoch 1/4:   5%|▌         | 63/1250 [01:40<30:55,  1.56s/it, loss=5.04]
Epoch 1/4:   5%|▌         | 63/1250 [01:41<30:55,  1.56s/it, loss=5.07]
Epoch 1/4:   5%|▌         | 64/1250 [01:41<31:42,  1.60s/it, loss=5.07]
Epoch 1/4:   5%|▌         | 64/1250 [01:43<31:42,  1.60s/it, loss=4.95]
Epoch 1/4:   5%|▌         | 65/1250 [01:43<31:01,  1.57s/it, loss=4.95]
Epoch 1/4:   5%|▌         | 65/1250 [01:44<31:01,  1.57s/it, loss=4.86]
Epoch 1/4:   5%|▌         | 66/1250 [01:44<30:56,  1.57s/it, loss=4.86]
Epoch 1/4:   5%|▌         | 66/1250 [01:46<30:56,  1.57s/it, loss=4.93]
Epoch 1/4:   5%|▌         | 67/1250 [01:46<31:22,  1.59s/it, loss=4.93]
Epoch 1/4:   5%|▌         | 67/1250 [01:48<31:22,  1.59s/it, loss=4.89]
Epoch 1/4:   5%|▌         | 68/1250 [01:48<33:17,  1.69s/it, loss=4.89]
Epoch 1/4:   5%|▌         | 68/1250 [01:50<33:17,  1.69s/it, loss=4.7] 
Epoch 1/4:   6%|▌         | 69/1250 [01:50<35:36,  1.81s/it, loss=4.7]
Epoch 1/4:   6%|▌         | 69/1250 [01:52<35:36,  1.81s/it, loss=4.71]
Epoch 1/4:   6%|▌         | 70/1250 [01:52<34:18,  1.74s/it, loss=4.71]
Epoch 1/4:   6%|▌         | 70/1250 [01:54<34:18,  1.74s/it, loss=4.62]
Epoch 1/4:   6%|▌         | 71/1250 [01:54<34:56,  1.78s/it, loss=4.62]
Epoch 1/4:   6%|▌         | 71/1250 [01:55<34:56,  1.78s/it, loss=4.89]
Epoch 1/4:   6%|▌         | 72/1250 [01:55<34:11,  1.74s/it, loss=4.89]
Epoch 1/4:   6%|▌         | 72/1250 [01:57<34:11,  1.74s/it, loss=4.39]
Epoch 1/4:   6%|▌         | 73/1250 [01:57<34:16,  1.75s/it, loss=4.39]
Epoch 1/4:   6%|▌         | 73/1250 [01:59<34:16,  1.75s/it, loss=4.6] 
Epoch 1/4:   6%|▌         | 74/1250 [01:59<35:10,  1.79s/it, loss=4.6]
Epoch 1/4:   6%|▌         | 74/1250 [02:01<35:10,  1.79s/it, loss=4.23]
Epoch 1/4:   6%|▌         | 75/1250 [02:01<34:47,  1.78s/it, loss=4.23]
Epoch 1/4:   6%|▌         | 75/1250 [02:03<34:47,  1.78s/it, loss=4.45]
Epoch 1/4:   6%|▌         | 76/1250 [02:03<37:33,  1.92s/it, loss=4.45]
Epoch 1/4:   6%|▌         | 76/1250 [02:04<37:33,  1.92s/it, loss=4.76]
Epoch 1/4:   6%|▌         | 77/1250 [02:04<35:10,  1.80s/it, loss=4.76]
Epoch 1/4:   6%|▌         | 77/1250 [02:06<35:10,  1.80s/it, loss=4.4] 
Epoch 1/4:   6%|▌         | 78/1250 [02:06<33:23,  1.71s/it, loss=4.4]
Epoch 1/4:   6%|▌         | 78/1250 [02:08<33:23,  1.71s/it, loss=4.4]
Epoch 1/4:   6%|▋         | 79/1250 [02:08<33:09,  1.70s/it, loss=4.4]
Epoch 1/4:   6%|▋         | 79/1250 [02:09<33:09,  1.70s/it, loss=4.3]
Epoch 1/4:   6%|▋         | 80/1250 [02:09<32:02,  1.64s/it, loss=4.3]
Epoch 1/4:   6%|▋         | 80/1250 [02:11<32:02,  1.64s/it, loss=4.36]
Epoch 1/4:   6%|▋         | 81/1250 [02:11<31:35,  1.62s/it, loss=4.36]
Epoch 1/4:   6%|▋         | 81/1250 [02:13<31:35,  1.62s/it, loss=4.22]
Epoch 1/4:   7%|▋         | 82/1250 [02:13<33:27,  1.72s/it, loss=4.22]
Epoch 1/4:   7%|▋         | 82/1250 [02:14<33:27,  1.72s/it, loss=4.6] 
Epoch 1/4:   7%|▋         | 83/1250 [02:14<32:14,  1.66s/it, loss=4.6]
Epoch 1/4:   7%|▋         | 83/1250 [02:16<32:14,  1.66s/it, loss=3.96]
Epoch 1/4:   7%|▋         | 84/1250 [02:16<33:23,  1.72s/it, loss=3.96]
Epoch 1/4:   7%|▋         | 84/1250 [02:18<33:23,  1.72s/it, loss=4.31]
Epoch 1/4:   7%|▋         | 85/1250 [02:18<35:21,  1.82s/it, loss=4.31]
Epoch 1/4:   7%|▋         | 85/1250 [02:20<35:21,  1.82s/it, loss=4.38]
Epoch 1/4:   7%|▋         | 86/1250 [02:20<35:54,  1.85s/it, loss=4.38]
Epoch 1/4:   7%|▋         | 86/1250 [02:22<35:54,  1.85s/it, loss=4.6] 
Epoch 1/4:   7%|▋         | 87/1250 [02:22<36:24,  1.88s/it, loss=4.6]
Epoch 1/4:   7%|▋         | 87/1250 [02:24<36:24,  1.88s/it, loss=4.18]
Epoch 1/4:   7%|▋         | 88/1250 [02:24<37:21,  1.93s/it, loss=4.18]
Epoch 1/4:   7%|▋         | 88/1250 [02:26<37:21,  1.93s/it, loss=4.29]
Epoch 1/4:   7%|▋         | 89/1250 [02:26<36:18,  1.88s/it, loss=4.29]
Epoch 1/4:   7%|▋         | 89/1250 [02:28<36:18,  1.88s/it, loss=4.11]
Epoch 1/4:   7%|▋         | 90/1250 [02:28<36:42,  1.90s/it, loss=4.11]
Epoch 1/4:   7%|▋         | 90/1250 [02:29<36:42,  1.90s/it, loss=4]   
Epoch 1/4:   7%|▋         | 91/1250 [02:29<35:14,  1.82s/it, loss=4]
Epoch 1/4:   7%|▋         | 91/1250 [02:31<35:14,  1.82s/it, loss=3.77]
Epoch 1/4:   7%|▋         | 92/1250 [02:31<34:19,  1.78s/it, loss=3.77]
Epoch 1/4:   7%|▋         | 92/1250 [02:33<34:19,  1.78s/it, loss=4.1] 
Epoch 1/4:   7%|▋         | 93/1250 [02:33<33:24,  1.73s/it, loss=4.1]
Epoch 1/4:   7%|▋         | 93/1250 [02:35<33:24,  1.73s/it, loss=4.26]
Epoch 1/4:   8%|▊         | 94/1250 [02:35<35:29,  1.84s/it, loss=4.26]
Epoch 1/4:   8%|▊         | 94/1250 [02:37<35:29,  1.84s/it, loss=3.86]
Epoch 1/4:   8%|▊         | 95/1250 [02:37<35:25,  1.84s/it, loss=3.86]
Epoch 1/4:   8%|▊         | 95/1250 [02:38<35:25,  1.84s/it, loss=4.75]
Epoch 1/4:   8%|▊         | 96/1250 [02:38<33:50,  1.76s/it, loss=4.75]
Epoch 1/4:   8%|▊         | 96/1250 [02:40<33:50,  1.76s/it, loss=3.94]
Epoch 1/4:   8%|▊         | 97/1250 [02:40<35:09,  1.83s/it, loss=3.94]
Epoch 1/4:   8%|▊         | 97/1250 [02:42<35:09,  1.83s/it, loss=3.52]
Epoch 1/4:   8%|▊         | 98/1250 [02:42<35:50,  1.87s/it, loss=3.52]
Epoch 1/4:   8%|▊         | 98/1250 [02:44<35:50,  1.87s/it, loss=3.65]
Epoch 1/4:   8%|▊         | 99/1250 [02:44<36:16,  1.89s/it, loss=3.65]
Epoch 1/4:   8%|▊         | 99/1250 [02:46<36:16,  1.89s/it, loss=3.85]
Epoch 1/4:   8%|▊         | 100/1250 [02:46<34:32,  1.80s/it, loss=3.85]
Epoch 1/4:   8%|▊         | 100/1250 [02:47<34:32,  1.80s/it, loss=3.71]
Epoch 1/4:   8%|▊         | 101/1250 [02:47<34:27,  1.80s/it, loss=3.71]
Epoch 1/4:   8%|▊         | 101/1250 [02:49<34:27,  1.80s/it, loss=3.48]
Epoch 1/4:   8%|▊         | 102/1250 [02:49<34:49,  1.82s/it, loss=3.48]
Epoch 1/4:   8%|▊         | 102/1250 [02:51<34:49,  1.82s/it, loss=3.56]
Epoch 1/4:   8%|▊         | 103/1250 [02:51<33:32,  1.75s/it, loss=3.56]
Epoch 1/4:   8%|▊         | 103/1250 [02:52<33:32,  1.75s/it, loss=3.65]
Epoch 1/4:   8%|▊         | 104/1250 [02:52<32:42,  1.71s/it, loss=3.65]
Epoch 1/4:   8%|▊         | 104/1250 [02:54<32:42,  1.71s/it, loss=3.9] 
Epoch 1/4:   8%|▊         | 105/1250 [02:54<32:18,  1.69s/it, loss=3.9]
Epoch 1/4:   8%|▊         | 105/1250 [02:56<32:18,  1.69s/it, loss=3.16]
Epoch 1/4:   8%|▊         | 106/1250 [02:56<33:58,  1.78s/it, loss=3.16]
Epoch 1/4:   8%|▊         | 106/1250 [02:58<33:58,  1.78s/it, loss=3.42]
Epoch 1/4:   9%|▊         | 107/1250 [02:58<31:59,  1.68s/it, loss=3.42]
Epoch 1/4:   9%|▊         | 107/1250 [02:59<31:59,  1.68s/it, loss=3.34]
Epoch 1/4:   9%|▊         | 108/1250 [02:59<31:31,  1.66s/it, loss=3.34]
Epoch 1/4:   9%|▊         | 108/1250 [03:01<31:31,  1.66s/it, loss=3.91]
Epoch 1/4:   9%|▊         | 109/1250 [03:01<33:12,  1.75s/it, loss=3.91]
Epoch 1/4:   9%|▊         | 109/1250 [03:03<33:12,  1.75s/it, loss=3.5] 
Epoch 1/4:   9%|▉         | 110/1250 [03:04<36:56,  1.94s/it, loss=3.5]
Epoch 1/4:   9%|▉         | 110/1250 [03:06<36:56,  1.94s/it, loss=3.45]
Epoch 1/4:   9%|▉         | 111/1250 [03:06<37:33,  1.98s/it, loss=3.45]
Epoch 1/4:   9%|▉         | 111/1250 [03:07<37:33,  1.98s/it, loss=3.28]
Epoch 1/4:   9%|▉         | 112/1250 [03:07<35:39,  1.88s/it, loss=3.28]
Epoch 1/4:   9%|▉         | 112/1250 [03:09<35:39,  1.88s/it, loss=3.37]
Epoch 1/4:   9%|▉         | 113/1250 [03:09<36:14,  1.91s/it, loss=3.37]
Epoch 1/4:   9%|▉         | 113/1250 [03:11<36:14,  1.91s/it, loss=2.96]
Epoch 1/4:   9%|▉         | 114/1250 [03:11<35:01,  1.85s/it, loss=2.96]
Epoch 1/4:   9%|▉         | 114/1250 [03:13<35:01,  1.85s/it, loss=3.27]
Epoch 1/4:   9%|▉         | 115/1250 [03:13<35:51,  1.90s/it, loss=3.27]
Epoch 1/4:   9%|▉         | 115/1250 [03:14<35:51,  1.90s/it, loss=3.17]
Epoch 1/4:   9%|▉         | 116/1250 [03:14<34:03,  1.80s/it, loss=3.17]
Epoch 1/4:   9%|▉         | 116/1250 [03:16<34:03,  1.80s/it, loss=3.02]
Epoch 1/4:   9%|▉         | 117/1250 [03:16<32:48,  1.74s/it, loss=3.02]
Epoch 1/4:   9%|▉         | 117/1250 [03:18<32:48,  1.74s/it, loss=3.15]
Epoch 1/4:   9%|▉         | 118/1250 [03:18<33:39,  1.78s/it, loss=3.15]
Epoch 1/4:   9%|▉         | 118/1250 [03:20<33:39,  1.78s/it, loss=3.27]
Epoch 1/4:  10%|▉         | 119/1250 [03:20<32:21,  1.72s/it, loss=3.27]
Epoch 1/4:  10%|▉         | 119/1250 [03:21<32:21,  1.72s/it, loss=3.33]
Epoch 1/4:  10%|▉         | 120/1250 [03:21<31:57,  1.70s/it, loss=3.33]
Epoch 1/4:  10%|▉         | 120/1250 [03:23<31:57,  1.70s/it, loss=3.12]
Epoch 1/4:  10%|▉         | 121/1250 [03:23<32:06,  1.71s/it, loss=3.12]
Epoch 1/4:  10%|▉         | 121/1250 [03:25<32:06,  1.71s/it, loss=3.14]
Epoch 1/4:  10%|▉         | 122/1250 [03:25<31:55,  1.70s/it, loss=3.14]
Epoch 1/4:  10%|▉         | 122/1250 [03:27<31:55,  1.70s/it, loss=3.48]
Epoch 1/4:  10%|▉         | 123/1250 [03:27<34:25,  1.83s/it, loss=3.48]
Epoch 1/4:  10%|▉         | 123/1250 [03:28<34:25,  1.83s/it, loss=2.92]
Epoch 1/4:  10%|▉         | 124/1250 [03:28<33:19,  1.78s/it, loss=2.92]
Epoch 1/4:  10%|▉         | 124/1250 [03:30<33:19,  1.78s/it, loss=2.87]
Epoch 1/4:  10%|█         | 125/1250 [03:30<34:32,  1.84s/it, loss=2.87]
Epoch 1/4:  10%|█         | 125/1250 [03:32<34:32,  1.84s/it, loss=3.27]
Epoch 1/4:  10%|█         | 126/1250 [03:32<32:59,  1.76s/it, loss=3.27]
Epoch 1/4:  10%|█         | 126/1250 [03:34<32:59,  1.76s/it, loss=3.14]
Epoch 1/4:  10%|█         | 127/1250 [03:34<33:14,  1.78s/it, loss=3.14]
Epoch 1/4:  10%|█         | 127/1250 [03:35<33:14,  1.78s/it, loss=2.72]
Epoch 1/4:  10%|█         | 128/1250 [03:35<32:28,  1.74s/it, loss=2.72]
Epoch 1/4:  10%|█         | 128/1250 [03:37<32:28,  1.74s/it, loss=2.56]
Epoch 1/4:  10%|█         | 129/1250 [03:37<31:33,  1.69s/it, loss=2.56]
Epoch 1/4:  10%|█         | 129/1250 [03:39<31:33,  1.69s/it, loss=3.4] 
Epoch 1/4:  10%|█         | 130/1250 [03:39<31:28,  1.69s/it, loss=3.4]
Epoch 1/4:  10%|█         | 130/1250 [03:40<31:28,  1.69s/it, loss=2.89]
Epoch 1/4:  10%|█         | 131/1250 [03:40<30:54,  1.66s/it, loss=2.89]
Epoch 1/4:  10%|█         | 131/1250 [03:42<30:54,  1.66s/it, loss=3.03]
Epoch 1/4:  11%|█         | 132/1250 [03:42<31:41,  1.70s/it, loss=3.03]
Epoch 1/4:  11%|█         | 132/1250 [03:44<31:41,  1.70s/it, loss=2.81]
Epoch 1/4:  11%|█         | 133/1250 [03:44<31:54,  1.71s/it, loss=2.81]
Epoch 1/4:  11%|█         | 133/1250 [03:45<31:54,  1.71s/it, loss=2.68]
Epoch 1/4:  11%|█         | 134/1250 [03:45<30:37,  1.65s/it, loss=2.68]
Epoch 1/4:  11%|█         | 134/1250 [03:47<30:37,  1.65s/it, loss=2.79]
Epoch 1/4:  11%|█         | 135/1250 [03:47<30:29,  1.64s/it, loss=2.79]
Epoch 1/4:  11%|█         | 135/1250 [03:48<30:29,  1.64s/it, loss=2.56]
Epoch 1/4:  11%|█         | 136/1250 [03:48<29:57,  1.61s/it, loss=2.56]
Epoch 1/4:  11%|█         | 136/1250 [03:51<29:57,  1.61s/it, loss=2.84]
Epoch 1/4:  11%|█         | 137/1250 [03:51<33:31,  1.81s/it, loss=2.84]
Epoch 1/4:  11%|█         | 137/1250 [03:52<33:31,  1.81s/it, loss=2.8] 
Epoch 1/4:  11%|█         | 138/1250 [03:52<31:50,  1.72s/it, loss=2.8]
Epoch 1/4:  11%|█         | 138/1250 [03:54<31:50,  1.72s/it, loss=2.42]
Epoch 1/4:  11%|█         | 139/1250 [03:54<31:26,  1.70s/it, loss=2.42]
Epoch 1/4:  11%|█         | 139/1250 [03:56<31:26,  1.70s/it, loss=2.5] 
Epoch 1/4:  11%|█         | 140/1250 [03:56<32:12,  1.74s/it, loss=2.5]
Epoch 1/4:  11%|█         | 140/1250 [03:57<32:12,  1.74s/it, loss=2.34]
Epoch 1/4:  11%|█▏        | 141/1250 [03:57<30:45,  1.66s/it, loss=2.34]
Epoch 1/4:  11%|█▏        | 141/1250 [03:59<30:45,  1.66s/it, loss=2.35]
Epoch 1/4:  11%|█▏        | 142/1250 [03:59<30:51,  1.67s/it, loss=2.35]
Epoch 1/4:  11%|█▏        | 142/1250 [04:01<30:51,  1.67s/it, loss=2.35]
Epoch 1/4:  11%|█▏        | 143/1250 [04:01<30:59,  1.68s/it, loss=2.35]
Epoch 1/4:  11%|█▏        | 143/1250 [04:02<30:59,  1.68s/it, loss=2.58]
Epoch 1/4:  12%|█▏        | 144/1250 [04:03<32:13,  1.75s/it, loss=2.58]
Epoch 1/4:  12%|█▏        | 144/1250 [04:04<32:13,  1.75s/it, loss=2.31]
Epoch 1/4:  12%|█▏        | 145/1250 [04:04<32:20,  1.76s/it, loss=2.31]
Epoch 1/4:  12%|█▏        | 145/1250 [04:06<32:20,  1.76s/it, loss=2.15]
Epoch 1/4:  12%|█▏        | 146/1250 [04:06<31:59,  1.74s/it, loss=2.15]
Epoch 1/4:  12%|█▏        | 146/1250 [04:08<31:59,  1.74s/it, loss=2.27]
Epoch 1/4:  12%|█▏        | 147/1250 [04:08<31:05,  1.69s/it, loss=2.27]
Epoch 1/4:  12%|█▏        | 147/1250 [04:09<31:05,  1.69s/it, loss=2.68]
Epoch 1/4:  12%|█▏        | 148/1250 [04:09<30:35,  1.67s/it, loss=2.68]
Epoch 1/4:  12%|█▏        | 148/1250 [04:11<30:35,  1.67s/it, loss=2.24]
Epoch 1/4:  12%|█▏        | 149/1250 [04:11<32:14,  1.76s/it, loss=2.24]
Epoch 1/4:  12%|█▏        | 149/1250 [04:13<32:14,  1.76s/it, loss=1.87]
Epoch 1/4:  12%|█▏        | 150/1250 [04:13<31:30,  1.72s/it, loss=1.87]
Epoch 1/4:  12%|█▏        | 150/1250 [04:14<31:30,  1.72s/it, loss=2.23]
Epoch 1/4:  12%|█▏        | 151/1250 [04:14<30:44,  1.68s/it, loss=2.23]
Epoch 1/4:  12%|█▏        | 151/1250 [04:16<30:44,  1.68s/it, loss=2.36]
Epoch 1/4:  12%|█▏        | 152/1250 [04:16<30:18,  1.66s/it, loss=2.36]
Epoch 1/4:  12%|█▏        | 152/1250 [04:18<30:18,  1.66s/it, loss=2.05]
Epoch 1/4:  12%|█▏        | 153/1250 [04:18<29:46,  1.63s/it, loss=2.05]
Epoch 1/4:  12%|█▏        | 153/1250 [04:19<29:46,  1.63s/it, loss=1.85]
Epoch 1/4:  12%|█▏        | 154/1250 [04:19<29:59,  1.64s/it, loss=1.85]
Epoch 1/4:  12%|█▏        | 154/1250 [04:21<29:59,  1.64s/it, loss=2.04]
Epoch 1/4:  12%|█▏        | 155/1250 [04:21<32:13,  1.77s/it, loss=2.04]
Epoch 1/4:  12%|█▏        | 155/1250 [04:23<32:13,  1.77s/it, loss=2.21]
Epoch 1/4:  12%|█▏        | 156/1250 [04:23<32:17,  1.77s/it, loss=2.21]
Epoch 1/4:  12%|█▏        | 156/1250 [04:25<32:17,  1.77s/it, loss=2.15]
Epoch 1/4:  13%|█▎        | 157/1250 [04:25<30:40,  1.68s/it, loss=2.15]
Epoch 1/4:  13%|█▎        | 157/1250 [04:26<30:40,  1.68s/it, loss=2.13]
Epoch 1/4:  13%|█▎        | 158/1250 [04:26<30:16,  1.66s/it, loss=2.13]
Epoch 1/4:  13%|█▎        | 158/1250 [04:28<30:16,  1.66s/it, loss=2.43]
Epoch 1/4:  13%|█▎        | 159/1250 [04:28<29:03,  1.60s/it, loss=2.43]
Epoch 1/4:  13%|█▎        | 159/1250 [04:29<29:03,  1.60s/it, loss=2.03]
Epoch 1/4:  13%|█▎        | 160/1250 [04:29<29:06,  1.60s/it, loss=2.03]
Epoch 1/4:  13%|█▎        | 160/1250 [04:31<29:06,  1.60s/it, loss=1.86]
Epoch 1/4:  13%|█▎        | 161/1250 [04:31<28:45,  1.58s/it, loss=1.86]
Epoch 1/4:  13%|█▎        | 161/1250 [04:32<28:45,  1.58s/it, loss=1.96]
Epoch 1/4:  13%|█▎        | 162/1250 [04:32<28:03,  1.55s/it, loss=1.96]
Epoch 1/4:  13%|█▎        | 162/1250 [04:34<28:03,  1.55s/it, loss=2.02]
Epoch 1/4:  13%|█▎        | 163/1250 [04:34<28:17,  1.56s/it, loss=2.02]
Epoch 1/4:  13%|█▎        | 163/1250 [04:35<28:17,  1.56s/it, loss=1.84]
Epoch 1/4:  13%|█▎        | 164/1250 [04:35<28:02,  1.55s/it, loss=1.84]
Epoch 1/4:  13%|█▎        | 164/1250 [04:37<28:02,  1.55s/it, loss=1.84]
Epoch 1/4:  13%|█▎        | 165/1250 [04:37<27:41,  1.53s/it, loss=1.84]
Epoch 1/4:  13%|█▎        | 165/1250 [04:38<27:41,  1.53s/it, loss=2.06]
Epoch 1/4:  13%|█▎        | 166/1250 [04:38<27:52,  1.54s/it, loss=2.06]
Epoch 1/4:  13%|█▎        | 166/1250 [04:40<27:52,  1.54s/it, loss=1.93]
Epoch 1/4:  13%|█▎        | 167/1250 [04:40<28:25,  1.57s/it, loss=1.93]
Epoch 1/4:  13%|█▎        | 167/1250 [04:42<28:25,  1.57s/it, loss=2.07]
Epoch 1/4:  13%|█▎        | 168/1250 [04:42<30:41,  1.70s/it, loss=2.07]
Epoch 1/4:  13%|█▎        | 168/1250 [04:43<30:41,  1.70s/it, loss=1.64]
Epoch 1/4:  14%|█▎        | 169/1250 [04:43<29:01,  1.61s/it, loss=1.64]
Epoch 1/4:  14%|█▎        | 169/1250 [04:45<29:01,  1.61s/it, loss=2.2] 
Epoch 1/4:  14%|█▎        | 170/1250 [04:45<28:49,  1.60s/it, loss=2.2]
Epoch 1/4:  14%|█▎        | 170/1250 [04:47<28:49,  1.60s/it, loss=1.64]
Epoch 1/4:  14%|█▎        | 171/1250 [04:47<28:47,  1.60s/it, loss=1.64]
Epoch 1/4:  14%|█▎        | 171/1250 [04:48<28:47,  1.60s/it, loss=1.64]
Epoch 1/4:  14%|█▍        | 172/1250 [04:48<28:31,  1.59s/it, loss=1.64]
Epoch 1/4:  14%|█▍        | 172/1250 [04:50<28:31,  1.59s/it, loss=1.59]
Epoch 1/4:  14%|█▍        | 173/1250 [04:50<28:19,  1.58s/it, loss=1.59]
Epoch 1/4:  14%|█▍        | 173/1250 [04:51<28:19,  1.58s/it, loss=1.7] 
Epoch 1/4:  14%|█▍        | 174/1250 [04:51<27:59,  1.56s/it, loss=1.7]
Epoch 1/4:  14%|█▍        | 174/1250 [04:53<27:59,  1.56s/it, loss=1.67]
Epoch 1/4:  14%|█▍        | 175/1250 [04:53<27:22,  1.53s/it, loss=1.67]
Epoch 1/4:  14%|█▍        | 175/1250 [04:54<27:22,  1.53s/it, loss=1.81]
Epoch 1/4:  14%|█▍        | 176/1250 [04:54<27:30,  1.54s/it, loss=1.81]
Epoch 1/4:  14%|█▍        | 176/1250 [04:56<27:30,  1.54s/it, loss=1.6] 
Epoch 1/4:  14%|█▍        | 177/1250 [04:56<27:26,  1.53s/it, loss=1.6]
Epoch 1/4:  14%|█▍        | 177/1250 [04:57<27:26,  1.53s/it, loss=1.62]
Epoch 1/4:  14%|█▍        | 178/1250 [04:57<27:01,  1.51s/it, loss=1.62]
Epoch 1/4:  14%|█▍        | 178/1250 [04:59<27:01,  1.51s/it, loss=2.07]
Epoch 1/4:  14%|█▍        | 179/1250 [04:59<27:01,  1.51s/it, loss=2.07]
Epoch 1/4:  14%|█▍        | 179/1250 [05:00<27:01,  1.51s/it, loss=1.99]
Epoch 1/4:  14%|█▍        | 180/1250 [05:00<26:54,  1.51s/it, loss=1.99]
Epoch 1/4:  14%|█▍        | 180/1250 [05:02<26:54,  1.51s/it, loss=1.68]
Epoch 1/4:  14%|█▍        | 181/1250 [05:02<26:55,  1.51s/it, loss=1.68]
Epoch 1/4:  14%|█▍        | 181/1250 [05:03<26:55,  1.51s/it, loss=1.6] 
Epoch 1/4:  15%|█▍        | 182/1250 [05:03<26:54,  1.51s/it, loss=1.6]
Epoch 1/4:  15%|█▍        | 182/1250 [05:05<26:54,  1.51s/it, loss=1.32]
Epoch 1/4:  15%|█▍        | 183/1250 [05:05<26:41,  1.50s/it, loss=1.32]
Epoch 1/4:  15%|█▍        | 183/1250 [05:06<26:41,  1.50s/it, loss=1.56]
Epoch 1/4:  15%|█▍        | 184/1250 [05:06<26:34,  1.50s/it, loss=1.56]
Epoch 1/4:  15%|█▍        | 184/1250 [05:08<26:34,  1.50s/it, loss=1.61]
Epoch 1/4:  15%|█▍        | 185/1250 [05:08<26:33,  1.50s/it, loss=1.61]
Epoch 1/4:  15%|█▍        | 185/1250 [05:09<26:33,  1.50s/it, loss=1.27]
Epoch 1/4:  15%|█▍        | 186/1250 [05:09<27:29,  1.55s/it, loss=1.27]
Epoch 1/4:  15%|█▍        | 186/1250 [05:11<27:29,  1.55s/it, loss=1.94]
Epoch 1/4:  15%|█▍        | 187/1250 [05:11<27:38,  1.56s/it, loss=1.94]
Epoch 1/4:  15%|█▍        | 187/1250 [05:13<27:38,  1.56s/it, loss=1.33]
Epoch 1/4:  15%|█▌        | 188/1250 [05:13<28:54,  1.63s/it, loss=1.33]
Epoch 1/4:  15%|█▌        | 188/1250 [05:15<28:54,  1.63s/it, loss=1.36]
Epoch 1/4:  15%|█▌        | 189/1250 [05:15<32:06,  1.82s/it, loss=1.36]
Epoch 1/4:  15%|█▌        | 189/1250 [05:16<32:06,  1.82s/it, loss=1.7] 
Epoch 1/4:  15%|█▌        | 190/1250 [05:16<29:49,  1.69s/it, loss=1.7]
Epoch 1/4:  15%|█▌        | 190/1250 [05:18<29:49,  1.69s/it, loss=1.72]
Epoch 1/4:  15%|█▌        | 191/1250 [05:18<29:08,  1.65s/it, loss=1.72]
Epoch 1/4:  15%|█▌        | 191/1250 [05:20<29:08,  1.65s/it, loss=1.43]
Epoch 1/4:  15%|█▌        | 192/1250 [05:20<29:28,  1.67s/it, loss=1.43]
Epoch 1/4:  15%|█▌        | 192/1250 [05:21<29:28,  1.67s/it, loss=1.42]
Epoch 1/4:  15%|█▌        | 193/1250 [05:21<29:52,  1.70s/it, loss=1.42]
Epoch 1/4:  15%|█▌        | 193/1250 [05:23<29:52,  1.70s/it, loss=1.42]
Epoch 1/4:  16%|█▌        | 194/1250 [05:23<28:02,  1.59s/it, loss=1.42]
Epoch 1/4:  16%|█▌        | 194/1250 [05:24<28:02,  1.59s/it, loss=1.23]
Epoch 1/4:  16%|█▌        | 195/1250 [05:24<28:07,  1.60s/it, loss=1.23]
Epoch 1/4:  16%|█▌        | 195/1250 [05:26<28:07,  1.60s/it, loss=1.39]
Epoch 1/4:  16%|█▌        | 196/1250 [05:26<28:04,  1.60s/it, loss=1.39]
Epoch 1/4:  16%|█▌        | 196/1250 [05:28<28:04,  1.60s/it, loss=1.3] 
Epoch 1/4:  16%|█▌        | 197/1250 [05:28<28:28,  1.62s/it, loss=1.3]
Epoch 1/4:  16%|█▌        | 197/1250 [05:30<28:28,  1.62s/it, loss=1.11]
Epoch 1/4:  16%|█▌        | 198/1250 [05:30<32:25,  1.85s/it, loss=1.11]
Epoch 1/4:  16%|█▌        | 198/1250 [05:31<32:25,  1.85s/it, loss=1.53]
Epoch 1/4:  16%|█▌        | 199/1250 [05:31<29:02,  1.66s/it, loss=1.53]
Epoch 1/4:  16%|█▌        | 199/1250 [05:34<29:02,  1.66s/it, loss=1.44]
Epoch 1/4:  16%|█▌        | 200/1250 [05:34<33:30,  1.91s/it, loss=1.44]
Epoch 1/4:  16%|█▌        | 200/1250 [05:37<33:30,  1.91s/it, loss=1.17]
Epoch 1/4:  16%|█▌        | 201/1250 [05:37<38:43,  2.21s/it, loss=1.17]
Epoch 1/4:  16%|█▌        | 201/1250 [05:40<38:43,  2.21s/it, loss=1.19]
Epoch 1/4:  16%|█▌        | 202/1250 [05:40<42:21,  2.42s/it, loss=1.19]
Epoch 1/4:  16%|█▌        | 202/1250 [05:42<42:21,  2.42s/it, loss=1.25]
Epoch 1/4:  16%|█▌        | 203/1250 [05:42<43:40,  2.50s/it, loss=1.25]
Epoch 1/4:  16%|█▌        | 203/1250 [05:44<43:40,  2.50s/it, loss=1.7] 
Epoch 1/4:  16%|█▋        | 204/1250 [05:44<38:31,  2.21s/it, loss=1.7]
Epoch 1/4:  16%|█▋        | 204/1250 [05:46<38:31,  2.21s/it, loss=1.04]
Epoch 1/4:  16%|█▋        | 205/1250 [05:46<40:04,  2.30s/it, loss=1.04]
Epoch 1/4:  16%|█▋        | 205/1250 [05:49<40:04,  2.30s/it, loss=1.29]
Epoch 1/4:  16%|█▋        | 206/1250 [05:49<40:47,  2.34s/it, loss=1.29]
Epoch 1/4:  16%|█▋        | 206/1250 [05:51<40:47,  2.34s/it, loss=1.35]
Epoch 1/4:  17%|█▋        | 207/1250 [05:51<42:28,  2.44s/it, loss=1.35]
Epoch 1/4:  17%|█▋        | 207/1250 [05:53<42:28,  2.44s/it, loss=1.56]
Epoch 1/4:  17%|█▋        | 208/1250 [05:53<35:49,  2.06s/it, loss=1.56]
Epoch 1/4:  17%|█▋        | 208/1250 [05:54<35:49,  2.06s/it, loss=1.22]
Epoch 1/4:  17%|█▋        | 209/1250 [05:54<33:38,  1.94s/it, loss=1.22]
Epoch 1/4:  17%|█▋        | 209/1250 [05:57<33:38,  1.94s/it, loss=1.65]
Epoch 1/4:  17%|█▋        | 210/1250 [05:57<36:21,  2.10s/it, loss=1.65]
Epoch 1/4:  17%|█▋        | 210/1250 [05:58<36:21,  2.10s/it, loss=1.13]
Epoch 1/4:  17%|█▋        | 211/1250 [05:58<32:45,  1.89s/it, loss=1.13]
Epoch 1/4:  17%|█▋        | 211/1250 [06:00<32:45,  1.89s/it, loss=1.34]
Epoch 1/4:  17%|█▋        | 212/1250 [06:00<31:10,  1.80s/it, loss=1.34]
Epoch 1/4:  17%|█▋        | 212/1250 [06:02<31:10,  1.80s/it, loss=1.19]
Epoch 1/4:  17%|█▋        | 213/1250 [06:02<31:33,  1.83s/it, loss=1.19]
Epoch 1/4:  17%|█▋        | 213/1250 [06:03<31:33,  1.83s/it, loss=1.25]
Epoch 1/4:  17%|█▋        | 214/1250 [06:03<30:17,  1.75s/it, loss=1.25]
Epoch 1/4:  17%|█▋        | 214/1250 [06:06<30:17,  1.75s/it, loss=1.14]
Epoch 1/4:  17%|█▋        | 215/1250 [06:06<36:02,  2.09s/it, loss=1.14]
Epoch 1/4:  17%|█▋        | 215/1250 [06:09<36:02,  2.09s/it, loss=1.22]
Epoch 1/4:  17%|█▋        | 216/1250 [06:09<42:05,  2.44s/it, loss=1.22]
Epoch 1/4:  17%|█▋        | 216/1250 [06:13<42:05,  2.44s/it, loss=1.41]
Epoch 1/4:  17%|█▋        | 217/1250 [06:13<46:13,  2.68s/it, loss=1.41]
Epoch 1/4:  17%|█▋        | 217/1250 [06:15<46:13,  2.68s/it, loss=1.21]
Epoch 1/4:  17%|█▋        | 218/1250 [06:15<46:22,  2.70s/it, loss=1.21]
Epoch 1/4:  17%|█▋        | 218/1250 [06:18<46:22,  2.70s/it, loss=0.994]
Epoch 1/4:  18%|█▊        | 219/1250 [06:18<46:11,  2.69s/it, loss=0.994]
Epoch 1/4:  18%|█▊        | 219/1250 [06:21<46:11,  2.69s/it, loss=1.11] 
Epoch 1/4:  18%|█▊        | 220/1250 [06:21<45:47,  2.67s/it, loss=1.11]
Epoch 1/4:  18%|█▊        | 220/1250 [06:22<45:47,  2.67s/it, loss=0.894]
Epoch 1/4:  18%|█▊        | 221/1250 [06:22<40:57,  2.39s/it, loss=0.894]
Epoch 1/4:  18%|█▊        | 221/1250 [06:25<40:57,  2.39s/it, loss=1.12] 
Epoch 1/4:  18%|█▊        | 222/1250 [06:25<40:25,  2.36s/it, loss=1.12]
Epoch 1/4:  18%|█▊        | 222/1250 [06:26<40:25,  2.36s/it, loss=1.11]
Epoch 1/4:  18%|█▊        | 223/1250 [06:26<36:40,  2.14s/it, loss=1.11]
Epoch 1/4:  18%|█▊        | 223/1250 [06:29<36:40,  2.14s/it, loss=1.71]
Epoch 1/4:  18%|█▊        | 224/1250 [06:29<37:07,  2.17s/it, loss=1.71]
Epoch 1/4:  18%|█▊        | 224/1250 [06:30<37:07,  2.17s/it, loss=1.15]
Epoch 1/4:  18%|█▊        | 225/1250 [06:30<35:26,  2.07s/it, loss=1.15]
Epoch 1/4:  18%|█▊        | 225/1250 [06:32<35:26,  2.07s/it, loss=1.16]
Epoch 1/4:  18%|█▊        | 226/1250 [06:32<34:02,  1.99s/it, loss=1.16]
Epoch 1/4:  18%|█▊        | 226/1250 [06:35<34:02,  1.99s/it, loss=0.943]
Epoch 1/4:  18%|█▊        | 227/1250 [06:35<36:53,  2.16s/it, loss=0.943]
Epoch 1/4:  18%|█▊        | 227/1250 [06:36<36:53,  2.16s/it, loss=0.825]
Epoch 1/4:  18%|█▊        | 228/1250 [06:36<33:22,  1.96s/it, loss=0.825]
Epoch 1/4:  18%|█▊        | 228/1250 [06:38<33:22,  1.96s/it, loss=0.762]
Epoch 1/4:  18%|█▊        | 229/1250 [06:38<33:15,  1.95s/it, loss=0.762]
Epoch 1/4:  18%|█▊        | 229/1250 [06:40<33:15,  1.95s/it, loss=1.38] 
Epoch 1/4:  18%|█▊        | 230/1250 [06:40<34:51,  2.05s/it, loss=1.38]
Epoch 1/4:  18%|█▊        | 230/1250 [06:42<34:51,  2.05s/it, loss=1.34]
Epoch 1/4:  18%|█▊        | 231/1250 [06:42<33:47,  1.99s/it, loss=1.34]
Epoch 1/4:  18%|█▊        | 231/1250 [06:45<33:47,  1.99s/it, loss=1.04]
Epoch 1/4:  19%|█▊        | 232/1250 [06:45<35:01,  2.06s/it, loss=1.04]
Epoch 1/4:  19%|█▊        | 232/1250 [06:46<35:01,  2.06s/it, loss=0.826]
Epoch 1/4:  19%|█▊        | 233/1250 [06:46<32:24,  1.91s/it, loss=0.826]
Epoch 1/4:  19%|█▊        | 233/1250 [06:48<32:24,  1.91s/it, loss=1.09] 
Epoch 1/4:  19%|█▊        | 234/1250 [06:48<31:11,  1.84s/it, loss=1.09]
Epoch 1/4:  19%|█▊        | 234/1250 [06:50<31:11,  1.84s/it, loss=1.31]
Epoch 1/4:  19%|█▉        | 235/1250 [06:50<32:44,  1.94s/it, loss=1.31]
Epoch 1/4:  19%|█▉        | 235/1250 [06:51<32:44,  1.94s/it, loss=1.76]
Epoch 1/4:  19%|█▉        | 236/1250 [06:51<29:51,  1.77s/it, loss=1.76]
Epoch 1/4:  19%|█▉        | 236/1250 [06:53<29:51,  1.77s/it, loss=0.966]
Epoch 1/4:  19%|█▉        | 237/1250 [06:53<30:07,  1.78s/it, loss=0.966]
Epoch 1/4:  19%|█▉        | 237/1250 [06:55<30:07,  1.78s/it, loss=1.11] 
Epoch 1/4:  19%|█▉        | 238/1250 [06:55<31:58,  1.90s/it, loss=1.11]
Epoch 1/4:  19%|█▉        | 238/1250 [06:57<31:58,  1.90s/it, loss=1.15]
Epoch 1/4:  19%|█▉        | 239/1250 [06:57<31:22,  1.86s/it, loss=1.15]
Epoch 1/4:  19%|█▉        | 239/1250 [06:59<31:22,  1.86s/it, loss=1.12]
Epoch 1/4:  19%|█▉        | 240/1250 [06:59<32:17,  1.92s/it, loss=1.12]
Epoch 1/4:  19%|█▉        | 240/1250 [07:01<32:17,  1.92s/it, loss=1.35]
Epoch 1/4:  19%|█▉        | 241/1250 [07:01<30:38,  1.82s/it, loss=1.35]
Epoch 1/4:  19%|█▉        | 241/1250 [07:03<30:38,  1.82s/it, loss=1.13]
Epoch 1/4:  19%|█▉        | 242/1250 [07:03<32:35,  1.94s/it, loss=1.13]
Epoch 1/4:  19%|█▉        | 242/1250 [07:05<32:35,  1.94s/it, loss=1.27]
Epoch 1/4:  19%|█▉        | 243/1250 [07:05<33:40,  2.01s/it, loss=1.27]
Epoch 1/4:  19%|█▉        | 243/1250 [07:07<33:40,  2.01s/it, loss=1.24]
Epoch 1/4:  20%|█▉        | 244/1250 [07:07<33:11,  1.98s/it, loss=1.24]
Epoch 1/4:  20%|█▉        | 244/1250 [07:09<33:11,  1.98s/it, loss=0.923]
Epoch 1/4:  20%|█▉        | 245/1250 [07:09<34:15,  2.05s/it, loss=0.923]
Epoch 1/4:  20%|█▉        | 245/1250 [07:11<34:15,  2.05s/it, loss=1.14] 
Epoch 1/4:  20%|█▉        | 246/1250 [07:11<32:19,  1.93s/it, loss=1.14]
Epoch 1/4:  20%|█▉        | 246/1250 [07:13<32:19,  1.93s/it, loss=1.02]
Epoch 1/4:  20%|█▉        | 247/1250 [07:13<34:41,  2.08s/it, loss=1.02]
Epoch 1/4:  20%|█▉        | 247/1250 [07:15<34:41,  2.08s/it, loss=0.899]
Epoch 1/4:  20%|█▉        | 248/1250 [07:15<31:55,  1.91s/it, loss=0.899]
Epoch 1/4:  20%|█▉        | 248/1250 [07:17<31:55,  1.91s/it, loss=1.45] 
Epoch 1/4:  20%|█▉        | 249/1250 [07:17<35:15,  2.11s/it, loss=1.45]
Epoch 1/4:  20%|█▉        | 249/1250 [07:19<35:15,  2.11s/it, loss=1.11]
Epoch 1/4:  20%|██        | 250/1250 [07:19<31:20,  1.88s/it, loss=1.11]
Epoch 1/4:  20%|██        | 250/1250 [07:21<31:20,  1.88s/it, loss=1.36]
Epoch 1/4:  20%|██        | 251/1250 [07:21<34:48,  2.09s/it, loss=1.36]
Epoch 1/4:  20%|██        | 251/1250 [07:23<34:48,  2.09s/it, loss=0.776]
Epoch 1/4:  20%|██        | 252/1250 [07:23<32:27,  1.95s/it, loss=0.776]
Epoch 1/4:  20%|██        | 252/1250 [07:25<32:27,  1.95s/it, loss=1.12] 
Epoch 1/4:  20%|██        | 253/1250 [07:25<31:24,  1.89s/it, loss=1.12]
Epoch 1/4:  20%|██        | 253/1250 [07:27<31:24,  1.89s/it, loss=0.893]
Epoch 1/4:  20%|██        | 254/1250 [07:27<31:46,  1.91s/it, loss=0.893]
Epoch 1/4:  20%|██        | 254/1250 [07:28<31:46,  1.91s/it, loss=1.4]  
Epoch 1/4:  20%|██        | 255/1250 [07:28<30:27,  1.84s/it, loss=1.4]
Epoch 1/4:  20%|██        | 255/1250 [07:30<30:27,  1.84s/it, loss=1.08]
Epoch 1/4:  20%|██        | 256/1250 [07:30<28:54,  1.75s/it, loss=1.08]
Epoch 1/4:  20%|██        | 256/1250 [07:32<28:54,  1.75s/it, loss=1.09]
Epoch 1/4:  21%|██        | 257/1250 [07:32<28:54,  1.75s/it, loss=1.09]
Epoch 1/4:  21%|██        | 257/1250 [07:34<28:54,  1.75s/it, loss=0.706]
Epoch 1/4:  21%|██        | 258/1250 [07:34<29:44,  1.80s/it, loss=0.706]
Epoch 1/4:  21%|██        | 258/1250 [07:35<29:44,  1.80s/it, loss=1.45] 
Epoch 1/4:  21%|██        | 259/1250 [07:35<27:54,  1.69s/it, loss=1.45]
Epoch 1/4:  21%|██        | 259/1250 [07:37<27:54,  1.69s/it, loss=0.747]
Epoch 1/4:  21%|██        | 260/1250 [07:37<30:25,  1.84s/it, loss=0.747]
Epoch 1/4:  21%|██        | 260/1250 [07:39<30:25,  1.84s/it, loss=0.997]
Epoch 1/4:  21%|██        | 261/1250 [07:39<32:00,  1.94s/it, loss=0.997]
Epoch 1/4:  21%|██        | 261/1250 [07:42<32:00,  1.94s/it, loss=0.988]
Epoch 1/4:  21%|██        | 262/1250 [07:42<33:27,  2.03s/it, loss=0.988]
Epoch 1/4:  21%|██        | 262/1250 [07:43<33:27,  2.03s/it, loss=1.05] 
Epoch 1/4:  21%|██        | 263/1250 [07:43<31:00,  1.89s/it, loss=1.05]
Epoch 1/4:  21%|██        | 263/1250 [07:46<31:00,  1.89s/it, loss=0.811]
Epoch 1/4:  21%|██        | 264/1250 [07:46<33:29,  2.04s/it, loss=0.811]
Epoch 1/4:  21%|██        | 264/1250 [07:47<33:29,  2.04s/it, loss=1.06] 
Epoch 1/4:  21%|██        | 265/1250 [07:47<29:08,  1.78s/it, loss=1.06]
Epoch 1/4:  21%|██        | 265/1250 [07:49<29:08,  1.78s/it, loss=0.768]
Epoch 1/4:  21%|██▏       | 266/1250 [07:49<31:37,  1.93s/it, loss=0.768]
Epoch 1/4:  21%|██▏       | 266/1250 [07:50<31:37,  1.93s/it, loss=0.8]  
Epoch 1/4:  21%|██▏       | 267/1250 [07:50<29:42,  1.81s/it, loss=0.8]
Epoch 1/4:  21%|██▏       | 267/1250 [07:52<29:42,  1.81s/it, loss=0.717]
Epoch 1/4:  21%|██▏       | 268/1250 [07:52<28:23,  1.73s/it, loss=0.717]
Epoch 1/4:  21%|██▏       | 268/1250 [07:54<28:23,  1.73s/it, loss=0.667]
Epoch 1/4:  22%|██▏       | 269/1250 [07:54<27:47,  1.70s/it, loss=0.667]
Epoch 1/4:  22%|██▏       | 269/1250 [07:56<27:47,  1.70s/it, loss=0.839]
Epoch 1/4:  22%|██▏       | 270/1250 [07:56<29:54,  1.83s/it, loss=0.839]
Epoch 1/4:  22%|██▏       | 270/1250 [07:57<29:54,  1.83s/it, loss=0.785]
Epoch 1/4:  22%|██▏       | 271/1250 [07:57<27:59,  1.72s/it, loss=0.785]
Epoch 1/4:  22%|██▏       | 271/1250 [07:59<27:59,  1.72s/it, loss=0.916]
Epoch 1/4:  22%|██▏       | 272/1250 [07:59<27:14,  1.67s/it, loss=0.916]
Epoch 1/4:  22%|██▏       | 272/1250 [08:01<27:14,  1.67s/it, loss=1.11] 
Epoch 1/4:  22%|██▏       | 273/1250 [08:01<29:49,  1.83s/it, loss=1.11]
Epoch 1/4:  22%|██▏       | 273/1250 [08:02<29:49,  1.83s/it, loss=0.966]
Epoch 1/4:  22%|██▏       | 274/1250 [08:02<27:50,  1.71s/it, loss=0.966]
Epoch 1/4:  22%|██▏       | 274/1250 [08:05<27:50,  1.71s/it, loss=1.26] 
Epoch 1/4:  22%|██▏       | 275/1250 [08:05<30:30,  1.88s/it, loss=1.26]
Epoch 1/4:  22%|██▏       | 275/1250 [08:06<30:30,  1.88s/it, loss=0.943]
Epoch 1/4:  22%|██▏       | 276/1250 [08:06<27:23,  1.69s/it, loss=0.943]
Epoch 1/4:  22%|██▏       | 276/1250 [08:09<27:23,  1.69s/it, loss=1.25] 
Epoch 1/4:  22%|██▏       | 277/1250 [08:09<32:06,  1.98s/it, loss=1.25]
Epoch 1/4:  22%|██▏       | 277/1250 [08:11<32:06,  1.98s/it, loss=0.841]
Epoch 1/4:  22%|██▏       | 278/1250 [08:11<31:40,  1.96s/it, loss=0.841]
Epoch 1/4:  22%|██▏       | 278/1250 [08:12<31:40,  1.96s/it, loss=0.886]
Epoch 1/4:  22%|██▏       | 279/1250 [08:12<27:40,  1.71s/it, loss=0.886]
Epoch 1/4:  22%|██▏       | 279/1250 [08:14<27:40,  1.71s/it, loss=1.06] 
Epoch 1/4:  22%|██▏       | 280/1250 [08:14<29:08,  1.80s/it, loss=1.06]
Epoch 1/4:  22%|██▏       | 280/1250 [08:15<29:08,  1.80s/it, loss=1.75]
Epoch 1/4:  22%|██▏       | 281/1250 [08:15<27:21,  1.69s/it, loss=1.75]
Epoch 1/4:  22%|██▏       | 281/1250 [08:17<27:21,  1.69s/it, loss=1.05]
Epoch 1/4:  23%|██▎       | 282/1250 [08:17<26:51,  1.66s/it, loss=1.05]
Epoch 1/4:  23%|██▎       | 282/1250 [08:18<26:51,  1.66s/it, loss=1.07]
Epoch 1/4:  23%|██▎       | 283/1250 [08:18<26:36,  1.65s/it, loss=1.07]
Epoch 1/4:  23%|██▎       | 283/1250 [08:20<26:36,  1.65s/it, loss=0.675]
Epoch 1/4:  23%|██▎       | 284/1250 [08:20<25:39,  1.59s/it, loss=0.675]
Epoch 1/4:  23%|██▎       | 284/1250 [08:22<25:39,  1.59s/it, loss=0.968]
Epoch 1/4:  23%|██▎       | 285/1250 [08:22<29:06,  1.81s/it, loss=0.968]
Epoch 1/4:  23%|██▎       | 285/1250 [08:23<29:06,  1.81s/it, loss=1.42] 
Epoch 1/4:  23%|██▎       | 286/1250 [08:23<26:07,  1.63s/it, loss=1.42]
Epoch 1/4:  23%|██▎       | 286/1250 [08:25<26:07,  1.63s/it, loss=0.767]
Epoch 1/4:  23%|██▎       | 287/1250 [08:25<25:01,  1.56s/it, loss=0.767]
Epoch 1/4:  23%|██▎       | 287/1250 [08:26<25:01,  1.56s/it, loss=1.26] 
Epoch 1/4:  23%|██▎       | 288/1250 [08:26<25:24,  1.59s/it, loss=1.26]
Epoch 1/4:  23%|██▎       | 288/1250 [08:28<25:24,  1.59s/it, loss=0.584]
Epoch 1/4:  23%|██▎       | 289/1250 [08:28<25:07,  1.57s/it, loss=0.584]
Epoch 1/4:  23%|██▎       | 289/1250 [08:29<25:07,  1.57s/it, loss=0.971]
Epoch 1/4:  23%|██▎       | 290/1250 [08:29<24:05,  1.51s/it, loss=0.971]
Epoch 1/4:  23%|██▎       | 290/1250 [08:31<24:05,  1.51s/it, loss=1.05] 
Epoch 1/4:  23%|██▎       | 291/1250 [08:31<23:34,  1.48s/it, loss=1.05]
Epoch 1/4:  23%|██▎       | 291/1250 [08:32<23:34,  1.48s/it, loss=0.741]
Epoch 1/4:  23%|██▎       | 292/1250 [08:32<24:33,  1.54s/it, loss=0.741]
Epoch 1/4:  23%|██▎       | 292/1250 [08:34<24:33,  1.54s/it, loss=1.01] 
Epoch 1/4:  23%|██▎       | 293/1250 [08:34<24:07,  1.51s/it, loss=1.01]
Epoch 1/4:  23%|██▎       | 293/1250 [08:35<24:07,  1.51s/it, loss=1.27]
Epoch 1/4:  24%|██▎       | 294/1250 [08:35<22:41,  1.42s/it, loss=1.27]
Epoch 1/4:  24%|██▎       | 294/1250 [08:37<22:41,  1.42s/it, loss=1.14]
Epoch 1/4:  24%|██▎       | 295/1250 [08:37<24:30,  1.54s/it, loss=1.14]
Epoch 1/4:  24%|██▎       | 295/1250 [08:40<24:30,  1.54s/it, loss=1.19]
Epoch 1/4:  24%|██▎       | 296/1250 [08:40<33:11,  2.09s/it, loss=1.19]
Epoch 1/4:  24%|██▎       | 296/1250 [08:43<33:11,  2.09s/it, loss=1.05]
Epoch 1/4:  24%|██▍       | 297/1250 [08:43<35:32,  2.24s/it, loss=1.05]
Epoch 1/4:  24%|██▍       | 297/1250 [08:45<35:32,  2.24s/it, loss=1]   
Epoch 1/4:  24%|██▍       | 298/1250 [08:45<35:19,  2.23s/it, loss=1]
Epoch 1/4:  24%|██▍       | 298/1250 [08:47<35:19,  2.23s/it, loss=1.49]
Epoch 1/4:  24%|██▍       | 299/1250 [08:47<34:25,  2.17s/it, loss=1.49]
Epoch 1/4:  24%|██▍       | 299/1250 [08:48<34:25,  2.17s/it, loss=0.851]
Epoch 1/4:  24%|██▍       | 300/1250 [08:48<30:21,  1.92s/it, loss=0.851]
Epoch 1/4:  24%|██▍       | 300/1250 [08:50<30:21,  1.92s/it, loss=0.675]
Epoch 1/4:  24%|██▍       | 301/1250 [08:50<27:57,  1.77s/it, loss=0.675]
Epoch 1/4:  24%|██▍       | 301/1250 [08:52<27:57,  1.77s/it, loss=0.926]
Epoch 1/4:  24%|██▍       | 302/1250 [08:52<30:31,  1.93s/it, loss=0.926]
Epoch 1/4:  24%|██▍       | 302/1250 [08:53<30:31,  1.93s/it, loss=0.94] 
Epoch 1/4:  24%|██▍       | 303/1250 [08:53<26:57,  1.71s/it, loss=0.94]
Epoch 1/4:  24%|██▍       | 303/1250 [08:55<26:57,  1.71s/it, loss=0.983]
Epoch 1/4:  24%|██▍       | 304/1250 [08:55<26:07,  1.66s/it, loss=0.983]
Epoch 1/4:  24%|██▍       | 304/1250 [08:56<26:07,  1.66s/it, loss=0.907]
Epoch 1/4:  24%|██▍       | 305/1250 [08:56<25:38,  1.63s/it, loss=0.907]
Epoch 1/4:  24%|██▍       | 305/1250 [08:58<25:38,  1.63s/it, loss=0.96] 
Epoch 1/4:  24%|██▍       | 306/1250 [08:58<24:18,  1.54s/it, loss=0.96]
Epoch 1/4:  24%|██▍       | 306/1250 [08:59<24:18,  1.54s/it, loss=0.826]
Epoch 1/4:  25%|██▍       | 307/1250 [08:59<23:34,  1.50s/it, loss=0.826]
Epoch 1/4:  25%|██▍       | 307/1250 [09:01<23:34,  1.50s/it, loss=1.38] 
Epoch 1/4:  25%|██▍       | 308/1250 [09:01<24:17,  1.55s/it, loss=1.38]
Epoch 1/4:  25%|██▍       | 308/1250 [09:02<24:17,  1.55s/it, loss=0.641]
Epoch 1/4:  25%|██▍       | 309/1250 [09:02<25:02,  1.60s/it, loss=0.641]
Epoch 1/4:  25%|██▍       | 309/1250 [09:04<25:02,  1.60s/it, loss=0.583]
Epoch 1/4:  25%|██▍       | 310/1250 [09:04<24:56,  1.59s/it, loss=0.583]
Epoch 1/4:  25%|██▍       | 310/1250 [09:05<24:56,  1.59s/it, loss=0.855]
Epoch 1/4:  25%|██▍       | 311/1250 [09:05<24:05,  1.54s/it, loss=0.855]
Epoch 1/4:  25%|██▍       | 311/1250 [09:08<24:05,  1.54s/it, loss=1.47] 
Epoch 1/4:  25%|██▍       | 312/1250 [09:08<26:58,  1.73s/it, loss=1.47]
Epoch 1/4:  25%|██▍       | 312/1250 [09:09<26:58,  1.73s/it, loss=0.657]
Epoch 1/4:  25%|██▌       | 313/1250 [09:09<25:30,  1.63s/it, loss=0.657]
Epoch 1/4:  25%|██▌       | 313/1250 [09:11<25:30,  1.63s/it, loss=1.35] 
Epoch 1/4:  25%|██▌       | 314/1250 [09:11<26:03,  1.67s/it, loss=1.35]
Epoch 1/4:  25%|██▌       | 314/1250 [09:12<26:03,  1.67s/it, loss=1.31]
Epoch 1/4:  25%|██▌       | 315/1250 [09:12<25:25,  1.63s/it, loss=1.31]
Epoch 1/4:  25%|██▌       | 315/1250 [09:14<25:25,  1.63s/it, loss=0.8] 
Epoch 1/4:  25%|██▌       | 316/1250 [09:14<26:57,  1.73s/it, loss=0.8]
Epoch 1/4:  25%|██▌       | 316/1250 [09:15<26:57,  1.73s/it, loss=0.906]
Epoch 1/4:  25%|██▌       | 317/1250 [09:15<24:07,  1.55s/it, loss=0.906]
Epoch 1/4:  25%|██▌       | 317/1250 [09:17<24:07,  1.55s/it, loss=0.768]
Epoch 1/4:  25%|██▌       | 318/1250 [09:17<24:30,  1.58s/it, loss=0.768]
Epoch 1/4:  25%|██▌       | 318/1250 [09:19<24:30,  1.58s/it, loss=0.927]
Epoch 1/4:  26%|██▌       | 319/1250 [09:19<24:34,  1.58s/it, loss=0.927]
Epoch 1/4:  26%|██▌       | 319/1250 [09:20<24:34,  1.58s/it, loss=0.953]
Epoch 1/4:  26%|██▌       | 320/1250 [09:20<23:38,  1.53s/it, loss=0.953]
Epoch 1/4:  26%|██▌       | 320/1250 [09:21<23:38,  1.53s/it, loss=0.644]
Epoch 1/4:  26%|██▌       | 321/1250 [09:21<22:58,  1.48s/it, loss=0.644]
Epoch 1/4:  26%|██▌       | 321/1250 [09:23<22:58,  1.48s/it, loss=0.981]
Epoch 1/4:  26%|██▌       | 322/1250 [09:23<23:01,  1.49s/it, loss=0.981]
Epoch 1/4:  26%|██▌       | 322/1250 [09:25<23:01,  1.49s/it, loss=0.861]
Epoch 1/4:  26%|██▌       | 323/1250 [09:25<23:35,  1.53s/it, loss=0.861]
Epoch 1/4:  26%|██▌       | 323/1250 [09:26<23:35,  1.53s/it, loss=1.01] 
Epoch 1/4:  26%|██▌       | 324/1250 [09:26<22:18,  1.45s/it, loss=1.01]
Epoch 1/4:  26%|██▌       | 324/1250 [09:27<22:18,  1.45s/it, loss=0.989]
Epoch 1/4:  26%|██▌       | 325/1250 [09:27<23:18,  1.51s/it, loss=0.989]
Epoch 1/4:  26%|██▌       | 325/1250 [09:29<23:18,  1.51s/it, loss=0.616]
Epoch 1/4:  26%|██▌       | 326/1250 [09:29<22:52,  1.49s/it, loss=0.616]
Epoch 1/4:  26%|██▌       | 326/1250 [09:30<22:52,  1.49s/it, loss=1.04] 
Epoch 1/4:  26%|██▌       | 327/1250 [09:30<23:11,  1.51s/it, loss=1.04]
Epoch 1/4:  26%|██▌       | 327/1250 [09:32<23:11,  1.51s/it, loss=1.15]
Epoch 1/4:  26%|██▌       | 328/1250 [09:32<23:03,  1.50s/it, loss=1.15]
Epoch 1/4:  26%|██▌       | 328/1250 [09:34<23:03,  1.50s/it, loss=1.06]
Epoch 1/4:  26%|██▋       | 329/1250 [09:34<25:44,  1.68s/it, loss=1.06]
Epoch 1/4:  26%|██▋       | 329/1250 [09:35<25:44,  1.68s/it, loss=0.875]
Epoch 1/4:  26%|██▋       | 330/1250 [09:35<23:03,  1.50s/it, loss=0.875]
Epoch 1/4:  26%|██▋       | 330/1250 [09:37<23:03,  1.50s/it, loss=0.692]
Epoch 1/4:  26%|██▋       | 331/1250 [09:37<23:17,  1.52s/it, loss=0.692]
Epoch 1/4:  26%|██▋       | 331/1250 [09:38<23:17,  1.52s/it, loss=1.17] 
Epoch 1/4:  27%|██▋       | 332/1250 [09:38<23:21,  1.53s/it, loss=1.17]
Epoch 1/4:  27%|██▋       | 332/1250 [09:40<23:21,  1.53s/it, loss=1.05]
Epoch 1/4:  27%|██▋       | 333/1250 [09:40<22:38,  1.48s/it, loss=1.05]
Epoch 1/4:  27%|██▋       | 333/1250 [09:41<22:38,  1.48s/it, loss=1.03]
Epoch 1/4:  27%|██▋       | 334/1250 [09:41<21:58,  1.44s/it, loss=1.03]
Epoch 1/4:  27%|██▋       | 334/1250 [09:42<21:58,  1.44s/it, loss=0.99]
Epoch 1/4:  27%|██▋       | 335/1250 [09:42<22:00,  1.44s/it, loss=0.99]
Epoch 1/4:  27%|██▋       | 335/1250 [09:44<22:00,  1.44s/it, loss=0.877]
Epoch 1/4:  27%|██▋       | 336/1250 [09:44<22:08,  1.45s/it, loss=0.877]
Epoch 1/4:  27%|██▋       | 336/1250 [09:45<22:08,  1.45s/it, loss=1.03] 
Epoch 1/4:  27%|██▋       | 337/1250 [09:45<22:11,  1.46s/it, loss=1.03]
Epoch 1/4:  27%|██▋       | 337/1250 [09:47<22:11,  1.46s/it, loss=1.08]
Epoch 1/4:  27%|██▋       | 338/1250 [09:47<21:54,  1.44s/it, loss=1.08]
Epoch 1/4:  27%|██▋       | 338/1250 [09:48<21:54,  1.44s/it, loss=1.22]
Epoch 1/4:  27%|██▋       | 339/1250 [09:48<21:51,  1.44s/it, loss=1.22]
Epoch 1/4:  27%|██▋       | 339/1250 [09:50<21:51,  1.44s/it, loss=1.58]
Epoch 1/4:  27%|██▋       | 340/1250 [09:50<22:24,  1.48s/it, loss=1.58]
Epoch 1/4:  27%|██▋       | 340/1250 [09:51<22:24,  1.48s/it, loss=0.652]
Epoch 1/4:  27%|██▋       | 341/1250 [09:51<22:10,  1.46s/it, loss=0.652]
Epoch 1/4:  27%|██▋       | 341/1250 [09:53<22:10,  1.46s/it, loss=0.868]
Epoch 1/4:  27%|██▋       | 342/1250 [09:53<22:21,  1.48s/it, loss=0.868]
Epoch 1/4:  27%|██▋       | 342/1250 [09:54<22:21,  1.48s/it, loss=0.808]
Epoch 1/4:  27%|██▋       | 343/1250 [09:54<23:01,  1.52s/it, loss=0.808]
Epoch 1/4:  27%|██▋       | 343/1250 [09:56<23:01,  1.52s/it, loss=0.915]
Epoch 1/4:  28%|██▊       | 344/1250 [09:56<22:02,  1.46s/it, loss=0.915]
Epoch 1/4:  28%|██▊       | 344/1250 [09:58<22:02,  1.46s/it, loss=1.22] 
Epoch 1/4:  28%|██▊       | 345/1250 [09:58<24:47,  1.64s/it, loss=1.22]
Epoch 1/4:  28%|██▊       | 345/1250 [09:59<24:47,  1.64s/it, loss=0.761]
Epoch 1/4:  28%|██▊       | 346/1250 [09:59<23:07,  1.53s/it, loss=0.761]
Epoch 1/4:  28%|██▊       | 346/1250 [10:00<23:07,  1.53s/it, loss=0.756]
Epoch 1/4:  28%|██▊       | 347/1250 [10:00<22:45,  1.51s/it, loss=0.756]
Epoch 1/4:  28%|██▊       | 347/1250 [10:02<22:45,  1.51s/it, loss=0.878]
Epoch 1/4:  28%|██▊       | 348/1250 [10:02<22:41,  1.51s/it, loss=0.878]
Epoch 1/4:  28%|██▊       | 348/1250 [10:03<22:41,  1.51s/it, loss=1.31] 
Epoch 1/4:  28%|██▊       | 349/1250 [10:03<21:32,  1.43s/it, loss=1.31]
Epoch 1/4:  28%|██▊       | 349/1250 [10:05<21:32,  1.43s/it, loss=1.01]
Epoch 1/4:  28%|██▊       | 350/1250 [10:05<21:03,  1.40s/it, loss=1.01]
Epoch 1/4:  28%|██▊       | 350/1250 [10:06<21:03,  1.40s/it, loss=0.706]
Epoch 1/4:  28%|██▊       | 351/1250 [10:06<20:23,  1.36s/it, loss=0.706]
Epoch 1/4:  28%|██▊       | 351/1250 [10:07<20:23,  1.36s/it, loss=2.12] 
Epoch 1/4:  28%|██▊       | 352/1250 [10:07<19:49,  1.32s/it, loss=2.12]
Epoch 1/4:  28%|██▊       | 352/1250 [10:08<19:49,  1.32s/it, loss=1.14]
Epoch 1/4:  28%|██▊       | 353/1250 [10:08<19:49,  1.33s/it, loss=1.14]
Epoch 1/4:  28%|██▊       | 353/1250 [10:10<19:49,  1.33s/it, loss=0.954]
Epoch 1/4:  28%|██▊       | 354/1250 [10:10<19:42,  1.32s/it, loss=0.954]
Epoch 1/4:  28%|██▊       | 354/1250 [10:11<19:42,  1.32s/it, loss=1.12] 
Epoch 1/4:  28%|██▊       | 355/1250 [10:11<19:09,  1.28s/it, loss=1.12]
Epoch 1/4:  28%|██▊       | 355/1250 [10:12<19:09,  1.28s/it, loss=0.95]
Epoch 1/4:  28%|██▊       | 356/1250 [10:12<18:54,  1.27s/it, loss=0.95]
Epoch 1/4:  28%|██▊       | 356/1250 [10:14<18:54,  1.27s/it, loss=0.753]
Epoch 1/4:  29%|██▊       | 357/1250 [10:14<19:39,  1.32s/it, loss=0.753]
Epoch 1/4:  29%|██▊       | 357/1250 [10:15<19:39,  1.32s/it, loss=1.29] 
Epoch 1/4:  29%|██▊       | 358/1250 [10:15<19:50,  1.33s/it, loss=1.29]
Epoch 1/4:  29%|██▊       | 358/1250 [10:16<19:50,  1.33s/it, loss=1.38]
Epoch 1/4:  29%|██▊       | 359/1250 [10:16<19:15,  1.30s/it, loss=1.38]
Epoch 1/4:  29%|██▊       | 359/1250 [10:17<19:15,  1.30s/it, loss=1.28]
Epoch 1/4:  29%|██▉       | 360/1250 [10:17<19:07,  1.29s/it, loss=1.28]
Epoch 1/4:  29%|██▉       | 360/1250 [10:19<19:07,  1.29s/it, loss=1.12]
Epoch 1/4:  29%|██▉       | 361/1250 [10:19<19:24,  1.31s/it, loss=1.12]
Epoch 1/4:  29%|██▉       | 361/1250 [10:20<19:24,  1.31s/it, loss=1.45]
Epoch 1/4:  29%|██▉       | 362/1250 [10:20<19:09,  1.29s/it, loss=1.45]
Epoch 1/4:  29%|██▉       | 362/1250 [10:21<19:09,  1.29s/it, loss=1.28]
Epoch 1/4:  29%|██▉       | 363/1250 [10:21<18:49,  1.27s/it, loss=1.28]
Epoch 1/4:  29%|██▉       | 363/1250 [10:23<18:49,  1.27s/it, loss=0.781]
Epoch 1/4:  29%|██▉       | 364/1250 [10:23<18:59,  1.29s/it, loss=0.781]
Epoch 1/4:  29%|██▉       | 364/1250 [10:24<18:59,  1.29s/it, loss=0.992]
Epoch 1/4:  29%|██▉       | 365/1250 [10:24<19:35,  1.33s/it, loss=0.992]
Epoch 1/4:  29%|██▉       | 365/1250 [10:25<19:35,  1.33s/it, loss=1.02] 
Epoch 1/4:  29%|██▉       | 366/1250 [10:25<19:29,  1.32s/it, loss=1.02]
Epoch 1/4:  29%|██▉       | 366/1250 [10:27<19:29,  1.32s/it, loss=1.08]
Epoch 1/4:  29%|██▉       | 367/1250 [10:27<19:21,  1.32s/it, loss=1.08]
Epoch 1/4:  29%|██▉       | 367/1250 [10:28<19:21,  1.32s/it, loss=0.691]
Epoch 1/4:  29%|██▉       | 368/1250 [10:28<19:30,  1.33s/it, loss=0.691]
Epoch 1/4:  29%|██▉       | 368/1250 [10:29<19:30,  1.33s/it, loss=0.951]
Epoch 1/4:  30%|██▉       | 369/1250 [10:29<19:42,  1.34s/it, loss=0.951]
Epoch 1/4:  30%|██▉       | 369/1250 [10:31<19:42,  1.34s/it, loss=1.1]  
Epoch 1/4:  30%|██▉       | 370/1250 [10:31<19:20,  1.32s/it, loss=1.1]
Epoch 1/4:  30%|██▉       | 370/1250 [10:32<19:20,  1.32s/it, loss=0.848]
Epoch 1/4:  30%|██▉       | 371/1250 [10:32<18:55,  1.29s/it, loss=0.848]
Epoch 1/4:  30%|██▉       | 371/1250 [10:33<18:55,  1.29s/it, loss=0.736]
Epoch 1/4:  30%|██▉       | 372/1250 [10:33<19:01,  1.30s/it, loss=0.736]
Epoch 1/4:  30%|██▉       | 372/1250 [10:34<19:01,  1.30s/it, loss=0.662]
Epoch 1/4:  30%|██▉       | 373/1250 [10:34<19:05,  1.31s/it, loss=0.662]
Epoch 1/4:  30%|██▉       | 373/1250 [10:36<19:05,  1.31s/it, loss=1.01] 
Epoch 1/4:  30%|██▉       | 374/1250 [10:36<18:39,  1.28s/it, loss=1.01]
Epoch 1/4:  30%|██▉       | 374/1250 [10:37<18:39,  1.28s/it, loss=0.799]
Epoch 1/4:  30%|███       | 375/1250 [10:37<18:13,  1.25s/it, loss=0.799]
Epoch 1/4:  30%|███       | 375/1250 [10:38<18:13,  1.25s/it, loss=0.811]
Epoch 1/4:  30%|███       | 376/1250 [10:38<18:37,  1.28s/it, loss=0.811]
Epoch 1/4:  30%|███       | 376/1250 [10:40<18:37,  1.28s/it, loss=1.07] 
Epoch 1/4:  30%|███       | 377/1250 [10:40<19:01,  1.31s/it, loss=1.07]
Epoch 1/4:  30%|███       | 377/1250 [10:41<19:01,  1.31s/it, loss=1.37]
Epoch 1/4:  30%|███       | 378/1250 [10:41<18:38,  1.28s/it, loss=1.37]
Epoch 1/4:  30%|███       | 378/1250 [10:42<18:38,  1.28s/it, loss=0.987]
Epoch 1/4:  30%|███       | 379/1250 [10:42<18:38,  1.28s/it, loss=0.987]
Epoch 1/4:  30%|███       | 379/1250 [10:43<18:38,  1.28s/it, loss=0.573]
Epoch 1/4:  30%|███       | 380/1250 [10:43<18:53,  1.30s/it, loss=0.573]
Epoch 1/4:  30%|███       | 380/1250 [10:45<18:53,  1.30s/it, loss=1.23] 
Epoch 1/4:  30%|███       | 381/1250 [10:45<19:09,  1.32s/it, loss=1.23]
Epoch 1/4:  30%|███       | 381/1250 [10:46<19:09,  1.32s/it, loss=0.833]
Epoch 1/4:  31%|███       | 382/1250 [10:46<20:04,  1.39s/it, loss=0.833]
Epoch 1/4:  31%|███       | 382/1250 [10:48<20:04,  1.39s/it, loss=1.31] 
Epoch 1/4:  31%|███       | 383/1250 [10:48<19:50,  1.37s/it, loss=1.31]
Epoch 1/4:  31%|███       | 383/1250 [10:49<19:50,  1.37s/it, loss=0.935]
Epoch 1/4:  31%|███       | 384/1250 [10:49<19:48,  1.37s/it, loss=0.935]
Epoch 1/4:  31%|███       | 384/1250 [10:50<19:48,  1.37s/it, loss=0.626]
Epoch 1/4:  31%|███       | 385/1250 [10:50<19:32,  1.36s/it, loss=0.626]
Epoch 1/4:  31%|███       | 385/1250 [10:52<19:32,  1.36s/it, loss=1.01] 
Epoch 1/4:  31%|███       | 386/1250 [10:52<18:50,  1.31s/it, loss=1.01]
Epoch 1/4:  31%|███       | 386/1250 [10:53<18:50,  1.31s/it, loss=0.874]
Epoch 1/4:  31%|███       | 387/1250 [10:53<18:39,  1.30s/it, loss=0.874]
Epoch 1/4:  31%|███       | 387/1250 [10:54<18:39,  1.30s/it, loss=1.02] 
Epoch 1/4:  31%|███       | 388/1250 [10:54<19:07,  1.33s/it, loss=1.02]
Epoch 1/4:  31%|███       | 388/1250 [10:55<19:07,  1.33s/it, loss=0.91]
Epoch 1/4:  31%|███       | 389/1250 [10:55<18:33,  1.29s/it, loss=0.91]
Epoch 1/4:  31%|███       | 389/1250 [10:57<18:33,  1.29s/it, loss=1.01]
Epoch 1/4:  31%|███       | 390/1250 [10:57<17:31,  1.22s/it, loss=1.01]
Epoch 1/4:  31%|███       | 390/1250 [10:58<17:31,  1.22s/it, loss=0.917]
Epoch 1/4:  31%|███▏      | 391/1250 [10:58<16:43,  1.17s/it, loss=0.917]
Epoch 1/4:  31%|███▏      | 391/1250 [10:59<16:43,  1.17s/it, loss=0.831]
Epoch 1/4:  31%|███▏      | 392/1250 [10:59<17:08,  1.20s/it, loss=0.831]
Epoch 1/4:  31%|███▏      | 392/1250 [11:00<17:08,  1.20s/it, loss=1.17] 
Epoch 1/4:  31%|███▏      | 393/1250 [11:00<17:04,  1.19s/it, loss=1.17]
Epoch 1/4:  31%|███▏      | 393/1250 [11:01<17:04,  1.19s/it, loss=0.924]
Epoch 1/4:  32%|███▏      | 394/1250 [11:01<16:52,  1.18s/it, loss=0.924]
Epoch 1/4:  32%|███▏      | 394/1250 [11:02<16:52,  1.18s/it, loss=1.01] 
Epoch 1/4:  32%|███▏      | 395/1250 [11:02<17:26,  1.22s/it, loss=1.01]
Epoch 1/4:  32%|███▏      | 395/1250 [11:04<17:26,  1.22s/it, loss=0.829]
Epoch 1/4:  32%|███▏      | 396/1250 [11:04<19:06,  1.34s/it, loss=0.829]
Epoch 1/4:  32%|███▏      | 396/1250 [11:06<19:06,  1.34s/it, loss=1.17] 
Epoch 1/4:  32%|███▏      | 397/1250 [11:06<20:53,  1.47s/it, loss=1.17]
Epoch 1/4:  32%|███▏      | 397/1250 [11:08<20:53,  1.47s/it, loss=1.09]
Epoch 1/4:  32%|███▏      | 398/1250 [11:08<23:11,  1.63s/it, loss=1.09]
Epoch 1/4:  32%|███▏      | 398/1250 [11:09<23:11,  1.63s/it, loss=0.709]
Epoch 1/4:  32%|███▏      | 399/1250 [11:09<21:31,  1.52s/it, loss=0.709]
Epoch 1/4:  32%|███▏      | 399/1250 [11:10<21:31,  1.52s/it, loss=0.815]
Epoch 1/4:  32%|███▏      | 400/1250 [11:10<20:39,  1.46s/it, loss=0.815]
Epoch 1/4:  32%|███▏      | 400/1250 [11:12<20:39,  1.46s/it, loss=0.893]
Epoch 1/4:  32%|███▏      | 401/1250 [11:12<20:37,  1.46s/it, loss=0.893]
Epoch 1/4:  32%|███▏      | 401/1250 [11:13<20:37,  1.46s/it, loss=0.57] 
Epoch 1/4:  32%|███▏      | 402/1250 [11:13<20:24,  1.44s/it, loss=0.57]
Epoch 1/4:  32%|███▏      | 402/1250 [11:15<20:24,  1.44s/it, loss=0.802]
Epoch 1/4:  32%|███▏      | 403/1250 [11:15<21:26,  1.52s/it, loss=0.802]
Epoch 1/4:  32%|███▏      | 403/1250 [11:16<21:26,  1.52s/it, loss=0.849]
Epoch 1/4:  32%|███▏      | 404/1250 [11:16<20:55,  1.48s/it, loss=0.849]
Epoch 1/4:  32%|███▏      | 404/1250 [11:18<20:55,  1.48s/it, loss=1.23] 
Epoch 1/4:  32%|███▏      | 405/1250 [11:18<20:19,  1.44s/it, loss=1.23]
Epoch 1/4:  32%|███▏      | 405/1250 [11:19<20:19,  1.44s/it, loss=0.851]
Epoch 1/4:  32%|███▏      | 406/1250 [11:19<19:40,  1.40s/it, loss=0.851]
Epoch 1/4:  32%|███▏      | 406/1250 [11:20<19:40,  1.40s/it, loss=0.849]
Epoch 1/4:  33%|███▎      | 407/1250 [11:20<19:16,  1.37s/it, loss=0.849]
Epoch 1/4:  33%|███▎      | 407/1250 [11:22<19:16,  1.37s/it, loss=0.849]
Epoch 1/4:  33%|███▎      | 408/1250 [11:22<18:44,  1.34s/it, loss=0.849]
Epoch 1/4:  33%|███▎      | 408/1250 [11:23<18:44,  1.34s/it, loss=0.845]
Epoch 1/4:  33%|███▎      | 409/1250 [11:23<18:00,  1.28s/it, loss=0.845]
Epoch 1/4:  33%|███▎      | 409/1250 [11:24<18:00,  1.28s/it, loss=0.946]
Epoch 1/4:  33%|███▎      | 410/1250 [11:24<17:42,  1.26s/it, loss=0.946]
Epoch 1/4:  33%|███▎      | 410/1250 [11:25<17:42,  1.26s/it, loss=0.881]
Epoch 1/4:  33%|███▎      | 411/1250 [11:25<17:33,  1.26s/it, loss=0.881]
Epoch 1/4:  33%|███▎      | 411/1250 [11:26<17:33,  1.26s/it, loss=0.956]
Epoch 1/4:  33%|███▎      | 412/1250 [11:26<17:27,  1.25s/it, loss=0.956]
Epoch 1/4:  33%|███▎      | 412/1250 [11:28<17:27,  1.25s/it, loss=0.703]
Epoch 1/4:  33%|███▎      | 413/1250 [11:28<17:32,  1.26s/it, loss=0.703]
Epoch 1/4:  33%|███▎      | 413/1250 [11:29<17:32,  1.26s/it, loss=0.872]
Epoch 1/4:  33%|███▎      | 414/1250 [11:29<17:56,  1.29s/it, loss=0.872]
Epoch 1/4:  33%|███▎      | 414/1250 [11:30<17:56,  1.29s/it, loss=0.911]
Epoch 1/4:  33%|███▎      | 415/1250 [11:30<18:03,  1.30s/it, loss=0.911]
Epoch 1/4:  33%|███▎      | 415/1250 [11:32<18:03,  1.30s/it, loss=0.993]
Epoch 1/4:  33%|███▎      | 416/1250 [11:32<17:45,  1.28s/it, loss=0.993]
Epoch 1/4:  33%|███▎      | 416/1250 [11:33<17:45,  1.28s/it, loss=0.916]
Epoch 1/4:  33%|███▎      | 417/1250 [11:33<17:29,  1.26s/it, loss=0.916]
Epoch 1/4:  33%|███▎      | 417/1250 [11:34<17:29,  1.26s/it, loss=0.985]
Epoch 1/4:  33%|███▎      | 418/1250 [11:34<17:38,  1.27s/it, loss=0.985]
Epoch 1/4:  33%|███▎      | 418/1250 [11:35<17:38,  1.27s/it, loss=0.891]
Epoch 1/4:  34%|███▎      | 419/1250 [11:35<17:40,  1.28s/it, loss=0.891]
Epoch 1/4:  34%|███▎      | 419/1250 [11:37<17:40,  1.28s/it, loss=0.717]
Epoch 1/4:  34%|███▎      | 420/1250 [11:37<17:16,  1.25s/it, loss=0.717]
Epoch 1/4:  34%|███▎      | 420/1250 [11:38<17:16,  1.25s/it, loss=0.75] 
Epoch 1/4:  34%|███▎      | 421/1250 [11:38<16:46,  1.21s/it, loss=0.75]
Epoch 1/4:  34%|███▎      | 421/1250 [11:39<16:46,  1.21s/it, loss=0.907]
Epoch 1/4:  34%|███▍      | 422/1250 [11:39<16:57,  1.23s/it, loss=0.907]
Epoch 1/4:  34%|███▍      | 422/1250 [11:40<16:57,  1.23s/it, loss=0.665]
Epoch 1/4:  34%|███▍      | 423/1250 [11:40<17:11,  1.25s/it, loss=0.665]
Epoch 1/4:  34%|███▍      | 423/1250 [11:41<17:11,  1.25s/it, loss=0.745]
Epoch 1/4:  34%|███▍      | 424/1250 [11:41<16:32,  1.20s/it, loss=0.745]
Epoch 1/4:  34%|███▍      | 424/1250 [11:42<16:32,  1.20s/it, loss=0.908]
Epoch 1/4:  34%|███▍      | 425/1250 [11:42<15:49,  1.15s/it, loss=0.908]
Epoch 1/4:  34%|███▍      | 425/1250 [11:44<15:49,  1.15s/it, loss=1.01] 
Epoch 1/4:  34%|███▍      | 426/1250 [11:44<15:48,  1.15s/it, loss=1.01]
Epoch 1/4:  34%|███▍      | 426/1250 [11:45<15:48,  1.15s/it, loss=0.735]
Epoch 1/4:  34%|███▍      | 427/1250 [11:45<18:33,  1.35s/it, loss=0.735]
Epoch 1/4:  34%|███▍      | 427/1250 [11:47<18:33,  1.35s/it, loss=1.19] 
Epoch 1/4:  34%|███▍      | 428/1250 [11:47<18:10,  1.33s/it, loss=1.19]
Epoch 1/4:  34%|███▍      | 428/1250 [11:48<18:10,  1.33s/it, loss=0.672]
Epoch 1/4:  34%|███▍      | 429/1250 [11:48<17:30,  1.28s/it, loss=0.672]
Epoch 1/4:  34%|███▍      | 429/1250 [11:49<17:30,  1.28s/it, loss=0.993]
Epoch 1/4:  34%|███▍      | 430/1250 [11:49<17:09,  1.26s/it, loss=0.993]
Epoch 1/4:  34%|███▍      | 430/1250 [11:50<17:09,  1.26s/it, loss=0.775]
Epoch 1/4:  34%|███▍      | 431/1250 [11:50<17:00,  1.25s/it, loss=0.775]
Epoch 1/4:  34%|███▍      | 431/1250 [11:51<17:00,  1.25s/it, loss=0.839]
Epoch 1/4:  35%|███▍      | 432/1250 [11:51<16:45,  1.23s/it, loss=0.839]
Epoch 1/4:  35%|███▍      | 432/1250 [11:53<16:45,  1.23s/it, loss=1.1]  
Epoch 1/4:  35%|███▍      | 433/1250 [11:53<16:30,  1.21s/it, loss=1.1]
Epoch 1/4:  35%|███▍      | 433/1250 [11:54<16:30,  1.21s/it, loss=0.71]
Epoch 1/4:  35%|███▍      | 434/1250 [11:54<16:35,  1.22s/it, loss=0.71]
Epoch 1/4:  35%|███▍      | 434/1250 [11:55<16:35,  1.22s/it, loss=1.14]
Epoch 1/4:  35%|███▍      | 435/1250 [11:55<16:50,  1.24s/it, loss=1.14]
Epoch 1/4:  35%|███▍      | 435/1250 [11:56<16:50,  1.24s/it, loss=0.986]
Epoch 1/4:  35%|███▍      | 436/1250 [11:56<16:44,  1.23s/it, loss=0.986]
Epoch 1/4:  35%|███▍      | 436/1250 [11:58<16:44,  1.23s/it, loss=1.15] 
Epoch 1/4:  35%|███▍      | 437/1250 [11:58<16:24,  1.21s/it, loss=1.15]
Epoch 1/4:  35%|███▍      | 437/1250 [11:59<16:24,  1.21s/it, loss=1.02]
Epoch 1/4:  35%|███▌      | 438/1250 [11:59<16:16,  1.20s/it, loss=1.02]
Epoch 1/4:  35%|███▌      | 438/1250 [12:00<16:16,  1.20s/it, loss=1]   
Epoch 1/4:  35%|███▌      | 439/1250 [12:00<16:22,  1.21s/it, loss=1]
Epoch 1/4:  35%|███▌      | 439/1250 [12:01<16:22,  1.21s/it, loss=0.643]
Epoch 1/4:  35%|███▌      | 440/1250 [12:01<16:19,  1.21s/it, loss=0.643]
Epoch 1/4:  35%|███▌      | 440/1250 [12:02<16:19,  1.21s/it, loss=0.864]
Epoch 1/4:  35%|███▌      | 441/1250 [12:02<16:19,  1.21s/it, loss=0.864]
Epoch 1/4:  35%|███▌      | 441/1250 [12:04<16:19,  1.21s/it, loss=1.21] 
Epoch 1/4:  35%|███▌      | 442/1250 [12:04<16:32,  1.23s/it, loss=1.21]
Epoch 1/4:  35%|███▌      | 442/1250 [12:05<16:32,  1.23s/it, loss=0.868]
Epoch 1/4:  35%|███▌      | 443/1250 [12:05<16:56,  1.26s/it, loss=0.868]
Epoch 1/4:  35%|███▌      | 443/1250 [12:06<16:56,  1.26s/it, loss=1.44] 
Epoch 1/4:  36%|███▌      | 444/1250 [12:06<16:42,  1.24s/it, loss=1.44]
Epoch 1/4:  36%|███▌      | 444/1250 [12:07<16:42,  1.24s/it, loss=0.774]
Epoch 1/4:  36%|███▌      | 445/1250 [12:07<16:09,  1.20s/it, loss=0.774]
Epoch 1/4:  36%|███▌      | 445/1250 [12:08<16:09,  1.20s/it, loss=0.808]
Epoch 1/4:  36%|███▌      | 446/1250 [12:08<15:44,  1.17s/it, loss=0.808]
Epoch 1/4:  36%|███▌      | 446/1250 [12:10<15:44,  1.17s/it, loss=1.01] 
Epoch 1/4:  36%|███▌      | 447/1250 [12:10<15:55,  1.19s/it, loss=1.01]
Epoch 1/4:  36%|███▌      | 447/1250 [12:11<15:55,  1.19s/it, loss=2.16]
Epoch 1/4:  36%|███▌      | 448/1250 [12:11<15:43,  1.18s/it, loss=2.16]
Epoch 1/4:  36%|███▌      | 448/1250 [12:12<15:43,  1.18s/it, loss=1.3] 
Epoch 1/4:  36%|███▌      | 449/1250 [12:12<14:59,  1.12s/it, loss=1.3]
Epoch 1/4:  36%|███▌      | 449/1250 [12:13<14:59,  1.12s/it, loss=1.13]
Epoch 1/4:  36%|███▌      | 450/1250 [12:13<14:19,  1.07s/it, loss=1.13]
Epoch 1/4:  36%|███▌      | 450/1250 [12:14<14:19,  1.07s/it, loss=0.707]
Epoch 1/4:  36%|███▌      | 451/1250 [12:14<14:14,  1.07s/it, loss=0.707]
Epoch 1/4:  36%|███▌      | 451/1250 [12:15<14:14,  1.07s/it, loss=0.956]
Epoch 1/4:  36%|███▌      | 452/1250 [12:15<14:59,  1.13s/it, loss=0.956]
Epoch 1/4:  36%|███▌      | 452/1250 [12:16<14:59,  1.13s/it, loss=0.59] 
Epoch 1/4:  36%|███▌      | 453/1250 [12:16<14:58,  1.13s/it, loss=0.59]
Epoch 1/4:  36%|███▌      | 453/1250 [12:17<14:58,  1.13s/it, loss=0.871]
Epoch 1/4:  36%|███▋      | 454/1250 [12:17<14:32,  1.10s/it, loss=0.871]
Epoch 1/4:  36%|███▋      | 454/1250 [12:18<14:32,  1.10s/it, loss=1.22] 
Epoch 1/4:  36%|███▋      | 455/1250 [12:18<14:12,  1.07s/it, loss=1.22]
Epoch 1/4:  36%|███▋      | 455/1250 [12:19<14:12,  1.07s/it, loss=0.621]
Epoch 1/4:  36%|███▋      | 456/1250 [12:19<14:36,  1.10s/it, loss=0.621]
Epoch 1/4:  36%|███▋      | 456/1250 [12:21<14:36,  1.10s/it, loss=0.916]
Epoch 1/4:  37%|███▋      | 457/1250 [12:21<15:05,  1.14s/it, loss=0.916]
Epoch 1/4:  37%|███▋      | 457/1250 [12:22<15:05,  1.14s/it, loss=0.73] 
Epoch 1/4:  37%|███▋      | 458/1250 [12:22<15:09,  1.15s/it, loss=0.73]
Epoch 1/4:  37%|███▋      | 458/1250 [12:23<15:09,  1.15s/it, loss=0.488]
Epoch 1/4:  37%|███▋      | 459/1250 [12:23<14:38,  1.11s/it, loss=0.488]
Epoch 1/4:  37%|███▋      | 459/1250 [12:24<14:38,  1.11s/it, loss=0.866]
Epoch 1/4:  37%|███▋      | 460/1250 [12:24<14:32,  1.10s/it, loss=0.866]
Epoch 1/4:  37%|███▋      | 460/1250 [12:25<14:32,  1.10s/it, loss=0.909]
Epoch 1/4:  37%|███▋      | 461/1250 [12:25<14:58,  1.14s/it, loss=0.909]
Epoch 1/4:  37%|███▋      | 461/1250 [12:26<14:58,  1.14s/it, loss=1.02] 
Epoch 1/4:  37%|███▋      | 462/1250 [12:26<14:37,  1.11s/it, loss=1.02]
Epoch 1/4:  37%|███▋      | 462/1250 [12:27<14:37,  1.11s/it, loss=0.974]
Epoch 1/4:  37%|███▋      | 463/1250 [12:27<14:09,  1.08s/it, loss=0.974]
Epoch 1/4:  37%|███▋      | 463/1250 [12:28<14:09,  1.08s/it, loss=1.39] 
Epoch 1/4:  37%|███▋      | 464/1250 [12:28<14:00,  1.07s/it, loss=1.39]
Epoch 1/4:  37%|███▋      | 464/1250 [12:30<14:00,  1.07s/it, loss=0.745]
Epoch 1/4:  37%|███▋      | 465/1250 [12:30<14:58,  1.14s/it, loss=0.745]
Epoch 1/4:  37%|███▋      | 465/1250 [12:31<14:58,  1.14s/it, loss=0.948]
Epoch 1/4:  37%|███▋      | 466/1250 [12:31<15:21,  1.18s/it, loss=0.948]
Epoch 1/4:  37%|███▋      | 466/1250 [12:32<15:21,  1.18s/it, loss=0.658]
Epoch 1/4:  37%|███▋      | 467/1250 [12:32<15:09,  1.16s/it, loss=0.658]
Epoch 1/4:  37%|███▋      | 467/1250 [12:33<15:09,  1.16s/it, loss=0.87] 
Epoch 1/4:  37%|███▋      | 468/1250 [12:33<14:39,  1.12s/it, loss=0.87]
Epoch 1/4:  37%|███▋      | 468/1250 [12:34<14:39,  1.12s/it, loss=0.821]
Epoch 1/4:  38%|███▊      | 469/1250 [12:34<14:45,  1.13s/it, loss=0.821]
Epoch 1/4:  38%|███▊      | 469/1250 [12:35<14:45,  1.13s/it, loss=1.78] 
Epoch 1/4:  38%|███▊      | 470/1250 [12:35<15:23,  1.18s/it, loss=1.78]
Epoch 1/4:  38%|███▊      | 470/1250 [12:36<15:23,  1.18s/it, loss=0.798]
Epoch 1/4:  38%|███▊      | 471/1250 [12:36<14:44,  1.14s/it, loss=0.798]
Epoch 1/4:  38%|███▊      | 471/1250 [12:37<14:44,  1.14s/it, loss=0.707]
Epoch 1/4:  38%|███▊      | 472/1250 [12:37<13:57,  1.08s/it, loss=0.707]
Epoch 1/4:  38%|███▊      | 472/1250 [12:39<13:57,  1.08s/it, loss=0.85] 
Epoch 1/4:  38%|███▊      | 473/1250 [12:39<14:30,  1.12s/it, loss=0.85]
Epoch 1/4:  38%|███▊      | 473/1250 [12:41<14:30,  1.12s/it, loss=0.947]
Epoch 1/4:  38%|███▊      | 474/1250 [12:41<18:19,  1.42s/it, loss=0.947]
Epoch 1/4:  38%|███▊      | 474/1250 [12:42<18:19,  1.42s/it, loss=0.68] 
Epoch 1/4:  38%|███▊      | 475/1250 [12:42<16:59,  1.32s/it, loss=0.68]
Epoch 1/4:  38%|███▊      | 475/1250 [12:43<16:59,  1.32s/it, loss=0.655]
Epoch 1/4:  38%|███▊      | 476/1250 [12:43<15:41,  1.22s/it, loss=0.655]
Epoch 1/4:  38%|███▊      | 476/1250 [12:44<15:41,  1.22s/it, loss=0.78] 
Epoch 1/4:  38%|███▊      | 477/1250 [12:44<14:28,  1.12s/it, loss=0.78]
Epoch 1/4:  38%|███▊      | 477/1250 [12:45<14:28,  1.12s/it, loss=0.986]
Epoch 1/4:  38%|███▊      | 478/1250 [12:45<13:57,  1.08s/it, loss=0.986]
Epoch 1/4:  38%|███▊      | 478/1250 [12:46<13:57,  1.08s/it, loss=1.12] 
Epoch 1/4:  38%|███▊      | 479/1250 [12:46<14:33,  1.13s/it, loss=1.12]
Epoch 1/4:  38%|███▊      | 479/1250 [12:47<14:33,  1.13s/it, loss=1.46]
Epoch 1/4:  38%|███▊      | 480/1250 [12:47<15:49,  1.23s/it, loss=1.46]
Epoch 1/4:  38%|███▊      | 480/1250 [12:50<15:49,  1.23s/it, loss=1.17]
Epoch 1/4:  38%|███▊      | 481/1250 [12:50<20:13,  1.58s/it, loss=1.17]
Epoch 1/4:  38%|███▊      | 481/1250 [12:51<20:13,  1.58s/it, loss=1.03]
Epoch 1/4:  39%|███▊      | 482/1250 [12:51<18:18,  1.43s/it, loss=1.03]
Epoch 1/4:  39%|███▊      | 482/1250 [12:52<18:18,  1.43s/it, loss=1.24]
Epoch 1/4:  39%|███▊      | 483/1250 [12:52<17:08,  1.34s/it, loss=1.24]
Epoch 1/4:  39%|███▊      | 483/1250 [12:53<17:08,  1.34s/it, loss=1.36]
Epoch 1/4:  39%|███▊      | 484/1250 [12:53<17:25,  1.36s/it, loss=1.36]
Epoch 1/4:  39%|███▊      | 484/1250 [12:55<17:25,  1.36s/it, loss=0.877]
Epoch 1/4:  39%|███▉      | 485/1250 [12:55<17:13,  1.35s/it, loss=0.877]
Epoch 1/4:  39%|███▉      | 485/1250 [12:56<17:13,  1.35s/it, loss=1]    
Epoch 1/4:  39%|███▉      | 486/1250 [12:56<16:25,  1.29s/it, loss=1]
Epoch 1/4:  39%|███▉      | 486/1250 [12:57<16:25,  1.29s/it, loss=0.979]
Epoch 1/4:  39%|███▉      | 487/1250 [12:57<15:54,  1.25s/it, loss=0.979]
Epoch 1/4:  39%|███▉      | 487/1250 [12:58<15:54,  1.25s/it, loss=0.98] 
Epoch 1/4:  39%|███▉      | 488/1250 [12:58<16:04,  1.27s/it, loss=0.98]
Epoch 1/4:  39%|███▉      | 488/1250 [12:59<16:04,  1.27s/it, loss=1.27]
Epoch 1/4:  39%|███▉      | 489/1250 [12:59<15:25,  1.22s/it, loss=1.27]
Epoch 1/4:  39%|███▉      | 489/1250 [13:01<15:25,  1.22s/it, loss=1.45]
Epoch 1/4:  39%|███▉      | 490/1250 [13:01<15:04,  1.19s/it, loss=1.45]
Epoch 1/4:  39%|███▉      | 490/1250 [13:02<15:04,  1.19s/it, loss=0.782]
Epoch 1/4:  39%|███▉      | 491/1250 [13:02<14:54,  1.18s/it, loss=0.782]
Epoch 1/4:  39%|███▉      | 491/1250 [13:03<14:54,  1.18s/it, loss=1.14] 
Epoch 1/4:  39%|███▉      | 492/1250 [13:03<15:33,  1.23s/it, loss=1.14]
Epoch 1/4:  39%|███▉      | 492/1250 [13:04<15:33,  1.23s/it, loss=0.923]
Epoch 1/4:  39%|███▉      | 493/1250 [13:04<15:08,  1.20s/it, loss=0.923]
Epoch 1/4:  39%|███▉      | 493/1250 [13:05<15:08,  1.20s/it, loss=0.713]
Epoch 1/4:  40%|███▉      | 494/1250 [13:05<14:40,  1.16s/it, loss=0.713]
Epoch 1/4:  40%|███▉      | 494/1250 [13:06<14:40,  1.16s/it, loss=1.31] 
Epoch 1/4:  40%|███▉      | 495/1250 [13:06<14:23,  1.14s/it, loss=1.31]
Epoch 1/4:  40%|███▉      | 495/1250 [13:08<14:23,  1.14s/it, loss=0.801]
Epoch 1/4:  40%|███▉      | 496/1250 [13:08<14:13,  1.13s/it, loss=0.801]
Epoch 1/4:  40%|███▉      | 496/1250 [13:09<14:13,  1.13s/it, loss=0.699]
Epoch 1/4:  40%|███▉      | 497/1250 [13:09<16:30,  1.31s/it, loss=0.699]
Epoch 1/4:  40%|███▉      | 497/1250 [13:10<16:30,  1.31s/it, loss=0.984]
Epoch 1/4:  40%|███▉      | 498/1250 [13:10<15:42,  1.25s/it, loss=0.984]
Epoch 1/4:  40%|███▉      | 498/1250 [13:11<15:42,  1.25s/it, loss=0.745]
Epoch 1/4:  40%|███▉      | 499/1250 [13:11<15:07,  1.21s/it, loss=0.745]
Epoch 1/4:  40%|███▉      | 499/1250 [13:13<15:07,  1.21s/it, loss=0.885]
Epoch 1/4:  40%|████      | 500/1250 [13:13<14:29,  1.16s/it, loss=0.885]
Epoch 1/4:  40%|████      | 500/1250 [13:14<14:29,  1.16s/it, loss=0.987]
Epoch 1/4:  40%|████      | 501/1250 [13:14<15:02,  1.21s/it, loss=0.987]
Epoch 1/4:  40%|████      | 501/1250 [13:15<15:02,  1.21s/it, loss=0.738]
Epoch 1/4:  40%|████      | 502/1250 [13:15<14:39,  1.18s/it, loss=0.738]
Epoch 1/4:  40%|████      | 502/1250 [13:16<14:39,  1.18s/it, loss=1.15] 
Epoch 1/4:  40%|████      | 503/1250 [13:16<14:10,  1.14s/it, loss=1.15]
Epoch 1/4:  40%|████      | 503/1250 [13:17<14:10,  1.14s/it, loss=0.804]
Epoch 1/4:  40%|████      | 504/1250 [13:17<13:37,  1.10s/it, loss=0.804]
Epoch 1/4:  40%|████      | 504/1250 [13:18<13:37,  1.10s/it, loss=0.742]
Epoch 1/4:  40%|████      | 505/1250 [13:18<13:01,  1.05s/it, loss=0.742]
Epoch 1/4:  40%|████      | 505/1250 [13:19<13:01,  1.05s/it, loss=0.808]
Epoch 1/4:  40%|████      | 506/1250 [13:19<13:05,  1.06s/it, loss=0.808]
Epoch 1/4:  40%|████      | 506/1250 [13:20<13:05,  1.06s/it, loss=1.06] 
Epoch 1/4:  41%|████      | 507/1250 [13:20<13:07,  1.06s/it, loss=1.06]
Epoch 1/4:  41%|████      | 507/1250 [13:21<13:07,  1.06s/it, loss=0.824]
Epoch 1/4:  41%|████      | 508/1250 [13:21<13:07,  1.06s/it, loss=0.824]
Epoch 1/4:  41%|████      | 508/1250 [13:22<13:07,  1.06s/it, loss=0.802]
Epoch 1/4:  41%|████      | 509/1250 [13:22<13:03,  1.06s/it, loss=0.802]
Epoch 1/4:  41%|████      | 509/1250 [13:23<13:03,  1.06s/it, loss=0.977]
Epoch 1/4:  41%|████      | 510/1250 [13:23<13:08,  1.07s/it, loss=0.977]
Epoch 1/4:  41%|████      | 510/1250 [13:24<13:08,  1.07s/it, loss=0.745]
Epoch 1/4:  41%|████      | 511/1250 [13:24<13:25,  1.09s/it, loss=0.745]
Epoch 1/4:  41%|████      | 511/1250 [13:26<13:25,  1.09s/it, loss=0.689]
Epoch 1/4:  41%|████      | 512/1250 [13:26<13:32,  1.10s/it, loss=0.689]
Epoch 1/4:  41%|████      | 512/1250 [13:27<13:32,  1.10s/it, loss=1.18] 
Epoch 1/4:  41%|████      | 513/1250 [13:27<13:36,  1.11s/it, loss=1.18]
Epoch 1/4:  41%|████      | 513/1250 [13:28<13:36,  1.11s/it, loss=1.02]
Epoch 1/4:  41%|████      | 514/1250 [13:28<13:26,  1.10s/it, loss=1.02]
Epoch 1/4:  41%|████      | 514/1250 [13:29<13:26,  1.10s/it, loss=0.813]
Epoch 1/4:  41%|████      | 515/1250 [13:29<13:14,  1.08s/it, loss=0.813]
Epoch 1/4:  41%|████      | 515/1250 [13:30<13:14,  1.08s/it, loss=0.925]
Epoch 1/4:  41%|████▏     | 516/1250 [13:30<13:23,  1.10s/it, loss=0.925]
Epoch 1/4:  41%|████▏     | 516/1250 [13:31<13:23,  1.10s/it, loss=0.55] 
Epoch 1/4:  41%|████▏     | 517/1250 [13:31<14:10,  1.16s/it, loss=0.55]
Epoch 1/4:  41%|████▏     | 517/1250 [13:33<14:10,  1.16s/it, loss=1.32]
Epoch 1/4:  41%|████▏     | 518/1250 [13:33<14:47,  1.21s/it, loss=1.32]
Epoch 1/4:  41%|████▏     | 518/1250 [13:34<14:47,  1.21s/it, loss=0.938]
Epoch 1/4:  42%|████▏     | 519/1250 [13:34<14:41,  1.21s/it, loss=0.938]
Epoch 1/4:  42%|████▏     | 519/1250 [13:35<14:41,  1.21s/it, loss=0.741]
Epoch 1/4:  42%|████▏     | 520/1250 [13:35<14:54,  1.22s/it, loss=0.741]
Epoch 1/4:  42%|████▏     | 520/1250 [13:36<14:54,  1.22s/it, loss=0.492]
Epoch 1/4:  42%|████▏     | 521/1250 [13:36<14:32,  1.20s/it, loss=0.492]
Epoch 1/4:  42%|████▏     | 521/1250 [13:37<14:32,  1.20s/it, loss=1.16] 
Epoch 1/4:  42%|████▏     | 522/1250 [13:37<14:06,  1.16s/it, loss=1.16]
Epoch 1/4:  42%|████▏     | 522/1250 [13:38<14:06,  1.16s/it, loss=0.849]
Epoch 1/4:  42%|████▏     | 523/1250 [13:38<13:36,  1.12s/it, loss=0.849]
Epoch 1/4:  42%|████▏     | 523/1250 [13:39<13:36,  1.12s/it, loss=1.15] 
Epoch 1/4:  42%|████▏     | 524/1250 [13:39<13:56,  1.15s/it, loss=1.15]
Epoch 1/4:  42%|████▏     | 524/1250 [13:41<13:56,  1.15s/it, loss=0.696]
Epoch 1/4:  42%|████▏     | 525/1250 [13:41<14:08,  1.17s/it, loss=0.696]
Epoch 1/4:  42%|████▏     | 525/1250 [13:42<14:08,  1.17s/it, loss=0.864]
Epoch 1/4:  42%|████▏     | 526/1250 [13:42<13:52,  1.15s/it, loss=0.864]
Epoch 1/4:  42%|████▏     | 526/1250 [13:43<13:52,  1.15s/it, loss=0.67] 
Epoch 1/4:  42%|████▏     | 527/1250 [13:43<13:23,  1.11s/it, loss=0.67]
Epoch 1/4:  42%|████▏     | 527/1250 [13:44<13:23,  1.11s/it, loss=0.82]
Epoch 1/4:  42%|████▏     | 528/1250 [13:44<12:57,  1.08s/it, loss=0.82]
Epoch 1/4:  42%|████▏     | 528/1250 [13:45<12:57,  1.08s/it, loss=0.871]
Epoch 1/4:  42%|████▏     | 529/1250 [13:45<14:01,  1.17s/it, loss=0.871]
Epoch 1/4:  42%|████▏     | 529/1250 [13:46<14:01,  1.17s/it, loss=0.769]
Epoch 1/4:  42%|████▏     | 530/1250 [13:46<13:58,  1.16s/it, loss=0.769]
Epoch 1/4:  42%|████▏     | 530/1250 [13:47<13:58,  1.16s/it, loss=0.811]
Epoch 1/4:  42%|████▏     | 531/1250 [13:47<13:30,  1.13s/it, loss=0.811]
Epoch 1/4:  42%|████▏     | 531/1250 [13:48<13:30,  1.13s/it, loss=1.06] 
Epoch 1/4:  43%|████▎     | 532/1250 [13:48<12:58,  1.08s/it, loss=1.06]
Epoch 1/4:  43%|████▎     | 532/1250 [13:49<12:58,  1.08s/it, loss=1.22]
Epoch 1/4:  43%|████▎     | 533/1250 [13:49<12:57,  1.08s/it, loss=1.22]
Epoch 1/4:  43%|████▎     | 533/1250 [13:51<12:57,  1.08s/it, loss=1.01]
Epoch 1/4:  43%|████▎     | 534/1250 [13:51<13:49,  1.16s/it, loss=1.01]
Epoch 1/4:  43%|████▎     | 534/1250 [13:52<13:49,  1.16s/it, loss=0.929]
Epoch 1/4:  43%|████▎     | 535/1250 [13:52<13:50,  1.16s/it, loss=0.929]
Epoch 1/4:  43%|████▎     | 535/1250 [13:53<13:50,  1.16s/it, loss=0.663]
Epoch 1/4:  43%|████▎     | 536/1250 [13:53<13:19,  1.12s/it, loss=0.663]
Epoch 1/4:  43%|████▎     | 536/1250 [13:54<13:19,  1.12s/it, loss=0.805]
Epoch 1/4:  43%|████▎     | 537/1250 [13:54<12:43,  1.07s/it, loss=0.805]
Epoch 1/4:  43%|████▎     | 537/1250 [13:55<12:43,  1.07s/it, loss=0.603]
Epoch 1/4:  43%|████▎     | 538/1250 [13:55<12:52,  1.09s/it, loss=0.603]
Epoch 1/4:  43%|████▎     | 538/1250 [13:56<12:52,  1.09s/it, loss=0.722]
Epoch 1/4:  43%|████▎     | 539/1250 [13:56<12:57,  1.09s/it, loss=0.722]
Epoch 1/4:  43%|████▎     | 539/1250 [13:57<12:57,  1.09s/it, loss=1]    
Epoch 1/4:  43%|████▎     | 540/1250 [13:57<12:30,  1.06s/it, loss=1]
Epoch 1/4:  43%|████▎     | 540/1250 [13:58<12:30,  1.06s/it, loss=1.04]
Epoch 1/4:  43%|████▎     | 541/1250 [13:58<11:56,  1.01s/it, loss=1.04]
Epoch 1/4:  43%|████▎     | 541/1250 [13:59<11:56,  1.01s/it, loss=0.828]
Epoch 1/4:  43%|████▎     | 542/1250 [13:59<11:26,  1.03it/s, loss=0.828]
Epoch 1/4:  43%|████▎     | 542/1250 [14:00<11:26,  1.03it/s, loss=0.866]
Epoch 1/4:  43%|████▎     | 543/1250 [14:00<11:43,  1.01it/s, loss=0.866]
Epoch 1/4:  43%|████▎     | 543/1250 [14:01<11:43,  1.01it/s, loss=0.776]
Epoch 1/4:  44%|████▎     | 544/1250 [14:01<12:38,  1.07s/it, loss=0.776]
Epoch 1/4:  44%|████▎     | 544/1250 [14:02<12:38,  1.07s/it, loss=0.911]
Epoch 1/4:  44%|████▎     | 545/1250 [14:02<12:46,  1.09s/it, loss=0.911]
Epoch 1/4:  44%|████▎     | 545/1250 [14:03<12:46,  1.09s/it, loss=1.02] 
Epoch 1/4:  44%|████▎     | 546/1250 [14:03<12:10,  1.04s/it, loss=1.02]
Epoch 1/4:  44%|████▎     | 546/1250 [14:04<12:10,  1.04s/it, loss=1.12]
Epoch 1/4:  44%|████▍     | 547/1250 [14:04<11:37,  1.01it/s, loss=1.12]
Epoch 1/4:  44%|████▍     | 547/1250 [14:05<11:37,  1.01it/s, loss=1.07]
Epoch 1/4:  44%|████▍     | 548/1250 [14:05<12:10,  1.04s/it, loss=1.07]
Epoch 1/4:  44%|████▍     | 548/1250 [14:07<12:10,  1.04s/it, loss=0.707]
Epoch 1/4:  44%|████▍     | 549/1250 [14:07<12:46,  1.09s/it, loss=0.707]
Epoch 1/4:  44%|████▍     | 549/1250 [14:07<12:46,  1.09s/it, loss=1.19] 
Epoch 1/4:  44%|████▍     | 550/1250 [14:07<12:19,  1.06s/it, loss=1.19]
Epoch 1/4:  44%|████▍     | 550/1250 [14:08<12:19,  1.06s/it, loss=0.737]
Epoch 1/4:  44%|████▍     | 551/1250 [14:08<11:55,  1.02s/it, loss=0.737]
Epoch 1/4:  44%|████▍     | 551/1250 [14:09<11:55,  1.02s/it, loss=0.661]
Epoch 1/4:  44%|████▍     | 552/1250 [14:09<11:28,  1.01it/s, loss=0.661]
Epoch 1/4:  44%|████▍     | 552/1250 [14:10<11:28,  1.01it/s, loss=1.08] 
Epoch 1/4:  44%|████▍     | 553/1250 [14:10<11:58,  1.03s/it, loss=1.08]
Epoch 1/4:  44%|████▍     | 553/1250 [14:12<11:58,  1.03s/it, loss=1.32]
Epoch 1/4:  44%|████▍     | 554/1250 [14:12<12:34,  1.08s/it, loss=1.32]
Epoch 1/4:  44%|████▍     | 554/1250 [14:13<12:34,  1.08s/it, loss=0.666]
Epoch 1/4:  44%|████▍     | 555/1250 [14:13<12:44,  1.10s/it, loss=0.666]
Epoch 1/4:  44%|████▍     | 555/1250 [14:14<12:44,  1.10s/it, loss=1.29] 
Epoch 1/4:  44%|████▍     | 556/1250 [14:14<12:11,  1.05s/it, loss=1.29]
Epoch 1/4:  44%|████▍     | 556/1250 [14:15<12:11,  1.05s/it, loss=0.76]
Epoch 1/4:  45%|████▍     | 557/1250 [14:15<11:35,  1.00s/it, loss=0.76]
Epoch 1/4:  45%|████▍     | 557/1250 [14:16<11:35,  1.00s/it, loss=0.956]
Epoch 1/4:  45%|████▍     | 558/1250 [14:16<12:21,  1.07s/it, loss=0.956]
Epoch 1/4:  45%|████▍     | 558/1250 [14:17<12:21,  1.07s/it, loss=0.803]
Epoch 1/4:  45%|████▍     | 559/1250 [14:17<12:58,  1.13s/it, loss=0.803]
Epoch 1/4:  45%|████▍     | 559/1250 [14:18<12:58,  1.13s/it, loss=0.823]
Epoch 1/4:  45%|████▍     | 560/1250 [14:18<12:17,  1.07s/it, loss=0.823]
Epoch 1/4:  45%|████▍     | 560/1250 [14:19<12:17,  1.07s/it, loss=0.719]
Epoch 1/4:  45%|████▍     | 561/1250 [14:19<11:30,  1.00s/it, loss=0.719]
Epoch 1/4:  45%|████▍     | 561/1250 [14:20<11:30,  1.00s/it, loss=1.44] 
Epoch 1/4:  45%|████▍     | 562/1250 [14:20<10:58,  1.05it/s, loss=1.44]
Epoch 1/4:  45%|████▍     | 562/1250 [14:21<10:58,  1.05it/s, loss=1.18]
Epoch 1/4:  45%|████▌     | 563/1250 [14:21<12:08,  1.06s/it, loss=1.18]
Epoch 1/4:  45%|████▌     | 563/1250 [14:22<12:08,  1.06s/it, loss=0.731]
Epoch 1/4:  45%|████▌     | 564/1250 [14:22<12:18,  1.08s/it, loss=0.731]
Epoch 1/4:  45%|████▌     | 564/1250 [14:23<12:18,  1.08s/it, loss=0.813]
Epoch 1/4:  45%|████▌     | 565/1250 [14:23<11:43,  1.03s/it, loss=0.813]
Epoch 1/4:  45%|████▌     | 565/1250 [14:24<11:43,  1.03s/it, loss=0.961]
Epoch 1/4:  45%|████▌     | 566/1250 [14:24<11:10,  1.02it/s, loss=0.961]
Epoch 1/4:  45%|████▌     | 566/1250 [14:25<11:10,  1.02it/s, loss=1.04] 
Epoch 1/4:  45%|████▌     | 567/1250 [14:25<10:37,  1.07it/s, loss=1.04]
Epoch 1/4:  45%|████▌     | 567/1250 [14:26<10:37,  1.07it/s, loss=1.04]
Epoch 1/4:  45%|████▌     | 568/1250 [14:26<11:09,  1.02it/s, loss=1.04]
Epoch 1/4:  45%|████▌     | 568/1250 [14:27<11:09,  1.02it/s, loss=0.791]
Epoch 1/4:  46%|████▌     | 569/1250 [14:27<12:37,  1.11s/it, loss=0.791]
Epoch 1/4:  46%|████▌     | 569/1250 [14:28<12:37,  1.11s/it, loss=0.979]
Epoch 1/4:  46%|████▌     | 570/1250 [14:28<12:02,  1.06s/it, loss=0.979]
Epoch 1/4:  46%|████▌     | 570/1250 [14:29<12:02,  1.06s/it, loss=0.891]
Epoch 1/4:  46%|████▌     | 571/1250 [14:29<11:29,  1.02s/it, loss=0.891]
Epoch 1/4:  46%|████▌     | 571/1250 [14:30<11:29,  1.02s/it, loss=0.985]
Epoch 1/4:  46%|████▌     | 572/1250 [14:30<11:05,  1.02it/s, loss=0.985]
Epoch 1/4:  46%|████▌     | 572/1250 [14:31<11:05,  1.02it/s, loss=1.35] 
Epoch 1/4:  46%|████▌     | 573/1250 [14:31<11:47,  1.05s/it, loss=1.35]
Epoch 1/4:  46%|████▌     | 573/1250 [14:32<11:47,  1.05s/it, loss=1.28]
Epoch 1/4:  46%|████▌     | 574/1250 [14:32<12:16,  1.09s/it, loss=1.28]
Epoch 1/4:  46%|████▌     | 574/1250 [14:33<12:16,  1.09s/it, loss=1.24]
Epoch 1/4:  46%|████▌     | 575/1250 [14:33<11:31,  1.02s/it, loss=1.24]
Epoch 1/4:  46%|████▌     | 575/1250 [14:34<11:31,  1.02s/it, loss=0.705]
Epoch 1/4:  46%|████▌     | 576/1250 [14:34<10:48,  1.04it/s, loss=0.705]
Epoch 1/4:  46%|████▌     | 576/1250 [14:35<10:48,  1.04it/s, loss=1.23] 
Epoch 1/4:  46%|████▌     | 577/1250 [14:35<10:26,  1.07it/s, loss=1.23]
Epoch 1/4:  46%|████▌     | 577/1250 [14:36<10:26,  1.07it/s, loss=0.881]
Epoch 1/4:  46%|████▌     | 578/1250 [14:36<11:11,  1.00it/s, loss=0.881]
Epoch 1/4:  46%|████▌     | 578/1250 [14:37<11:11,  1.00it/s, loss=0.973]
Epoch 1/4:  46%|████▋     | 579/1250 [14:37<11:50,  1.06s/it, loss=0.973]
Epoch 1/4:  46%|████▋     | 579/1250 [14:38<11:50,  1.06s/it, loss=0.851]
Epoch 1/4:  46%|████▋     | 580/1250 [14:38<11:20,  1.02s/it, loss=0.851]
Epoch 1/4:  46%|████▋     | 580/1250 [14:39<11:20,  1.02s/it, loss=0.991]
Epoch 1/4:  46%|████▋     | 581/1250 [14:39<11:02,  1.01it/s, loss=0.991]
Epoch 1/4:  46%|████▋     | 581/1250 [14:40<11:02,  1.01it/s, loss=0.861]
Epoch 1/4:  47%|████▋     | 582/1250 [14:40<10:25,  1.07it/s, loss=0.861]
Epoch 1/4:  47%|████▋     | 582/1250 [14:41<10:25,  1.07it/s, loss=1.13] 
Epoch 1/4:  47%|████▋     | 583/1250 [14:41<10:33,  1.05it/s, loss=1.13]
Epoch 1/4:  47%|████▋     | 583/1250 [14:42<10:33,  1.05it/s, loss=1.1] 
Epoch 1/4:  47%|████▋     | 584/1250 [14:42<11:36,  1.05s/it, loss=1.1]
Epoch 1/4:  47%|████▋     | 584/1250 [14:43<11:36,  1.05s/it, loss=0.767]
Epoch 1/4:  47%|████▋     | 585/1250 [14:43<11:01,  1.01it/s, loss=0.767]
Epoch 1/4:  47%|████▋     | 585/1250 [14:44<11:01,  1.01it/s, loss=0.757]
Epoch 1/4:  47%|████▋     | 586/1250 [14:44<10:25,  1.06it/s, loss=0.757]
Epoch 1/4:  47%|████▋     | 586/1250 [14:45<10:25,  1.06it/s, loss=0.745]
Epoch 1/4:  47%|████▋     | 587/1250 [14:45<10:01,  1.10it/s, loss=0.745]
Epoch 1/4:  47%|████▋     | 587/1250 [14:46<10:01,  1.10it/s, loss=1.06] 
Epoch 1/4:  47%|████▋     | 588/1250 [14:46<10:12,  1.08it/s, loss=1.06]
Epoch 1/4:  47%|████▋     | 588/1250 [14:47<10:12,  1.08it/s, loss=0.496]
Epoch 1/4:  47%|████▋     | 589/1250 [14:47<11:10,  1.01s/it, loss=0.496]
Epoch 1/4:  47%|████▋     | 589/1250 [14:48<11:10,  1.01s/it, loss=0.629]
Epoch 1/4:  47%|████▋     | 590/1250 [14:48<10:59,  1.00it/s, loss=0.629]
Epoch 1/4:  47%|████▋     | 590/1250 [14:49<10:59,  1.00it/s, loss=0.575]
Epoch 1/4:  47%|████▋     | 591/1250 [14:49<10:22,  1.06it/s, loss=0.575]
Epoch 1/4:  47%|████▋     | 591/1250 [14:50<10:22,  1.06it/s, loss=1.62] 
Epoch 1/4:  47%|████▋     | 592/1250 [14:50<09:56,  1.10it/s, loss=1.62]
Epoch 1/4:  47%|████▋     | 592/1250 [14:50<09:56,  1.10it/s, loss=0.742]
Epoch 1/4:  47%|████▋     | 593/1250 [14:50<09:56,  1.10it/s, loss=0.742]
Epoch 1/4:  47%|████▋     | 593/1250 [14:52<09:56,  1.10it/s, loss=1.02] 
Epoch 1/4:  48%|████▊     | 594/1250 [14:52<10:46,  1.01it/s, loss=1.02]
Epoch 1/4:  48%|████▊     | 594/1250 [14:53<10:46,  1.01it/s, loss=1]   
Epoch 1/4:  48%|████▊     | 595/1250 [14:53<11:29,  1.05s/it, loss=1]
Epoch 1/4:  48%|████▊     | 595/1250 [14:54<11:29,  1.05s/it, loss=0.843]
Epoch 1/4:  48%|████▊     | 596/1250 [14:54<10:44,  1.01it/s, loss=0.843]
Epoch 1/4:  48%|████▊     | 596/1250 [14:54<10:44,  1.01it/s, loss=0.874]
Epoch 1/4:  48%|████▊     | 597/1250 [14:54<10:00,  1.09it/s, loss=0.874]
Epoch 1/4:  48%|████▊     | 597/1250 [14:55<10:00,  1.09it/s, loss=1.21] 
Epoch 1/4:  48%|████▊     | 598/1250 [14:55<09:47,  1.11it/s, loss=1.21]
Epoch 1/4:  48%|████▊     | 598/1250 [14:56<09:47,  1.11it/s, loss=0.979]
Epoch 1/4:  48%|████▊     | 599/1250 [14:56<10:26,  1.04it/s, loss=0.979]
Epoch 1/4:  48%|████▊     | 599/1250 [14:58<10:26,  1.04it/s, loss=1]    
Epoch 1/4:  48%|████▊     | 600/1250 [14:58<11:41,  1.08s/it, loss=1]
Epoch 1/4:  48%|████▊     | 600/1250 [14:59<11:41,  1.08s/it, loss=0.802]
Epoch 1/4:  48%|████▊     | 601/1250 [14:59<11:08,  1.03s/it, loss=0.802]
Epoch 1/4:  48%|████▊     | 601/1250 [14:59<11:08,  1.03s/it, loss=0.684]
Epoch 1/4:  48%|████▊     | 602/1250 [14:59<10:25,  1.04it/s, loss=0.684]
Epoch 1/4:  48%|████▊     | 602/1250 [15:00<10:25,  1.04it/s, loss=1.01] 
Epoch 1/4:  48%|████▊     | 603/1250 [15:00<10:04,  1.07it/s, loss=1.01]
Epoch 1/4:  48%|████▊     | 603/1250 [15:02<10:04,  1.07it/s, loss=1.15]
Epoch 1/4:  48%|████▊     | 604/1250 [15:02<10:55,  1.01s/it, loss=1.15]
Epoch 1/4:  48%|████▊     | 604/1250 [15:03<10:55,  1.01s/it, loss=1.13]
Epoch 1/4:  48%|████▊     | 605/1250 [15:03<12:21,  1.15s/it, loss=1.13]
Epoch 1/4:  48%|████▊     | 605/1250 [15:04<12:21,  1.15s/it, loss=1.19]
Epoch 1/4:  48%|████▊     | 606/1250 [15:04<12:04,  1.12s/it, loss=1.19]
Epoch 1/4:  48%|████▊     | 606/1250 [15:05<12:04,  1.12s/it, loss=0.724]
Epoch 1/4:  49%|████▊     | 607/1250 [15:05<11:19,  1.06s/it, loss=0.724]
Epoch 1/4:  49%|████▊     | 607/1250 [15:06<11:19,  1.06s/it, loss=0.712]
Epoch 1/4:  49%|████▊     | 608/1250 [15:06<10:48,  1.01s/it, loss=0.712]
Epoch 1/4:  49%|████▊     | 608/1250 [15:07<10:48,  1.01s/it, loss=0.784]
Epoch 1/4:  49%|████▊     | 609/1250 [15:07<10:45,  1.01s/it, loss=0.784]
Epoch 1/4:  49%|████▊     | 609/1250 [15:08<10:45,  1.01s/it, loss=1.08] 
Epoch 1/4:  49%|████▉     | 610/1250 [15:08<12:03,  1.13s/it, loss=1.08]
Epoch 1/4:  49%|████▉     | 610/1250 [15:10<12:03,  1.13s/it, loss=1.05]
Epoch 1/4:  49%|████▉     | 611/1250 [15:10<12:31,  1.18s/it, loss=1.05]
Epoch 1/4:  49%|████▉     | 611/1250 [15:11<12:31,  1.18s/it, loss=0.633]
Epoch 1/4:  49%|████▉     | 612/1250 [15:11<11:59,  1.13s/it, loss=0.633]
Epoch 1/4:  49%|████▉     | 612/1250 [15:12<11:59,  1.13s/it, loss=1.26] 
Epoch 1/4:  49%|████▉     | 613/1250 [15:12<11:29,  1.08s/it, loss=1.26]
Epoch 1/4:  49%|████▉     | 613/1250 [15:13<11:29,  1.08s/it, loss=1.33]
Epoch 1/4:  49%|████▉     | 614/1250 [15:13<11:06,  1.05s/it, loss=1.33]
Epoch 1/4:  49%|████▉     | 614/1250 [15:14<11:06,  1.05s/it, loss=0.804]
Epoch 1/4:  49%|████▉     | 615/1250 [15:14<11:50,  1.12s/it, loss=0.804]
Epoch 1/4:  49%|████▉     | 615/1250 [15:15<11:50,  1.12s/it, loss=0.955]
Epoch 1/4:  49%|████▉     | 616/1250 [15:15<12:12,  1.16s/it, loss=0.955]
Epoch 1/4:  49%|████▉     | 616/1250 [15:16<12:12,  1.16s/it, loss=0.598]
Epoch 1/4:  49%|████▉     | 617/1250 [15:16<11:49,  1.12s/it, loss=0.598]
Epoch 1/4:  49%|████▉     | 617/1250 [15:18<11:49,  1.12s/it, loss=1.6]  
Epoch 1/4:  49%|████▉     | 618/1250 [15:18<13:28,  1.28s/it, loss=1.6]
Epoch 1/4:  49%|████▉     | 618/1250 [15:19<13:28,  1.28s/it, loss=0.929]
Epoch 1/4:  50%|████▉     | 619/1250 [15:19<13:36,  1.29s/it, loss=0.929]
Epoch 1/4:  50%|████▉     | 619/1250 [15:20<13:36,  1.29s/it, loss=1.5]  
Epoch 1/4:  50%|████▉     | 620/1250 [15:20<12:39,  1.21s/it, loss=1.5]
Epoch 1/4:  50%|████▉     | 620/1250 [15:21<12:39,  1.21s/it, loss=0.572]
Epoch 1/4:  50%|████▉     | 621/1250 [15:21<12:34,  1.20s/it, loss=0.572]
Epoch 1/4:  50%|████▉     | 621/1250 [15:23<12:34,  1.20s/it, loss=1.6]  
Epoch 1/4:  50%|████▉     | 622/1250 [15:23<14:45,  1.41s/it, loss=1.6]
Epoch 1/4:  50%|████▉     | 622/1250 [15:24<14:45,  1.41s/it, loss=1.09]
Epoch 1/4:  50%|████▉     | 623/1250 [15:24<14:17,  1.37s/it, loss=1.09]
Epoch 1/4:  50%|████▉     | 623/1250 [15:26<14:17,  1.37s/it, loss=0.723]
Epoch 1/4:  50%|████▉     | 624/1250 [15:26<13:36,  1.30s/it, loss=0.723]
Epoch 1/4:  50%|████▉     | 624/1250 [15:27<13:36,  1.30s/it, loss=1.14] 
Epoch 1/4:  50%|█████     | 625/1250 [15:27<13:12,  1.27s/it, loss=1.14]
Epoch 1/4:  50%|█████     | 625/1250 [15:28<13:12,  1.27s/it, loss=0.701]
Epoch 1/4:  50%|█████     | 626/1250 [15:28<12:39,  1.22s/it, loss=0.701]
Epoch 1/4:  50%|█████     | 626/1250 [15:29<12:39,  1.22s/it, loss=1.45] 
Epoch 1/4:  50%|█████     | 627/1250 [15:29<12:29,  1.20s/it, loss=1.45]
Epoch 1/4:  50%|█████     | 627/1250 [15:30<12:29,  1.20s/it, loss=0.917]
Epoch 1/4:  50%|█████     | 628/1250 [15:30<12:16,  1.18s/it, loss=0.917]
Epoch 1/4:  50%|█████     | 628/1250 [15:31<12:16,  1.18s/it, loss=1.08] 
Epoch 1/4:  50%|█████     | 629/1250 [15:31<11:36,  1.12s/it, loss=1.08]
Epoch 1/4:  50%|█████     | 629/1250 [15:32<11:36,  1.12s/it, loss=0.815]
Epoch 1/4:  50%|█████     | 630/1250 [15:32<11:58,  1.16s/it, loss=0.815]
Epoch 1/4:  50%|█████     | 630/1250 [15:34<11:58,  1.16s/it, loss=0.957]
Epoch 1/4:  50%|█████     | 631/1250 [15:34<11:56,  1.16s/it, loss=0.957]
Epoch 1/4:  50%|█████     | 631/1250 [15:35<11:56,  1.16s/it, loss=0.582]
Epoch 1/4:  51%|█████     | 632/1250 [15:35<12:16,  1.19s/it, loss=0.582]
Epoch 1/4:  51%|█████     | 632/1250 [15:36<12:16,  1.19s/it, loss=0.702]
Epoch 1/4:  51%|█████     | 633/1250 [15:36<12:07,  1.18s/it, loss=0.702]
Epoch 1/4:  51%|█████     | 633/1250 [15:37<12:07,  1.18s/it, loss=0.676]
Epoch 1/4:  51%|█████     | 634/1250 [15:37<11:50,  1.15s/it, loss=0.676]
Epoch 1/4:  51%|█████     | 634/1250 [15:38<11:50,  1.15s/it, loss=1.42] 
Epoch 1/4:  51%|█████     | 635/1250 [15:38<11:25,  1.12s/it, loss=1.42]
Epoch 1/4:  51%|█████     | 635/1250 [15:39<11:25,  1.12s/it, loss=1.01]
Epoch 1/4:  51%|█████     | 636/1250 [15:39<11:00,  1.08s/it, loss=1.01]
Epoch 1/4:  51%|█████     | 636/1250 [15:40<11:00,  1.08s/it, loss=0.79]
Epoch 1/4:  51%|█████     | 637/1250 [15:40<11:10,  1.09s/it, loss=0.79]
Epoch 1/4:  51%|█████     | 637/1250 [15:41<11:10,  1.09s/it, loss=0.979]
Epoch 1/4:  51%|█████     | 638/1250 [15:41<10:53,  1.07s/it, loss=0.979]
Epoch 1/4:  51%|█████     | 638/1250 [15:42<10:53,  1.07s/it, loss=1.01] 
Epoch 1/4:  51%|█████     | 639/1250 [15:42<10:31,  1.03s/it, loss=1.01]
Epoch 1/4:  51%|█████     | 639/1250 [15:43<10:31,  1.03s/it, loss=0.896]
Epoch 1/4:  51%|█████     | 640/1250 [15:43<10:21,  1.02s/it, loss=0.896]
Epoch 1/4:  51%|█████     | 640/1250 [15:44<10:21,  1.02s/it, loss=1.15] 
Epoch 1/4:  51%|█████▏    | 641/1250 [15:44<10:34,  1.04s/it, loss=1.15]
Epoch 1/4:  51%|█████▏    | 641/1250 [15:46<10:34,  1.04s/it, loss=0.651]
Epoch 1/4:  51%|█████▏    | 642/1250 [15:46<11:24,  1.13s/it, loss=0.651]
Epoch 1/4:  51%|█████▏    | 642/1250 [15:47<11:24,  1.13s/it, loss=0.8]  
Epoch 1/4:  51%|█████▏    | 643/1250 [15:47<11:04,  1.09s/it, loss=0.8]
Epoch 1/4:  51%|█████▏    | 643/1250 [15:48<11:04,  1.09s/it, loss=0.544]
Epoch 1/4:  52%|█████▏    | 644/1250 [15:48<10:45,  1.06s/it, loss=0.544]
Epoch 1/4:  52%|█████▏    | 644/1250 [15:49<10:45,  1.06s/it, loss=0.831]
Epoch 1/4:  52%|█████▏    | 645/1250 [15:49<10:25,  1.03s/it, loss=0.831]
Epoch 1/4:  52%|█████▏    | 645/1250 [15:50<10:25,  1.03s/it, loss=1.18] 
Epoch 1/4:  52%|█████▏    | 646/1250 [15:50<10:19,  1.02s/it, loss=1.18]
Epoch 1/4:  52%|█████▏    | 646/1250 [15:51<10:19,  1.02s/it, loss=1.18]
Epoch 1/4:  52%|█████▏    | 647/1250 [15:51<10:50,  1.08s/it, loss=1.18]
Epoch 1/4:  52%|█████▏    | 647/1250 [15:52<10:50,  1.08s/it, loss=1.02]
Epoch 1/4:  52%|█████▏    | 648/1250 [15:52<10:37,  1.06s/it, loss=1.02]
Epoch 1/4:  52%|█████▏    | 648/1250 [15:53<10:37,  1.06s/it, loss=1.23]
Epoch 1/4:  52%|█████▏    | 649/1250 [15:53<10:22,  1.04s/it, loss=1.23]
Epoch 1/4:  52%|█████▏    | 649/1250 [15:54<10:22,  1.04s/it, loss=0.655]
Epoch 1/4:  52%|█████▏    | 650/1250 [15:54<10:08,  1.01s/it, loss=0.655]
Epoch 1/4:  52%|█████▏    | 650/1250 [15:55<10:08,  1.01s/it, loss=1.23] 
Epoch 1/4:  52%|█████▏    | 651/1250 [15:55<10:02,  1.01s/it, loss=1.23]
Epoch 1/4:  52%|█████▏    | 651/1250 [15:56<10:02,  1.01s/it, loss=0.904]
Epoch 1/4:  52%|█████▏    | 652/1250 [15:56<10:41,  1.07s/it, loss=0.904]
Epoch 1/4:  52%|█████▏    | 652/1250 [15:57<10:41,  1.07s/it, loss=0.699]
Epoch 1/4:  52%|█████▏    | 653/1250 [15:57<10:42,  1.08s/it, loss=0.699]
Epoch 1/4:  52%|█████▏    | 653/1250 [15:58<10:42,  1.08s/it, loss=0.635]
Epoch 1/4:  52%|█████▏    | 654/1250 [15:58<10:44,  1.08s/it, loss=0.635]
Epoch 1/4:  52%|█████▏    | 654/1250 [15:59<10:44,  1.08s/it, loss=1.2]  
Epoch 1/4:  52%|█████▏    | 655/1250 [15:59<10:34,  1.07s/it, loss=1.2]
Epoch 1/4:  52%|█████▏    | 655/1250 [16:00<10:34,  1.07s/it, loss=0.849]
Epoch 1/4:  52%|█████▏    | 656/1250 [16:00<10:31,  1.06s/it, loss=0.849]
Epoch 1/4:  52%|█████▏    | 656/1250 [16:01<10:31,  1.06s/it, loss=0.946]
Epoch 1/4:  53%|█████▎    | 657/1250 [16:01<10:45,  1.09s/it, loss=0.946]
Epoch 1/4:  53%|█████▎    | 657/1250 [16:02<10:45,  1.09s/it, loss=0.636]
Epoch 1/4:  53%|█████▎    | 658/1250 [16:02<10:35,  1.07s/it, loss=0.636]
Epoch 1/4:  53%|█████▎    | 658/1250 [16:03<10:35,  1.07s/it, loss=0.915]
Epoch 1/4:  53%|█████▎    | 659/1250 [16:03<10:20,  1.05s/it, loss=0.915]
Epoch 1/4:  53%|█████▎    | 659/1250 [16:04<10:20,  1.05s/it, loss=1.14] 
Epoch 1/4:  53%|█████▎    | 660/1250 [16:04<10:01,  1.02s/it, loss=1.14]
Epoch 1/4:  53%|█████▎    | 660/1250 [16:06<10:01,  1.02s/it, loss=0.782]
Epoch 1/4:  53%|█████▎    | 661/1250 [16:06<10:31,  1.07s/it, loss=0.782]
Epoch 1/4:  53%|█████▎    | 661/1250 [16:07<10:31,  1.07s/it, loss=0.806]
Epoch 1/4:  53%|█████▎    | 662/1250 [16:07<10:49,  1.10s/it, loss=0.806]
Epoch 1/4:  53%|█████▎    | 662/1250 [16:08<10:49,  1.10s/it, loss=0.708]
Epoch 1/4:  53%|█████▎    | 663/1250 [16:08<10:35,  1.08s/it, loss=0.708]
Epoch 1/4:  53%|█████▎    | 663/1250 [16:09<10:35,  1.08s/it, loss=0.803]
Epoch 1/4:  53%|█████▎    | 664/1250 [16:09<10:06,  1.04s/it, loss=0.803]
Epoch 1/4:  53%|█████▎    | 664/1250 [16:10<10:06,  1.04s/it, loss=0.846]
Epoch 1/4:  53%|█████▎    | 665/1250 [16:10<09:34,  1.02it/s, loss=0.846]
Epoch 1/4:  53%|█████▎    | 665/1250 [16:10<09:34,  1.02it/s, loss=0.685]
Epoch 1/4:  53%|█████▎    | 666/1250 [16:10<09:05,  1.07it/s, loss=0.685]
Epoch 1/4:  53%|█████▎    | 666/1250 [16:12<09:05,  1.07it/s, loss=1.26] 
Epoch 1/4:  53%|█████▎    | 667/1250 [16:12<10:20,  1.06s/it, loss=1.26]
Epoch 1/4:  53%|█████▎    | 667/1250 [16:13<10:20,  1.06s/it, loss=0.677]
Epoch 1/4:  53%|█████▎    | 668/1250 [16:13<10:43,  1.11s/it, loss=0.677]
Epoch 1/4:  53%|█████▎    | 668/1250 [16:14<10:43,  1.11s/it, loss=0.977]
Epoch 1/4:  54%|█████▎    | 669/1250 [16:14<10:10,  1.05s/it, loss=0.977]
Epoch 1/4:  54%|█████▎    | 669/1250 [16:15<10:10,  1.05s/it, loss=0.898]
Epoch 1/4:  54%|█████▎    | 670/1250 [16:15<09:31,  1.02it/s, loss=0.898]
Epoch 1/4:  54%|█████▎    | 670/1250 [16:15<09:31,  1.02it/s, loss=0.639]
Epoch 1/4:  54%|█████▎    | 671/1250 [16:15<08:52,  1.09it/s, loss=0.639]
Epoch 1/4:  54%|█████▎    | 671/1250 [16:17<08:52,  1.09it/s, loss=0.772]
Epoch 1/4:  54%|█████▍    | 672/1250 [16:17<09:38,  1.00s/it, loss=0.772]
Epoch 1/4:  54%|█████▍    | 672/1250 [16:18<09:38,  1.00s/it, loss=0.979]
Epoch 1/4:  54%|█████▍    | 673/1250 [16:18<10:03,  1.05s/it, loss=0.979]
Epoch 1/4:  54%|█████▍    | 673/1250 [16:19<10:03,  1.05s/it, loss=0.856]
Epoch 1/4:  54%|█████▍    | 674/1250 [16:19<10:13,  1.06s/it, loss=0.856]
Epoch 1/4:  54%|█████▍    | 674/1250 [16:20<10:13,  1.06s/it, loss=0.778]
Epoch 1/4:  54%|█████▍    | 675/1250 [16:20<09:32,  1.00it/s, loss=0.778]
Epoch 1/4:  54%|█████▍    | 675/1250 [16:21<09:32,  1.00it/s, loss=0.724]
Epoch 1/4:  54%|█████▍    | 676/1250 [16:21<09:20,  1.02it/s, loss=0.724]
Epoch 1/4:  54%|█████▍    | 676/1250 [16:22<09:20,  1.02it/s, loss=0.939]
Epoch 1/4:  54%|█████▍    | 677/1250 [16:22<10:42,  1.12s/it, loss=0.939]
Epoch 1/4:  54%|█████▍    | 677/1250 [16:23<10:42,  1.12s/it, loss=0.914]
Epoch 1/4:  54%|█████▍    | 678/1250 [16:23<10:42,  1.12s/it, loss=0.914]
Epoch 1/4:  54%|█████▍    | 678/1250 [16:24<10:42,  1.12s/it, loss=0.662]
Epoch 1/4:  54%|█████▍    | 679/1250 [16:24<09:55,  1.04s/it, loss=0.662]
Epoch 1/4:  54%|█████▍    | 679/1250 [16:25<09:55,  1.04s/it, loss=1.01] 
Epoch 1/4:  54%|█████▍    | 680/1250 [16:25<09:11,  1.03it/s, loss=1.01]
Epoch 1/4:  54%|█████▍    | 680/1250 [16:26<09:11,  1.03it/s, loss=0.6] 
Epoch 1/4:  54%|█████▍    | 681/1250 [16:26<08:38,  1.10it/s, loss=0.6]
Epoch 1/4:  54%|█████▍    | 681/1250 [16:27<08:38,  1.10it/s, loss=0.952]
Epoch 1/4:  55%|█████▍    | 682/1250 [16:27<09:29,  1.00s/it, loss=0.952]
Epoch 1/4:  55%|█████▍    | 682/1250 [16:28<09:29,  1.00s/it, loss=0.96] 
Epoch 1/4:  55%|█████▍    | 683/1250 [16:28<10:10,  1.08s/it, loss=0.96]
Epoch 1/4:  55%|█████▍    | 683/1250 [16:29<10:10,  1.08s/it, loss=0.917]
Epoch 1/4:  55%|█████▍    | 684/1250 [16:29<09:44,  1.03s/it, loss=0.917]
Epoch 1/4:  55%|█████▍    | 684/1250 [16:30<09:44,  1.03s/it, loss=0.698]
Epoch 1/4:  55%|█████▍    | 685/1250 [16:30<09:29,  1.01s/it, loss=0.698]
Epoch 1/4:  55%|█████▍    | 685/1250 [16:31<09:29,  1.01s/it, loss=0.898]
Epoch 1/4:  55%|█████▍    | 686/1250 [16:31<09:21,  1.00it/s, loss=0.898]
Epoch 1/4:  55%|█████▍    | 686/1250 [16:32<09:21,  1.00it/s, loss=1.03] 
Epoch 1/4:  55%|█████▍    | 687/1250 [16:32<10:42,  1.14s/it, loss=1.03]
Epoch 1/4:  55%|█████▍    | 687/1250 [16:34<10:42,  1.14s/it, loss=1.04]
Epoch 1/4:  55%|█████▌    | 688/1250 [16:34<10:48,  1.15s/it, loss=1.04]
Epoch 1/4:  55%|█████▌    | 688/1250 [16:35<10:48,  1.15s/it, loss=1.16]
Epoch 1/4:  55%|█████▌    | 689/1250 [16:35<10:17,  1.10s/it, loss=1.16]
Epoch 1/4:  55%|█████▌    | 689/1250 [16:36<10:17,  1.10s/it, loss=0.761]
Epoch 1/4:  55%|█████▌    | 690/1250 [16:36<09:43,  1.04s/it, loss=0.761]
Epoch 1/4:  55%|█████▌    | 690/1250 [16:36<09:43,  1.04s/it, loss=0.848]
Epoch 1/4:  55%|█████▌    | 691/1250 [16:36<09:19,  1.00s/it, loss=0.848]
Epoch 1/4:  55%|█████▌    | 691/1250 [16:38<09:19,  1.00s/it, loss=0.786]
Epoch 1/4:  55%|█████▌    | 692/1250 [16:38<10:06,  1.09s/it, loss=0.786]
Epoch 1/4:  55%|█████▌    | 692/1250 [16:39<10:06,  1.09s/it, loss=0.954]
Epoch 1/4:  55%|█████▌    | 693/1250 [16:39<10:07,  1.09s/it, loss=0.954]
Epoch 1/4:  55%|█████▌    | 693/1250 [16:40<10:07,  1.09s/it, loss=0.875]
Epoch 1/4:  56%|█████▌    | 694/1250 [16:40<09:48,  1.06s/it, loss=0.875]
Epoch 1/4:  56%|█████▌    | 694/1250 [16:41<09:48,  1.06s/it, loss=0.833]
Epoch 1/4:  56%|█████▌    | 695/1250 [16:41<09:23,  1.02s/it, loss=0.833]
Epoch 1/4:  56%|█████▌    | 695/1250 [16:42<09:23,  1.02s/it, loss=0.753]
Epoch 1/4:  56%|█████▌    | 696/1250 [16:42<09:16,  1.00s/it, loss=0.753]
Epoch 1/4:  56%|█████▌    | 696/1250 [16:43<09:16,  1.00s/it, loss=0.988]
Epoch 1/4:  56%|█████▌    | 697/1250 [16:43<10:38,  1.15s/it, loss=0.988]
Epoch 1/4:  56%|█████▌    | 697/1250 [16:44<10:38,  1.15s/it, loss=1.08] 
Epoch 1/4:  56%|█████▌    | 698/1250 [16:44<10:23,  1.13s/it, loss=1.08]
Epoch 1/4:  56%|█████▌    | 698/1250 [16:45<10:23,  1.13s/it, loss=1.03]
Epoch 1/4:  56%|█████▌    | 699/1250 [16:45<09:51,  1.07s/it, loss=1.03]
Epoch 1/4:  56%|█████▌    | 699/1250 [16:46<09:51,  1.07s/it, loss=0.81]
Epoch 1/4:  56%|█████▌    | 700/1250 [16:46<09:25,  1.03s/it, loss=0.81]
Epoch 1/4:  56%|█████▌    | 700/1250 [16:47<09:25,  1.03s/it, loss=0.908]
Epoch 1/4:  56%|█████▌    | 701/1250 [16:47<09:08,  1.00it/s, loss=0.908]
Epoch 1/4:  56%|█████▌    | 701/1250 [16:48<09:08,  1.00it/s, loss=0.71] 
Epoch 1/4:  56%|█████▌    | 702/1250 [16:48<10:09,  1.11s/it, loss=0.71]
Epoch 1/4:  56%|█████▌    | 702/1250 [16:49<10:09,  1.11s/it, loss=0.961]
Epoch 1/4:  56%|█████▌    | 703/1250 [16:49<09:57,  1.09s/it, loss=0.961]
Epoch 1/4:  56%|█████▌    | 703/1250 [16:50<09:57,  1.09s/it, loss=0.976]
Epoch 1/4:  56%|█████▋    | 704/1250 [16:50<09:30,  1.05s/it, loss=0.976]
Epoch 1/4:  56%|█████▋    | 704/1250 [16:51<09:30,  1.05s/it, loss=1.43] 
Epoch 1/4:  56%|█████▋    | 705/1250 [16:51<09:10,  1.01s/it, loss=1.43]
Epoch 1/4:  56%|█████▋    | 705/1250 [16:52<09:10,  1.01s/it, loss=1.07]
Epoch 1/4:  56%|█████▋    | 706/1250 [16:52<09:05,  1.00s/it, loss=1.07]
Epoch 1/4:  56%|█████▋    | 706/1250 [16:54<09:05,  1.00s/it, loss=0.958]
Epoch 1/4:  57%|█████▋    | 707/1250 [16:54<09:50,  1.09s/it, loss=0.958]
Epoch 1/4:  57%|█████▋    | 707/1250 [16:54<09:50,  1.09s/it, loss=0.7]  
Epoch 1/4:  57%|█████▋    | 708/1250 [16:54<09:11,  1.02s/it, loss=0.7]
Epoch 1/4:  57%|█████▋    | 708/1250 [16:55<09:11,  1.02s/it, loss=0.648]
Epoch 1/4:  57%|█████▋    | 709/1250 [16:55<08:39,  1.04it/s, loss=0.648]
Epoch 1/4:  57%|█████▋    | 709/1250 [16:56<08:39,  1.04it/s, loss=0.737]
Epoch 1/4:  57%|█████▋    | 710/1250 [16:56<08:09,  1.10it/s, loss=0.737]
Epoch 1/4:  57%|█████▋    | 710/1250 [16:57<08:09,  1.10it/s, loss=0.669]
Epoch 1/4:  57%|█████▋    | 711/1250 [16:57<08:11,  1.10it/s, loss=0.669]
Epoch 1/4:  57%|█████▋    | 711/1250 [16:58<08:11,  1.10it/s, loss=0.656]
Epoch 1/4:  57%|█████▋    | 712/1250 [16:58<09:11,  1.02s/it, loss=0.656]
Epoch 1/4:  57%|█████▋    | 712/1250 [16:59<09:11,  1.02s/it, loss=0.889]
Epoch 1/4:  57%|█████▋    | 713/1250 [16:59<09:01,  1.01s/it, loss=0.889]
Epoch 1/4:  57%|█████▋    | 713/1250 [17:00<09:01,  1.01s/it, loss=0.768]
Epoch 1/4:  57%|█████▋    | 714/1250 [17:00<08:37,  1.04it/s, loss=0.768]
Epoch 1/4:  57%|█████▋    | 714/1250 [17:01<08:37,  1.04it/s, loss=0.796]
Epoch 1/4:  57%|█████▋    | 715/1250 [17:01<08:19,  1.07it/s, loss=0.796]
Epoch 1/4:  57%|█████▋    | 715/1250 [17:02<08:19,  1.07it/s, loss=0.616]
Epoch 1/4:  57%|█████▋    | 716/1250 [17:02<08:32,  1.04it/s, loss=0.616]
Epoch 1/4:  57%|█████▋    | 716/1250 [17:03<08:32,  1.04it/s, loss=0.765]
Epoch 1/4:  57%|█████▋    | 717/1250 [17:03<09:15,  1.04s/it, loss=0.765]
Epoch 1/4:  57%|█████▋    | 717/1250 [17:04<09:15,  1.04s/it, loss=1.02] 
Epoch 1/4:  57%|█████▋    | 718/1250 [17:04<09:14,  1.04s/it, loss=1.02]
Epoch 1/4:  57%|█████▋    | 718/1250 [17:05<09:14,  1.04s/it, loss=0.896]
Epoch 1/4:  58%|█████▊    | 719/1250 [17:05<08:49,  1.00it/s, loss=0.896]
Epoch 1/4:  58%|█████▊    | 719/1250 [17:06<08:49,  1.00it/s, loss=0.601]
Epoch 1/4:  58%|█████▊    | 720/1250 [17:06<08:16,  1.07it/s, loss=0.601]
Epoch 1/4:  58%|█████▊    | 720/1250 [17:07<08:16,  1.07it/s, loss=0.674]
Epoch 1/4:  58%|█████▊    | 721/1250 [17:07<08:01,  1.10it/s, loss=0.674]
Epoch 1/4:  58%|█████▊    | 721/1250 [17:08<08:01,  1.10it/s, loss=0.912]
Epoch 1/4:  58%|█████▊    | 722/1250 [17:08<08:16,  1.06it/s, loss=0.912]
Epoch 1/4:  58%|█████▊    | 722/1250 [17:09<08:16,  1.06it/s, loss=1.01] 
Epoch 1/4:  58%|█████▊    | 723/1250 [17:09<09:44,  1.11s/it, loss=1.01]
Epoch 1/4:  58%|█████▊    | 723/1250 [17:10<09:44,  1.11s/it, loss=0.727]
Epoch 1/4:  58%|█████▊    | 724/1250 [17:10<09:08,  1.04s/it, loss=0.727]
Epoch 1/4:  58%|█████▊    | 724/1250 [17:11<09:08,  1.04s/it, loss=0.94] 
Epoch 1/4:  58%|█████▊    | 725/1250 [17:11<08:29,  1.03it/s, loss=0.94]
Epoch 1/4:  58%|█████▊    | 725/1250 [17:12<08:29,  1.03it/s, loss=0.887]
Epoch 1/4:  58%|█████▊    | 726/1250 [17:12<08:13,  1.06it/s, loss=0.887]
Epoch 1/4:  58%|█████▊    | 726/1250 [17:13<08:13,  1.06it/s, loss=1.01] 
Epoch 1/4:  58%|█████▊    | 727/1250 [17:13<08:35,  1.01it/s, loss=1.01]
Epoch 1/4:  58%|█████▊    | 727/1250 [17:15<08:35,  1.01it/s, loss=1.45]
Epoch 1/4:  58%|█████▊    | 728/1250 [17:15<10:16,  1.18s/it, loss=1.45]
Epoch 1/4:  58%|█████▊    | 728/1250 [17:16<10:16,  1.18s/it, loss=0.775]
Epoch 1/4:  58%|█████▊    | 729/1250 [17:16<09:49,  1.13s/it, loss=0.775]
Epoch 1/4:  58%|█████▊    | 729/1250 [17:17<09:49,  1.13s/it, loss=0.798]
Epoch 1/4:  58%|█████▊    | 730/1250 [17:17<09:13,  1.06s/it, loss=0.798]
Epoch 1/4:  58%|█████▊    | 730/1250 [17:17<09:13,  1.06s/it, loss=0.681]
Epoch 1/4:  58%|█████▊    | 731/1250 [17:17<08:49,  1.02s/it, loss=0.681]
Epoch 1/4:  58%|█████▊    | 731/1250 [17:19<08:49,  1.02s/it, loss=0.62] 
Epoch 1/4:  59%|█████▊    | 732/1250 [17:19<08:57,  1.04s/it, loss=0.62]
Epoch 1/4:  59%|█████▊    | 732/1250 [17:20<08:57,  1.04s/it, loss=1.03]
Epoch 1/4:  59%|█████▊    | 733/1250 [17:20<09:29,  1.10s/it, loss=1.03]
Epoch 1/4:  59%|█████▊    | 733/1250 [17:21<09:29,  1.10s/it, loss=0.824]
Epoch 1/4:  59%|█████▊    | 734/1250 [17:21<09:09,  1.07s/it, loss=0.824]
Epoch 1/4:  59%|█████▊    | 734/1250 [17:22<09:09,  1.07s/it, loss=0.674]
Epoch 1/4:  59%|█████▉    | 735/1250 [17:22<08:51,  1.03s/it, loss=0.674]
Epoch 1/4:  59%|█████▉    | 735/1250 [17:23<08:51,  1.03s/it, loss=0.504]
Epoch 1/4:  59%|█████▉    | 736/1250 [17:23<08:47,  1.03s/it, loss=0.504]
Epoch 1/4:  59%|█████▉    | 736/1250 [17:24<08:47,  1.03s/it, loss=0.694]
Epoch 1/4:  59%|█████▉    | 737/1250 [17:24<09:20,  1.09s/it, loss=0.694]
Epoch 1/4:  59%|█████▉    | 737/1250 [17:26<09:20,  1.09s/it, loss=0.663]
Epoch 1/4:  59%|█████▉    | 738/1250 [17:26<10:24,  1.22s/it, loss=0.663]
Epoch 1/4:  59%|█████▉    | 738/1250 [17:26<10:24,  1.22s/it, loss=0.795]
Epoch 1/4:  59%|█████▉    | 739/1250 [17:26<09:46,  1.15s/it, loss=0.795]
Epoch 1/4:  59%|█████▉    | 739/1250 [17:27<09:46,  1.15s/it, loss=0.872]
Epoch 1/4:  59%|█████▉    | 740/1250 [17:27<09:20,  1.10s/it, loss=0.872]
Epoch 1/4:  59%|█████▉    | 740/1250 [17:29<09:20,  1.10s/it, loss=0.923]
Epoch 1/4:  59%|█████▉    | 741/1250 [17:29<09:12,  1.08s/it, loss=0.923]
Epoch 1/4:  59%|█████▉    | 741/1250 [17:30<09:12,  1.08s/it, loss=0.692]
Epoch 1/4:  59%|█████▉    | 742/1250 [17:30<09:47,  1.16s/it, loss=0.692]
Epoch 1/4:  59%|█████▉    | 742/1250 [17:31<09:47,  1.16s/it, loss=0.919]
Epoch 1/4:  59%|█████▉    | 743/1250 [17:31<09:18,  1.10s/it, loss=0.919]
Epoch 1/4:  59%|█████▉    | 743/1250 [17:32<09:18,  1.10s/it, loss=0.817]
Epoch 1/4:  60%|█████▉    | 744/1250 [17:32<08:52,  1.05s/it, loss=0.817]
Epoch 1/4:  60%|█████▉    | 744/1250 [17:33<08:52,  1.05s/it, loss=1.06] 
Epoch 1/4:  60%|█████▉    | 745/1250 [17:33<08:41,  1.03s/it, loss=1.06]
Epoch 1/4:  60%|█████▉    | 745/1250 [17:34<08:41,  1.03s/it, loss=0.83]
Epoch 1/4:  60%|█████▉    | 746/1250 [17:34<08:52,  1.06s/it, loss=0.83]
Epoch 1/4:  60%|█████▉    | 746/1250 [17:35<08:52,  1.06s/it, loss=0.755]
Epoch 1/4:  60%|█████▉    | 747/1250 [17:35<09:20,  1.11s/it, loss=0.755]
Epoch 1/4:  60%|█████▉    | 747/1250 [17:36<09:20,  1.11s/it, loss=1.25] 
Epoch 1/4:  60%|█████▉    | 748/1250 [17:36<08:45,  1.05s/it, loss=1.25]
Epoch 1/4:  60%|█████▉    | 748/1250 [17:37<08:45,  1.05s/it, loss=0.703]
Epoch 1/4:  60%|█████▉    | 749/1250 [17:37<08:04,  1.03it/s, loss=0.703]
Epoch 1/4:  60%|█████▉    | 749/1250 [17:38<08:04,  1.03it/s, loss=0.863]
Epoch 1/4:  60%|██████    | 750/1250 [17:38<07:56,  1.05it/s, loss=0.863]
Epoch 1/4:  60%|██████    | 750/1250 [17:39<07:56,  1.05it/s, loss=0.989]
Epoch 1/4:  60%|██████    | 751/1250 [17:39<08:04,  1.03it/s, loss=0.989]
Epoch 1/4:  60%|██████    | 751/1250 [17:40<08:04,  1.03it/s, loss=0.985]
Epoch 1/4:  60%|██████    | 752/1250 [17:40<08:38,  1.04s/it, loss=0.985]
Epoch 1/4:  60%|██████    | 752/1250 [17:41<08:38,  1.04s/it, loss=0.858]
Epoch 1/4:  60%|██████    | 753/1250 [17:41<08:11,  1.01it/s, loss=0.858]
Epoch 1/4:  60%|██████    | 753/1250 [17:42<08:11,  1.01it/s, loss=0.907]
Epoch 1/4:  60%|██████    | 754/1250 [17:42<07:42,  1.07it/s, loss=0.907]
Epoch 1/4:  60%|██████    | 754/1250 [17:42<07:42,  1.07it/s, loss=1.11] 
Epoch 1/4:  60%|██████    | 755/1250 [17:42<07:38,  1.08it/s, loss=1.11]
Epoch 1/4:  60%|██████    | 755/1250 [17:43<07:38,  1.08it/s, loss=1.11]
Epoch 1/4:  60%|██████    | 756/1250 [17:43<07:49,  1.05it/s, loss=1.11]
Epoch 1/4:  60%|██████    | 756/1250 [17:45<07:49,  1.05it/s, loss=0.79]
Epoch 1/4:  61%|██████    | 757/1250 [17:45<08:05,  1.02it/s, loss=0.79]
Epoch 1/4:  61%|██████    | 757/1250 [17:46<08:05,  1.02it/s, loss=1.18]
Epoch 1/4:  61%|██████    | 758/1250 [17:46<08:23,  1.02s/it, loss=1.18]
Epoch 1/4:  61%|██████    | 758/1250 [17:47<08:23,  1.02s/it, loss=0.892]
Epoch 1/4:  61%|██████    | 759/1250 [17:47<07:56,  1.03it/s, loss=0.892]
Epoch 1/4:  61%|██████    | 759/1250 [17:47<07:56,  1.03it/s, loss=0.66] 
Epoch 1/4:  61%|██████    | 760/1250 [17:47<07:47,  1.05it/s, loss=0.66]
Epoch 1/4:  61%|██████    | 760/1250 [17:48<07:47,  1.05it/s, loss=0.951]
Epoch 1/4:  61%|██████    | 761/1250 [17:48<07:57,  1.02it/s, loss=0.951]
Epoch 1/4:  61%|██████    | 761/1250 [17:49<07:57,  1.02it/s, loss=1.34] 
Epoch 1/4:  61%|██████    | 762/1250 [17:49<07:59,  1.02it/s, loss=1.34]
Epoch 1/4:  61%|██████    | 762/1250 [17:51<07:59,  1.02it/s, loss=1.75]
Epoch 1/4:  61%|██████    | 763/1250 [17:51<08:16,  1.02s/it, loss=1.75]
Epoch 1/4:  61%|██████    | 763/1250 [17:51<08:16,  1.02s/it, loss=0.849]
Epoch 1/4:  61%|██████    | 764/1250 [17:51<07:48,  1.04it/s, loss=0.849]
Epoch 1/4:  61%|██████    | 764/1250 [17:52<07:48,  1.04it/s, loss=0.751]
Epoch 1/4:  61%|██████    | 765/1250 [17:52<07:32,  1.07it/s, loss=0.751]
Epoch 1/4:  61%|██████    | 765/1250 [17:53<07:32,  1.07it/s, loss=1.03] 
Epoch 1/4:  61%|██████▏   | 766/1250 [17:53<07:43,  1.04it/s, loss=1.03]
Epoch 1/4:  61%|██████▏   | 766/1250 [17:54<07:43,  1.04it/s, loss=1.09]
Epoch 1/4:  61%|██████▏   | 767/1250 [17:54<07:44,  1.04it/s, loss=1.09]
Epoch 1/4:  61%|██████▏   | 767/1250 [17:55<07:44,  1.04it/s, loss=0.887]
Epoch 1/4:  61%|██████▏   | 768/1250 [17:55<08:13,  1.02s/it, loss=0.887]
Epoch 1/4:  61%|██████▏   | 768/1250 [17:56<08:13,  1.02s/it, loss=0.81] 
Epoch 1/4:  62%|██████▏   | 769/1250 [17:56<07:52,  1.02it/s, loss=0.81]
Epoch 1/4:  62%|██████▏   | 769/1250 [17:57<07:52,  1.02it/s, loss=1.11]
Epoch 1/4:  62%|██████▏   | 770/1250 [17:57<07:34,  1.06it/s, loss=1.11]
Epoch 1/4:  62%|██████▏   | 770/1250 [17:58<07:34,  1.06it/s, loss=0.821]
Epoch 1/4:  62%|██████▏   | 771/1250 [17:58<07:43,  1.03it/s, loss=0.821]
Epoch 1/4:  62%|██████▏   | 771/1250 [17:59<07:43,  1.03it/s, loss=1.06] 
Epoch 1/4:  62%|██████▏   | 772/1250 [17:59<07:45,  1.03it/s, loss=1.06]
Epoch 1/4:  62%|██████▏   | 772/1250 [18:00<07:45,  1.03it/s, loss=0.827]
Epoch 1/4:  62%|██████▏   | 773/1250 [18:00<07:31,  1.06it/s, loss=0.827]
Epoch 1/4:  62%|██████▏   | 773/1250 [18:01<07:31,  1.06it/s, loss=0.661]
Epoch 1/4:  62%|██████▏   | 774/1250 [18:01<07:51,  1.01it/s, loss=0.661]
Epoch 1/4:  62%|██████▏   | 774/1250 [18:02<07:51,  1.01it/s, loss=1.02] 
Epoch 1/4:  62%|██████▏   | 775/1250 [18:02<07:53,  1.00it/s, loss=1.02]
Epoch 1/4:  62%|██████▏   | 775/1250 [18:03<07:53,  1.00it/s, loss=0.817]
Epoch 1/4:  62%|██████▏   | 776/1250 [18:03<08:02,  1.02s/it, loss=0.817]
Epoch 1/4:  62%|██████▏   | 776/1250 [18:04<08:02,  1.02s/it, loss=0.555]
Epoch 1/4:  62%|██████▏   | 777/1250 [18:04<07:58,  1.01s/it, loss=0.555]
Epoch 1/4:  62%|██████▏   | 777/1250 [18:05<07:58,  1.01s/it, loss=0.672]
Epoch 1/4:  62%|██████▏   | 778/1250 [18:05<07:29,  1.05it/s, loss=0.672]
Epoch 1/4:  62%|██████▏   | 778/1250 [18:06<07:29,  1.05it/s, loss=0.75] 
Epoch 1/4:  62%|██████▏   | 779/1250 [18:06<07:48,  1.00it/s, loss=0.75]
Epoch 1/4:  62%|██████▏   | 779/1250 [18:07<07:48,  1.00it/s, loss=0.939]
Epoch 1/4:  62%|██████▏   | 780/1250 [18:07<07:32,  1.04it/s, loss=0.939]
Epoch 1/4:  62%|██████▏   | 780/1250 [18:08<07:32,  1.04it/s, loss=0.862]
Epoch 1/4:  62%|██████▏   | 781/1250 [18:08<07:35,  1.03it/s, loss=0.862]
Epoch 1/4:  62%|██████▏   | 781/1250 [18:09<07:35,  1.03it/s, loss=0.539]
Epoch 1/4:  63%|██████▎   | 782/1250 [18:09<07:54,  1.01s/it, loss=0.539]
Epoch 1/4:  63%|██████▎   | 782/1250 [18:10<07:54,  1.01s/it, loss=0.889]
Epoch 1/4:  63%|██████▎   | 783/1250 [18:10<07:41,  1.01it/s, loss=0.889]
Epoch 1/4:  63%|██████▎   | 783/1250 [18:11<07:41,  1.01it/s, loss=0.584]
Epoch 1/4:  63%|██████▎   | 784/1250 [18:11<08:21,  1.08s/it, loss=0.584]
Epoch 1/4:  63%|██████▎   | 784/1250 [18:12<08:21,  1.08s/it, loss=1.04] 
Epoch 1/4:  63%|██████▎   | 785/1250 [18:12<08:28,  1.09s/it, loss=1.04]
Epoch 1/4:  63%|██████▎   | 785/1250 [18:14<08:28,  1.09s/it, loss=0.661]
Epoch 1/4:  63%|██████▎   | 786/1250 [18:14<08:34,  1.11s/it, loss=0.661]
Epoch 1/4:  63%|██████▎   | 786/1250 [18:15<08:34,  1.11s/it, loss=0.905]
Epoch 1/4:  63%|██████▎   | 787/1250 [18:15<08:26,  1.09s/it, loss=0.905]
Epoch 1/4:  63%|██████▎   | 787/1250 [18:16<08:26,  1.09s/it, loss=1.18] 
Epoch 1/4:  63%|██████▎   | 788/1250 [18:16<08:08,  1.06s/it, loss=1.18]
Epoch 1/4:  63%|██████▎   | 788/1250 [18:17<08:08,  1.06s/it, loss=1.3] 
Epoch 1/4:  63%|██████▎   | 789/1250 [18:17<08:40,  1.13s/it, loss=1.3]
Epoch 1/4:  63%|██████▎   | 789/1250 [18:18<08:40,  1.13s/it, loss=0.992]
Epoch 1/4:  63%|██████▎   | 790/1250 [18:18<08:24,  1.10s/it, loss=0.992]
Epoch 1/4:  63%|██████▎   | 790/1250 [18:19<08:24,  1.10s/it, loss=0.634]
Epoch 1/4:  63%|██████▎   | 791/1250 [18:19<08:24,  1.10s/it, loss=0.634]
Epoch 1/4:  63%|██████▎   | 791/1250 [18:20<08:24,  1.10s/it, loss=0.829]
Epoch 1/4:  63%|██████▎   | 792/1250 [18:20<08:16,  1.08s/it, loss=0.829]
Epoch 1/4:  63%|██████▎   | 792/1250 [18:21<08:16,  1.08s/it, loss=0.663]
Epoch 1/4:  63%|██████▎   | 793/1250 [18:21<08:13,  1.08s/it, loss=0.663]
Epoch 1/4:  63%|██████▎   | 793/1250 [18:22<08:13,  1.08s/it, loss=1.1]  
Epoch 1/4:  64%|██████▎   | 794/1250 [18:22<08:18,  1.09s/it, loss=1.1]
Epoch 1/4:  64%|██████▎   | 794/1250 [18:23<08:18,  1.09s/it, loss=0.772]
Epoch 1/4:  64%|██████▎   | 795/1250 [18:23<08:09,  1.08s/it, loss=0.772]
Epoch 1/4:  64%|██████▎   | 795/1250 [18:25<08:09,  1.08s/it, loss=1.04] 
Epoch 1/4:  64%|██████▎   | 796/1250 [18:25<08:49,  1.17s/it, loss=1.04]
Epoch 1/4:  64%|██████▎   | 796/1250 [18:26<08:49,  1.17s/it, loss=0.753]
Epoch 1/4:  64%|██████▍   | 797/1250 [18:26<08:38,  1.14s/it, loss=0.753]
Epoch 1/4:  64%|██████▍   | 797/1250 [18:27<08:38,  1.14s/it, loss=0.869]
Epoch 1/4:  64%|██████▍   | 798/1250 [18:27<08:48,  1.17s/it, loss=0.869]
Epoch 1/4:  64%|██████▍   | 798/1250 [18:28<08:48,  1.17s/it, loss=0.857]
Epoch 1/4:  64%|██████▍   | 799/1250 [18:28<09:04,  1.21s/it, loss=0.857]
Epoch 1/4:  64%|██████▍   | 799/1250 [18:30<09:04,  1.21s/it, loss=1.7]  
Epoch 1/4:  64%|██████▍   | 800/1250 [18:30<09:22,  1.25s/it, loss=1.7]
Epoch 1/4:  64%|██████▍   | 800/1250 [18:31<09:22,  1.25s/it, loss=0.908]
Epoch 1/4:  64%|██████▍   | 801/1250 [18:31<09:15,  1.24s/it, loss=0.908]
Epoch 1/4:  64%|██████▍   | 801/1250 [18:32<09:15,  1.24s/it, loss=1.01] 
Epoch 1/4:  64%|██████▍   | 802/1250 [18:32<09:09,  1.23s/it, loss=1.01]
Epoch 1/4:  64%|██████▍   | 802/1250 [18:33<09:09,  1.23s/it, loss=0.703]
Epoch 1/4:  64%|██████▍   | 803/1250 [18:33<09:03,  1.22s/it, loss=0.703]
Epoch 1/4:  64%|██████▍   | 803/1250 [18:34<09:03,  1.22s/it, loss=0.67] 
Epoch 1/4:  64%|██████▍   | 804/1250 [18:34<08:38,  1.16s/it, loss=0.67]
Epoch 1/4:  64%|██████▍   | 804/1250 [18:35<08:38,  1.16s/it, loss=0.865]
Epoch 1/4:  64%|██████▍   | 805/1250 [18:35<08:18,  1.12s/it, loss=0.865]
Epoch 1/4:  64%|██████▍   | 805/1250 [18:36<08:18,  1.12s/it, loss=0.669]
Epoch 1/4:  64%|██████▍   | 806/1250 [18:36<08:13,  1.11s/it, loss=0.669]
Epoch 1/4:  64%|██████▍   | 806/1250 [18:38<08:13,  1.11s/it, loss=0.72] 
Epoch 1/4:  65%|██████▍   | 807/1250 [18:38<08:59,  1.22s/it, loss=0.72]
Epoch 1/4:  65%|██████▍   | 807/1250 [18:39<08:59,  1.22s/it, loss=0.602]
Epoch 1/4:  65%|██████▍   | 808/1250 [18:39<08:37,  1.17s/it, loss=0.602]
Epoch 1/4:  65%|██████▍   | 808/1250 [18:40<08:37,  1.17s/it, loss=0.97] 
Epoch 1/4:  65%|██████▍   | 809/1250 [18:40<08:09,  1.11s/it, loss=0.97]
Epoch 1/4:  65%|██████▍   | 809/1250 [18:41<08:09,  1.11s/it, loss=1.26]
Epoch 1/4:  65%|██████▍   | 810/1250 [18:41<08:03,  1.10s/it, loss=1.26]
Epoch 1/4:  65%|██████▍   | 810/1250 [18:42<08:03,  1.10s/it, loss=0.717]
Epoch 1/4:  65%|██████▍   | 811/1250 [18:42<07:51,  1.07s/it, loss=0.717]
Epoch 1/4:  65%|██████▍   | 811/1250 [18:43<07:51,  1.07s/it, loss=0.894]
Epoch 1/4:  65%|██████▍   | 812/1250 [18:43<08:31,  1.17s/it, loss=0.894]
Epoch 1/4:  65%|██████▍   | 812/1250 [18:45<08:31,  1.17s/it, loss=1.06] 
Epoch 1/4:  65%|██████▌   | 813/1250 [18:45<08:21,  1.15s/it, loss=1.06]
Epoch 1/4:  65%|██████▌   | 813/1250 [18:46<08:21,  1.15s/it, loss=0.762]
Epoch 1/4:  65%|██████▌   | 814/1250 [18:46<07:59,  1.10s/it, loss=0.762]
Epoch 1/4:  65%|██████▌   | 814/1250 [18:46<07:59,  1.10s/it, loss=0.499]
Epoch 1/4:  65%|██████▌   | 815/1250 [18:46<07:38,  1.06s/it, loss=0.499]
Epoch 1/4:  65%|██████▌   | 815/1250 [18:47<07:38,  1.06s/it, loss=0.815]
Epoch 1/4:  65%|██████▌   | 816/1250 [18:47<07:28,  1.03s/it, loss=0.815]
Epoch 1/4:  65%|██████▌   | 816/1250 [18:49<07:28,  1.03s/it, loss=0.953]
Epoch 1/4:  65%|██████▌   | 817/1250 [18:49<07:54,  1.10s/it, loss=0.953]
Epoch 1/4:  65%|██████▌   | 817/1250 [18:50<07:54,  1.10s/it, loss=1.22] 
Epoch 1/4:  65%|██████▌   | 818/1250 [18:50<07:49,  1.09s/it, loss=1.22]
Epoch 1/4:  65%|██████▌   | 818/1250 [18:51<07:49,  1.09s/it, loss=0.576]
Epoch 1/4:  66%|██████▌   | 819/1250 [18:51<07:28,  1.04s/it, loss=0.576]
Epoch 1/4:  66%|██████▌   | 819/1250 [18:52<07:28,  1.04s/it, loss=0.835]
Epoch 1/4:  66%|██████▌   | 820/1250 [18:52<07:02,  1.02it/s, loss=0.835]
Epoch 1/4:  66%|██████▌   | 820/1250 [18:52<07:02,  1.02it/s, loss=0.724]
Epoch 1/4:  66%|██████▌   | 821/1250 [18:52<06:43,  1.06it/s, loss=0.724]
Epoch 1/4:  66%|██████▌   | 821/1250 [18:54<06:43,  1.06it/s, loss=0.877]
Epoch 1/4:  66%|██████▌   | 822/1250 [18:54<07:53,  1.11s/it, loss=0.877]
Epoch 1/4:  66%|██████▌   | 822/1250 [18:56<07:53,  1.11s/it, loss=1.22] 
Epoch 1/4:  66%|██████▌   | 823/1250 [18:56<09:03,  1.27s/it, loss=1.22]
Epoch 1/4:  66%|██████▌   | 823/1250 [18:57<09:03,  1.27s/it, loss=0.596]
Epoch 1/4:  66%|██████▌   | 824/1250 [18:57<08:43,  1.23s/it, loss=0.596]
Epoch 1/4:  66%|██████▌   | 824/1250 [18:58<08:43,  1.23s/it, loss=0.917]
Epoch 1/4:  66%|██████▌   | 825/1250 [18:58<08:42,  1.23s/it, loss=0.917]
Epoch 1/4:  66%|██████▌   | 825/1250 [18:59<08:42,  1.23s/it, loss=1.05] 
Epoch 1/4:  66%|██████▌   | 826/1250 [18:59<09:18,  1.32s/it, loss=1.05]
Epoch 1/4:  66%|██████▌   | 826/1250 [19:01<09:18,  1.32s/it, loss=1.02]
Epoch 1/4:  66%|██████▌   | 827/1250 [19:01<09:10,  1.30s/it, loss=1.02]
Epoch 1/4:  66%|██████▌   | 827/1250 [19:02<09:10,  1.30s/it, loss=0.844]
Epoch 1/4:  66%|██████▌   | 828/1250 [19:02<08:41,  1.24s/it, loss=0.844]
Epoch 1/4:  66%|██████▌   | 828/1250 [19:03<08:41,  1.24s/it, loss=0.907]
Epoch 1/4:  66%|██████▋   | 829/1250 [19:03<08:03,  1.15s/it, loss=0.907]
Epoch 1/4:  66%|██████▋   | 829/1250 [19:04<08:03,  1.15s/it, loss=0.582]
Epoch 1/4:  66%|██████▋   | 830/1250 [19:04<08:21,  1.19s/it, loss=0.582]
Epoch 1/4:  66%|██████▋   | 830/1250 [19:05<08:21,  1.19s/it, loss=0.877]
Epoch 1/4:  66%|██████▋   | 831/1250 [19:05<08:13,  1.18s/it, loss=0.877]
Epoch 1/4:  66%|██████▋   | 831/1250 [19:06<08:13,  1.18s/it, loss=0.581]
Epoch 1/4:  67%|██████▋   | 832/1250 [19:06<07:56,  1.14s/it, loss=0.581]
Epoch 1/4:  67%|██████▋   | 832/1250 [19:07<07:56,  1.14s/it, loss=1.1]  
Epoch 1/4:  67%|██████▋   | 833/1250 [19:07<07:39,  1.10s/it, loss=1.1]
Epoch 1/4:  67%|██████▋   | 833/1250 [19:08<07:39,  1.10s/it, loss=1.2]
Epoch 1/4:  67%|██████▋   | 834/1250 [19:08<07:36,  1.10s/it, loss=1.2]
Epoch 1/4:  67%|██████▋   | 834/1250 [19:10<07:36,  1.10s/it, loss=0.688]
Epoch 1/4:  67%|██████▋   | 835/1250 [19:10<08:01,  1.16s/it, loss=0.688]
Epoch 1/4:  67%|██████▋   | 835/1250 [19:11<08:01,  1.16s/it, loss=0.714]
Epoch 1/4:  67%|██████▋   | 836/1250 [19:11<07:49,  1.13s/it, loss=0.714]
Epoch 1/4:  67%|██████▋   | 836/1250 [19:12<07:49,  1.13s/it, loss=1.25] 
Epoch 1/4:  67%|██████▋   | 837/1250 [19:12<07:43,  1.12s/it, loss=1.25]
Epoch 1/4:  67%|██████▋   | 837/1250 [19:13<07:43,  1.12s/it, loss=0.77]
Epoch 1/4:  67%|██████▋   | 838/1250 [19:13<07:59,  1.16s/it, loss=0.77]
Epoch 1/4:  67%|██████▋   | 838/1250 [19:14<07:59,  1.16s/it, loss=0.829]
Epoch 1/4:  67%|██████▋   | 839/1250 [19:14<08:22,  1.22s/it, loss=0.829]
Epoch 1/4:  67%|██████▋   | 839/1250 [19:16<08:22,  1.22s/it, loss=0.554]
Epoch 1/4:  67%|██████▋   | 840/1250 [19:16<08:18,  1.22s/it, loss=0.554]
Epoch 1/4:  67%|██████▋   | 840/1250 [19:17<08:18,  1.22s/it, loss=0.86] 
Epoch 1/4:  67%|██████▋   | 841/1250 [19:17<08:07,  1.19s/it, loss=0.86]
Epoch 1/4:  67%|██████▋   | 841/1250 [19:18<08:07,  1.19s/it, loss=0.54]
Epoch 1/4:  67%|██████▋   | 842/1250 [19:18<07:50,  1.15s/it, loss=0.54]
Epoch 1/4:  67%|██████▋   | 842/1250 [19:19<07:50,  1.15s/it, loss=0.836]
Epoch 1/4:  67%|██████▋   | 843/1250 [19:19<07:40,  1.13s/it, loss=0.836]
Epoch 1/4:  67%|██████▋   | 843/1250 [19:20<07:40,  1.13s/it, loss=0.978]
Epoch 1/4:  68%|██████▊   | 844/1250 [19:20<07:52,  1.16s/it, loss=0.978]
Epoch 1/4:  68%|██████▊   | 844/1250 [19:22<07:52,  1.16s/it, loss=1.14] 
Epoch 1/4:  68%|██████▊   | 845/1250 [19:22<09:36,  1.42s/it, loss=1.14]
Epoch 1/4:  68%|██████▊   | 845/1250 [19:24<09:36,  1.42s/it, loss=0.887]
Epoch 1/4:  68%|██████▊   | 846/1250 [19:24<09:31,  1.42s/it, loss=0.887]
Epoch 1/4:  68%|██████▊   | 846/1250 [19:24<09:31,  1.42s/it, loss=1.35] 
Epoch 1/4:  68%|██████▊   | 847/1250 [19:24<08:34,  1.28s/it, loss=1.35]
Epoch 1/4:  68%|██████▊   | 847/1250 [19:26<08:34,  1.28s/it, loss=0.882]
Epoch 1/4:  68%|██████▊   | 848/1250 [19:26<08:12,  1.23s/it, loss=0.882]
Epoch 1/4:  68%|██████▊   | 848/1250 [19:27<08:12,  1.23s/it, loss=0.982]
Epoch 1/4:  68%|██████▊   | 849/1250 [19:27<08:14,  1.23s/it, loss=0.982]
Epoch 1/4:  68%|██████▊   | 849/1250 [19:28<08:14,  1.23s/it, loss=0.869]
Epoch 1/4:  68%|██████▊   | 850/1250 [19:28<07:44,  1.16s/it, loss=0.869]
Epoch 1/4:  68%|██████▊   | 850/1250 [19:29<07:44,  1.16s/it, loss=0.959]
Epoch 1/4:  68%|██████▊   | 851/1250 [19:29<07:14,  1.09s/it, loss=0.959]
Epoch 1/4:  68%|██████▊   | 851/1250 [19:30<07:14,  1.09s/it, loss=0.969]
Epoch 1/4:  68%|██████▊   | 852/1250 [19:30<07:19,  1.10s/it, loss=0.969]
Epoch 1/4:  68%|██████▊   | 852/1250 [19:31<07:19,  1.10s/it, loss=0.757]
Epoch 1/4:  68%|██████▊   | 853/1250 [19:31<07:07,  1.08s/it, loss=0.757]
Epoch 1/4:  68%|██████▊   | 853/1250 [19:32<07:07,  1.08s/it, loss=1.25] 
Epoch 1/4:  68%|██████▊   | 854/1250 [19:32<06:55,  1.05s/it, loss=1.25]
Epoch 1/4:  68%|██████▊   | 854/1250 [19:33<06:55,  1.05s/it, loss=1.46]
Epoch 1/4:  68%|██████▊   | 855/1250 [19:33<06:45,  1.03s/it, loss=1.46]
Epoch 1/4:  68%|██████▊   | 855/1250 [19:34<06:45,  1.03s/it, loss=0.73]
Epoch 1/4:  68%|██████▊   | 856/1250 [19:34<07:13,  1.10s/it, loss=0.73]
Epoch 1/4:  68%|██████▊   | 856/1250 [19:35<07:13,  1.10s/it, loss=0.834]
Epoch 1/4:  69%|██████▊   | 857/1250 [19:35<07:39,  1.17s/it, loss=0.834]
Epoch 1/4:  69%|██████▊   | 857/1250 [19:36<07:39,  1.17s/it, loss=0.614]
Epoch 1/4:  69%|██████▊   | 858/1250 [19:36<07:00,  1.07s/it, loss=0.614]
Epoch 1/4:  69%|██████▊   | 858/1250 [19:37<07:00,  1.07s/it, loss=0.63] 
Epoch 1/4:  69%|██████▊   | 859/1250 [19:37<06:32,  1.00s/it, loss=0.63]
Epoch 1/4:  69%|██████▊   | 859/1250 [19:38<06:32,  1.00s/it, loss=0.869]
Epoch 1/4:  69%|██████▉   | 860/1250 [19:38<06:07,  1.06it/s, loss=0.869]
Epoch 1/4:  69%|██████▉   | 860/1250 [19:39<06:07,  1.06it/s, loss=0.682]
Epoch 1/4:  69%|██████▉   | 861/1250 [19:39<06:03,  1.07it/s, loss=0.682]
Epoch 1/4:  69%|██████▉   | 861/1250 [19:40<06:03,  1.07it/s, loss=0.947]
Epoch 1/4:  69%|██████▉   | 862/1250 [19:40<06:44,  1.04s/it, loss=0.947]
Epoch 1/4:  69%|██████▉   | 862/1250 [19:41<06:44,  1.04s/it, loss=1.13] 
Epoch 1/4:  69%|██████▉   | 863/1250 [19:41<06:24,  1.01it/s, loss=1.13]
Epoch 1/4:  69%|██████▉   | 863/1250 [19:42<06:24,  1.01it/s, loss=0.817]
Epoch 1/4:  69%|██████▉   | 864/1250 [19:42<06:07,  1.05it/s, loss=0.817]
Epoch 1/4:  69%|██████▉   | 864/1250 [19:43<06:07,  1.05it/s, loss=1.01] 
Epoch 1/4:  69%|██████▉   | 865/1250 [19:43<05:50,  1.10it/s, loss=1.01]
Epoch 1/4:  69%|██████▉   | 865/1250 [19:44<05:50,  1.10it/s, loss=1.21]
Epoch 1/4:  69%|██████▉   | 866/1250 [19:44<05:48,  1.10it/s, loss=1.21]
Epoch 1/4:  69%|██████▉   | 866/1250 [19:45<05:48,  1.10it/s, loss=0.805]
Epoch 1/4:  69%|██████▉   | 867/1250 [19:45<06:18,  1.01it/s, loss=0.805]
Epoch 1/4:  69%|██████▉   | 867/1250 [19:46<06:18,  1.01it/s, loss=1.46] 
Epoch 1/4:  69%|██████▉   | 868/1250 [19:46<06:31,  1.02s/it, loss=1.46]
Epoch 1/4:  69%|██████▉   | 868/1250 [19:47<06:31,  1.02s/it, loss=0.864]
Epoch 1/4:  70%|██████▉   | 869/1250 [19:47<06:06,  1.04it/s, loss=0.864]
Epoch 1/4:  70%|██████▉   | 869/1250 [19:48<06:06,  1.04it/s, loss=0.777]
Epoch 1/4:  70%|██████▉   | 870/1250 [19:48<05:47,  1.09it/s, loss=0.777]
Epoch 1/4:  70%|██████▉   | 870/1250 [19:48<05:47,  1.09it/s, loss=0.786]
Epoch 1/4:  70%|██████▉   | 871/1250 [19:48<05:51,  1.08it/s, loss=0.786]
Epoch 1/4:  70%|██████▉   | 871/1250 [19:50<05:51,  1.08it/s, loss=1.26] 
Epoch 1/4:  70%|██████▉   | 872/1250 [19:50<07:00,  1.11s/it, loss=1.26]
Epoch 1/4:  70%|██████▉   | 872/1250 [19:51<07:00,  1.11s/it, loss=0.826]
Epoch 1/4:  70%|██████▉   | 873/1250 [19:51<07:27,  1.19s/it, loss=0.826]
Epoch 1/4:  70%|██████▉   | 873/1250 [19:52<07:27,  1.19s/it, loss=1.09] 
Epoch 1/4:  70%|██████▉   | 874/1250 [19:52<06:59,  1.12s/it, loss=1.09]
Epoch 1/4:  70%|██████▉   | 874/1250 [19:53<06:59,  1.12s/it, loss=0.968]
Epoch 1/4:  70%|███████   | 875/1250 [19:53<06:34,  1.05s/it, loss=0.968]
Epoch 1/4:  70%|███████   | 875/1250 [19:54<06:34,  1.05s/it, loss=0.949]
Epoch 1/4:  70%|███████   | 876/1250 [19:54<06:20,  1.02s/it, loss=0.949]
Epoch 1/4:  70%|███████   | 876/1250 [19:55<06:20,  1.02s/it, loss=0.943]
Epoch 1/4:  70%|███████   | 877/1250 [19:55<06:54,  1.11s/it, loss=0.943]
Epoch 1/4:  70%|███████   | 877/1250 [19:57<06:54,  1.11s/it, loss=0.66] 
Epoch 1/4:  70%|███████   | 878/1250 [19:57<06:59,  1.13s/it, loss=0.66]
Epoch 1/4:  70%|███████   | 878/1250 [19:58<06:59,  1.13s/it, loss=1.25]
Epoch 1/4:  70%|███████   | 879/1250 [19:58<06:39,  1.08s/it, loss=1.25]
Epoch 1/4:  70%|███████   | 879/1250 [19:59<06:39,  1.08s/it, loss=0.852]
Epoch 1/4:  70%|███████   | 880/1250 [19:59<06:42,  1.09s/it, loss=0.852]
Epoch 1/4:  70%|███████   | 880/1250 [20:00<06:42,  1.09s/it, loss=0.831]
Epoch 1/4:  70%|███████   | 881/1250 [20:00<07:23,  1.20s/it, loss=0.831]
Epoch 1/4:  70%|███████   | 881/1250 [20:02<07:23,  1.20s/it, loss=0.967]
Epoch 1/4:  71%|███████   | 882/1250 [20:02<07:44,  1.26s/it, loss=0.967]
Epoch 1/4:  71%|███████   | 882/1250 [20:03<07:44,  1.26s/it, loss=1.01] 
Epoch 1/4:  71%|███████   | 883/1250 [20:03<07:15,  1.19s/it, loss=1.01]
Epoch 1/4:  71%|███████   | 883/1250 [20:04<07:15,  1.19s/it, loss=1.11]
Epoch 1/4:  71%|███████   | 884/1250 [20:04<06:52,  1.13s/it, loss=1.11]
Epoch 1/4:  71%|███████   | 884/1250 [20:05<06:52,  1.13s/it, loss=0.753]
Epoch 1/4:  71%|███████   | 885/1250 [20:05<07:20,  1.21s/it, loss=0.753]
Epoch 1/4:  71%|███████   | 885/1250 [20:06<07:20,  1.21s/it, loss=0.876]
Epoch 1/4:  71%|███████   | 886/1250 [20:06<07:50,  1.29s/it, loss=0.876]
Epoch 1/4:  71%|███████   | 886/1250 [20:08<07:50,  1.29s/it, loss=1.41] 
Epoch 1/4:  71%|███████   | 887/1250 [20:08<07:20,  1.21s/it, loss=1.41]
Epoch 1/4:  71%|███████   | 887/1250 [20:08<07:20,  1.21s/it, loss=0.816]
Epoch 1/4:  71%|███████   | 888/1250 [20:08<06:47,  1.12s/it, loss=0.816]
Epoch 1/4:  71%|███████   | 888/1250 [20:09<06:47,  1.12s/it, loss=1.34] 
Epoch 1/4:  71%|███████   | 889/1250 [20:09<06:23,  1.06s/it, loss=1.34]
Epoch 1/4:  71%|███████   | 889/1250 [20:10<06:23,  1.06s/it, loss=0.916]
Epoch 1/4:  71%|███████   | 890/1250 [20:10<06:25,  1.07s/it, loss=0.916]
Epoch 1/4:  71%|███████   | 890/1250 [20:12<06:25,  1.07s/it, loss=1]    
Epoch 1/4:  71%|███████▏  | 891/1250 [20:12<06:47,  1.13s/it, loss=1]
Epoch 1/4:  71%|███████▏  | 891/1250 [20:13<06:47,  1.13s/it, loss=1.03]
Epoch 1/4:  71%|███████▏  | 892/1250 [20:13<06:28,  1.09s/it, loss=1.03]
Epoch 1/4:  71%|███████▏  | 892/1250 [20:14<06:28,  1.09s/it, loss=0.687]
Epoch 1/4:  71%|███████▏  | 893/1250 [20:14<06:01,  1.01s/it, loss=0.687]
Epoch 1/4:  71%|███████▏  | 893/1250 [20:14<06:01,  1.01s/it, loss=0.711]
Epoch 1/4:  72%|███████▏  | 894/1250 [20:14<05:55,  1.00it/s, loss=0.711]
Epoch 1/4:  72%|███████▏  | 894/1250 [20:16<05:55,  1.00it/s, loss=1.16] 
Epoch 1/4:  72%|███████▏  | 895/1250 [20:16<05:58,  1.01s/it, loss=1.16]
Epoch 1/4:  72%|███████▏  | 895/1250 [20:17<05:58,  1.01s/it, loss=0.816]
Epoch 1/4:  72%|███████▏  | 896/1250 [20:17<06:07,  1.04s/it, loss=0.816]
Epoch 1/4:  72%|███████▏  | 896/1250 [20:17<06:07,  1.04s/it, loss=0.928]
Epoch 1/4:  72%|███████▏  | 897/1250 [20:17<05:41,  1.03it/s, loss=0.928]
Epoch 1/4:  72%|███████▏  | 897/1250 [20:18<05:41,  1.03it/s, loss=1.27] 
Epoch 1/4:  72%|███████▏  | 898/1250 [20:18<05:23,  1.09it/s, loss=1.27]
Epoch 1/4:  72%|███████▏  | 898/1250 [20:19<05:23,  1.09it/s, loss=0.958]
Epoch 1/4:  72%|███████▏  | 899/1250 [20:19<05:26,  1.07it/s, loss=0.958]
Epoch 1/4:  72%|███████▏  | 899/1250 [20:20<05:26,  1.07it/s, loss=1.08] 
Epoch 1/4:  72%|███████▏  | 900/1250 [20:20<05:42,  1.02it/s, loss=1.08]
Epoch 1/4:  72%|███████▏  | 900/1250 [20:22<05:42,  1.02it/s, loss=0.698]
Epoch 1/4:  72%|███████▏  | 901/1250 [20:22<06:05,  1.05s/it, loss=0.698]
Epoch 1/4:  72%|███████▏  | 901/1250 [20:22<06:05,  1.05s/it, loss=1.33] 
Epoch 1/4:  72%|███████▏  | 902/1250 [20:22<05:56,  1.02s/it, loss=1.33]
Epoch 1/4:  72%|███████▏  | 902/1250 [20:23<05:56,  1.02s/it, loss=1.18]
Epoch 1/4:  72%|███████▏  | 903/1250 [20:23<05:34,  1.04it/s, loss=1.18]
Epoch 1/4:  72%|███████▏  | 903/1250 [20:24<05:34,  1.04it/s, loss=0.854]
Epoch 1/4:  72%|███████▏  | 904/1250 [20:24<05:28,  1.05it/s, loss=0.854]
Epoch 1/4:  72%|███████▏  | 904/1250 [20:25<05:28,  1.05it/s, loss=1.01] 
Epoch 1/4:  72%|███████▏  | 905/1250 [20:25<05:32,  1.04it/s, loss=1.01]
Epoch 1/4:  72%|███████▏  | 905/1250 [20:26<05:32,  1.04it/s, loss=0.667]
Epoch 1/4:  72%|███████▏  | 906/1250 [20:26<05:37,  1.02it/s, loss=0.667]
Epoch 1/4:  72%|███████▏  | 906/1250 [20:27<05:37,  1.02it/s, loss=0.792]
Epoch 1/4:  73%|███████▎  | 907/1250 [20:27<05:44,  1.01s/it, loss=0.792]
Epoch 1/4:  73%|███████▎  | 907/1250 [20:28<05:44,  1.01s/it, loss=0.85] 
Epoch 1/4:  73%|███████▎  | 908/1250 [20:28<05:31,  1.03it/s, loss=0.85]
Epoch 1/4:  73%|███████▎  | 908/1250 [20:29<05:31,  1.03it/s, loss=1.68]
Epoch 1/4:  73%|███████▎  | 909/1250 [20:29<05:25,  1.05it/s, loss=1.68]
Epoch 1/4:  73%|███████▎  | 909/1250 [20:30<05:25,  1.05it/s, loss=1.18]
Epoch 1/4:  73%|███████▎  | 910/1250 [20:30<05:33,  1.02it/s, loss=1.18]
Epoch 1/4:  73%|███████▎  | 910/1250 [20:31<05:33,  1.02it/s, loss=1.26]
Epoch 1/4:  73%|███████▎  | 911/1250 [20:31<05:53,  1.04s/it, loss=1.26]
Epoch 1/4:  73%|███████▎  | 911/1250 [20:32<05:53,  1.04s/it, loss=0.492]
Epoch 1/4:  73%|███████▎  | 912/1250 [20:32<05:55,  1.05s/it, loss=0.492]
Epoch 1/4:  73%|███████▎  | 912/1250 [20:33<05:55,  1.05s/it, loss=1]    
Epoch 1/4:  73%|███████▎  | 913/1250 [20:33<05:32,  1.01it/s, loss=1]
Epoch 1/4:  73%|███████▎  | 913/1250 [20:34<05:32,  1.01it/s, loss=1.03]
Epoch 1/4:  73%|███████▎  | 914/1250 [20:34<05:23,  1.04it/s, loss=1.03]
Epoch 1/4:  73%|███████▎  | 914/1250 [20:35<05:23,  1.04it/s, loss=0.814]
Epoch 1/4:  73%|███████▎  | 915/1250 [20:35<05:36,  1.01s/it, loss=0.814]
Epoch 1/4:  73%|███████▎  | 915/1250 [20:36<05:36,  1.01s/it, loss=0.766]
Epoch 1/4:  73%|███████▎  | 916/1250 [20:36<05:52,  1.06s/it, loss=0.766]
Epoch 1/4:  73%|███████▎  | 916/1250 [20:38<05:52,  1.06s/it, loss=0.807]
Epoch 1/4:  73%|███████▎  | 917/1250 [20:38<06:35,  1.19s/it, loss=0.807]
Epoch 1/4:  73%|███████▎  | 917/1250 [20:39<06:35,  1.19s/it, loss=0.639]
Epoch 1/4:  73%|███████▎  | 918/1250 [20:39<06:19,  1.14s/it, loss=0.639]
Epoch 1/4:  73%|███████▎  | 918/1250 [20:40<06:19,  1.14s/it, loss=0.901]
Epoch 1/4:  74%|███████▎  | 919/1250 [20:40<06:06,  1.11s/it, loss=0.901]
Epoch 1/4:  74%|███████▎  | 919/1250 [20:41<06:06,  1.11s/it, loss=0.617]
Epoch 1/4:  74%|███████▎  | 920/1250 [20:41<06:04,  1.10s/it, loss=0.617]
Epoch 1/4:  74%|███████▎  | 920/1250 [20:42<06:04,  1.10s/it, loss=0.597]
Epoch 1/4:  74%|███████▎  | 921/1250 [20:42<06:22,  1.16s/it, loss=0.597]
Epoch 1/4:  74%|███████▎  | 921/1250 [20:43<06:22,  1.16s/it, loss=1.29] 
Epoch 1/4:  74%|███████▍  | 922/1250 [20:43<06:09,  1.13s/it, loss=1.29]
Epoch 1/4:  74%|███████▍  | 922/1250 [20:44<06:09,  1.13s/it, loss=1.63]
Epoch 1/4:  74%|███████▍  | 923/1250 [20:44<05:53,  1.08s/it, loss=1.63]
Epoch 1/4:  74%|███████▍  | 923/1250 [20:45<05:53,  1.08s/it, loss=0.593]
Epoch 1/4:  74%|███████▍  | 924/1250 [20:45<05:46,  1.06s/it, loss=0.593]
Epoch 1/4:  74%|███████▍  | 924/1250 [20:46<05:46,  1.06s/it, loss=0.891]
Epoch 1/4:  74%|███████▍  | 925/1250 [20:46<05:44,  1.06s/it, loss=0.891]
Epoch 1/4:  74%|███████▍  | 925/1250 [20:48<05:44,  1.06s/it, loss=0.596]
Epoch 1/4:  74%|███████▍  | 926/1250 [20:48<06:08,  1.14s/it, loss=0.596]
Epoch 1/4:  74%|███████▍  | 926/1250 [20:49<06:08,  1.14s/it, loss=0.739]
Epoch 1/4:  74%|███████▍  | 927/1250 [20:49<06:22,  1.18s/it, loss=0.739]
Epoch 1/4:  74%|███████▍  | 927/1250 [20:50<06:22,  1.18s/it, loss=0.827]
Epoch 1/4:  74%|███████▍  | 928/1250 [20:50<06:12,  1.16s/it, loss=0.827]
Epoch 1/4:  74%|███████▍  | 928/1250 [20:51<06:12,  1.16s/it, loss=1.46] 
Epoch 1/4:  74%|███████▍  | 929/1250 [20:51<06:02,  1.13s/it, loss=1.46]
Epoch 1/4:  74%|███████▍  | 929/1250 [20:53<06:02,  1.13s/it, loss=0.705]
Epoch 1/4:  74%|███████▍  | 930/1250 [20:53<06:17,  1.18s/it, loss=0.705]
Epoch 1/4:  74%|███████▍  | 930/1250 [20:54<06:17,  1.18s/it, loss=0.733]
Epoch 1/4:  74%|███████▍  | 931/1250 [20:54<06:01,  1.13s/it, loss=0.733]
Epoch 1/4:  74%|███████▍  | 931/1250 [20:55<06:01,  1.13s/it, loss=1.26] 
Epoch 1/4:  75%|███████▍  | 932/1250 [20:55<05:46,  1.09s/it, loss=1.26]
Epoch 1/4:  75%|███████▍  | 932/1250 [20:56<05:46,  1.09s/it, loss=1.03]
Epoch 1/4:  75%|███████▍  | 933/1250 [20:56<06:21,  1.20s/it, loss=1.03]
Epoch 1/4:  75%|███████▍  | 933/1250 [20:57<06:21,  1.20s/it, loss=0.659]
Epoch 1/4:  75%|███████▍  | 934/1250 [20:57<06:36,  1.25s/it, loss=0.659]
Epoch 1/4:  75%|███████▍  | 934/1250 [20:59<06:36,  1.25s/it, loss=0.817]
Epoch 1/4:  75%|███████▍  | 935/1250 [20:59<06:28,  1.23s/it, loss=0.817]
Epoch 1/4:  75%|███████▍  | 935/1250 [21:00<06:28,  1.23s/it, loss=0.68] 
Epoch 1/4:  75%|███████▍  | 936/1250 [21:00<06:01,  1.15s/it, loss=0.68]
Epoch 1/4:  75%|███████▍  | 936/1250 [21:01<06:01,  1.15s/it, loss=0.677]
Epoch 1/4:  75%|███████▍  | 937/1250 [21:01<05:43,  1.10s/it, loss=0.677]
Epoch 1/4:  75%|███████▍  | 937/1250 [21:01<05:43,  1.10s/it, loss=1.14] 
Epoch 1/4:  75%|███████▌  | 938/1250 [21:01<05:30,  1.06s/it, loss=1.14]
Epoch 1/4:  75%|███████▌  | 938/1250 [21:02<05:30,  1.06s/it, loss=0.6] 
Epoch 1/4:  75%|███████▌  | 939/1250 [21:02<05:18,  1.02s/it, loss=0.6]
Epoch 1/4:  75%|███████▌  | 939/1250 [21:03<05:18,  1.02s/it, loss=0.905]
Epoch 1/4:  75%|███████▌  | 940/1250 [21:03<05:19,  1.03s/it, loss=0.905]
Epoch 1/4:  75%|███████▌  | 940/1250 [21:04<05:19,  1.03s/it, loss=0.854]
Epoch 1/4:  75%|███████▌  | 941/1250 [21:04<05:11,  1.01s/it, loss=0.854]
Epoch 1/4:  75%|███████▌  | 941/1250 [21:06<05:11,  1.01s/it, loss=0.899]
Epoch 1/4:  75%|███████▌  | 942/1250 [21:06<05:26,  1.06s/it, loss=0.899]
Epoch 1/4:  75%|███████▌  | 942/1250 [21:07<05:26,  1.06s/it, loss=0.889]
Epoch 1/4:  75%|███████▌  | 943/1250 [21:07<05:17,  1.03s/it, loss=0.889]
Epoch 1/4:  75%|███████▌  | 943/1250 [21:07<05:17,  1.03s/it, loss=0.981]
Epoch 1/4:  76%|███████▌  | 944/1250 [21:07<05:00,  1.02it/s, loss=0.981]
Epoch 1/4:  76%|███████▌  | 944/1250 [21:09<05:00,  1.02it/s, loss=0.853]
Epoch 1/4:  76%|███████▌  | 945/1250 [21:09<05:13,  1.03s/it, loss=0.853]
Epoch 1/4:  76%|███████▌  | 945/1250 [21:10<05:13,  1.03s/it, loss=0.711]
Epoch 1/4:  76%|███████▌  | 946/1250 [21:10<05:07,  1.01s/it, loss=0.711]
Epoch 1/4:  76%|███████▌  | 946/1250 [21:11<05:07,  1.01s/it, loss=0.645]
Epoch 1/4:  76%|███████▌  | 947/1250 [21:11<05:08,  1.02s/it, loss=0.645]
Epoch 1/4:  76%|███████▌  | 947/1250 [21:11<05:08,  1.02s/it, loss=0.61] 
Epoch 1/4:  76%|███████▌  | 948/1250 [21:11<04:56,  1.02it/s, loss=0.61]
Epoch 1/4:  76%|███████▌  | 948/1250 [21:12<04:56,  1.02it/s, loss=0.661]
Epoch 1/4:  76%|███████▌  | 949/1250 [21:12<04:41,  1.07it/s, loss=0.661]
Epoch 1/4:  76%|███████▌  | 949/1250 [21:13<04:41,  1.07it/s, loss=0.702]
Epoch 1/4:  76%|███████▌  | 950/1250 [21:13<04:59,  1.00it/s, loss=0.702]
Epoch 1/4:  76%|███████▌  | 950/1250 [21:14<04:59,  1.00it/s, loss=1.28] 
Epoch 1/4:  76%|███████▌  | 951/1250 [21:14<05:03,  1.01s/it, loss=1.28]
Epoch 1/4:  76%|███████▌  | 951/1250 [21:16<05:03,  1.01s/it, loss=1.26]
Epoch 1/4:  76%|███████▌  | 952/1250 [21:16<05:07,  1.03s/it, loss=1.26]
Epoch 1/4:  76%|███████▌  | 952/1250 [21:16<05:07,  1.03s/it, loss=0.934]
Epoch 1/4:  76%|███████▌  | 953/1250 [21:16<04:56,  1.00it/s, loss=0.934]
Epoch 1/4:  76%|███████▌  | 953/1250 [21:17<04:56,  1.00it/s, loss=0.91] 
Epoch 1/4:  76%|███████▋  | 954/1250 [21:17<04:44,  1.04it/s, loss=0.91]
Epoch 1/4:  76%|███████▋  | 954/1250 [21:18<04:44,  1.04it/s, loss=1.19]
Epoch 1/4:  76%|███████▋  | 955/1250 [21:18<04:52,  1.01it/s, loss=1.19]
Epoch 1/4:  76%|███████▋  | 955/1250 [21:20<04:52,  1.01it/s, loss=1.04]
Epoch 1/4:  76%|███████▋  | 956/1250 [21:20<05:15,  1.07s/it, loss=1.04]
Epoch 1/4:  76%|███████▋  | 956/1250 [21:21<05:15,  1.07s/it, loss=0.95]
Epoch 1/4:  77%|███████▋  | 957/1250 [21:21<05:17,  1.08s/it, loss=0.95]
Epoch 1/4:  77%|███████▋  | 957/1250 [21:22<05:17,  1.08s/it, loss=0.582]
Epoch 1/4:  77%|███████▋  | 958/1250 [21:22<05:01,  1.03s/it, loss=0.582]
Epoch 1/4:  77%|███████▋  | 958/1250 [21:23<05:01,  1.03s/it, loss=0.537]
Epoch 1/4:  77%|███████▋  | 959/1250 [21:23<04:44,  1.02it/s, loss=0.537]
Epoch 1/4:  77%|███████▋  | 959/1250 [21:23<04:44,  1.02it/s, loss=0.702]
Epoch 1/4:  77%|███████▋  | 960/1250 [21:23<04:36,  1.05it/s, loss=0.702]
Epoch 1/4:  77%|███████▋  | 960/1250 [21:25<04:36,  1.05it/s, loss=0.631]
Epoch 1/4:  77%|███████▋  | 961/1250 [21:25<04:58,  1.03s/it, loss=0.631]
Epoch 1/4:  77%|███████▋  | 961/1250 [21:26<04:58,  1.03s/it, loss=1.04] 
Epoch 1/4:  77%|███████▋  | 962/1250 [21:26<05:29,  1.14s/it, loss=1.04]
Epoch 1/4:  77%|███████▋  | 962/1250 [21:27<05:29,  1.14s/it, loss=1.21]
Epoch 1/4:  77%|███████▋  | 963/1250 [21:27<05:18,  1.11s/it, loss=1.21]
Epoch 1/4:  77%|███████▋  | 963/1250 [21:28<05:18,  1.11s/it, loss=0.799]
Epoch 1/4:  77%|███████▋  | 964/1250 [21:28<05:02,  1.06s/it, loss=0.799]
Epoch 1/4:  77%|███████▋  | 964/1250 [21:29<05:02,  1.06s/it, loss=1.34] 
Epoch 1/4:  77%|███████▋  | 965/1250 [21:29<05:10,  1.09s/it, loss=1.34]
Epoch 1/4:  77%|███████▋  | 965/1250 [21:30<05:10,  1.09s/it, loss=0.739]
Epoch 1/4:  77%|███████▋  | 966/1250 [21:30<05:05,  1.08s/it, loss=0.739]
Epoch 1/4:  77%|███████▋  | 966/1250 [21:31<05:05,  1.08s/it, loss=0.757]
Epoch 1/4:  77%|███████▋  | 967/1250 [21:31<05:03,  1.07s/it, loss=0.757]
Epoch 1/4:  77%|███████▋  | 967/1250 [21:32<05:03,  1.07s/it, loss=1.04] 
Epoch 1/4:  77%|███████▋  | 968/1250 [21:32<04:59,  1.06s/it, loss=1.04]
Epoch 1/4:  77%|███████▋  | 968/1250 [21:33<04:59,  1.06s/it, loss=1.06]
Epoch 1/4:  78%|███████▊  | 969/1250 [21:33<04:40,  1.00it/s, loss=1.06]
Epoch 1/4:  78%|███████▊  | 969/1250 [21:34<04:40,  1.00it/s, loss=0.737]
Epoch 1/4:  78%|███████▊  | 970/1250 [21:34<04:48,  1.03s/it, loss=0.737]
Epoch 1/4:  78%|███████▊  | 970/1250 [21:35<04:48,  1.03s/it, loss=0.704]
Epoch 1/4:  78%|███████▊  | 971/1250 [21:35<04:54,  1.05s/it, loss=0.704]
Epoch 1/4:  78%|███████▊  | 971/1250 [21:37<04:54,  1.05s/it, loss=0.693]
Epoch 1/4:  78%|███████▊  | 972/1250 [21:37<05:02,  1.09s/it, loss=0.693]
Epoch 1/4:  78%|███████▊  | 972/1250 [21:38<05:02,  1.09s/it, loss=0.84] 
Epoch 1/4:  78%|███████▊  | 973/1250 [21:38<05:08,  1.11s/it, loss=0.84]
Epoch 1/4:  78%|███████▊  | 973/1250 [21:39<05:08,  1.11s/it, loss=0.548]
Epoch 1/4:  78%|███████▊  | 974/1250 [21:39<05:04,  1.10s/it, loss=0.548]
Epoch 1/4:  78%|███████▊  | 974/1250 [21:40<05:04,  1.10s/it, loss=1.24] 
Epoch 1/4:  78%|███████▊  | 975/1250 [21:40<05:23,  1.17s/it, loss=1.24]
Epoch 1/4:  78%|███████▊  | 975/1250 [21:42<05:23,  1.17s/it, loss=0.788]
Epoch 1/4:  78%|███████▊  | 976/1250 [21:42<06:12,  1.36s/it, loss=0.788]
Epoch 1/4:  78%|███████▊  | 976/1250 [21:43<06:12,  1.36s/it, loss=1.61] 
Epoch 1/4:  78%|███████▊  | 977/1250 [21:43<05:35,  1.23s/it, loss=1.61]
Epoch 1/4:  78%|███████▊  | 977/1250 [21:44<05:35,  1.23s/it, loss=1.13]
Epoch 1/4:  78%|███████▊  | 978/1250 [21:44<05:12,  1.15s/it, loss=1.13]
Epoch 1/4:  78%|███████▊  | 978/1250 [21:45<05:12,  1.15s/it, loss=1.11]
Epoch 1/4:  78%|███████▊  | 979/1250 [21:45<05:26,  1.21s/it, loss=1.11]
Epoch 1/4:  78%|███████▊  | 979/1250 [21:46<05:26,  1.21s/it, loss=0.802]
Epoch 1/4:  78%|███████▊  | 980/1250 [21:46<05:26,  1.21s/it, loss=0.802]
Epoch 1/4:  78%|███████▊  | 980/1250 [21:48<05:26,  1.21s/it, loss=1]    
Epoch 1/4:  78%|███████▊  | 981/1250 [21:48<05:29,  1.23s/it, loss=1]
Epoch 1/4:  78%|███████▊  | 981/1250 [21:49<05:29,  1.23s/it, loss=0.772]
Epoch 1/4:  79%|███████▊  | 982/1250 [21:49<05:37,  1.26s/it, loss=0.772]
Epoch 1/4:  79%|███████▊  | 982/1250 [21:50<05:37,  1.26s/it, loss=0.973]
Epoch 1/4:  79%|███████▊  | 983/1250 [21:50<05:22,  1.21s/it, loss=0.973]
Epoch 1/4:  79%|███████▊  | 983/1250 [21:51<05:22,  1.21s/it, loss=0.646]
Epoch 1/4:  79%|███████▊  | 984/1250 [21:51<05:11,  1.17s/it, loss=0.646]
Epoch 1/4:  79%|███████▊  | 984/1250 [21:52<05:11,  1.17s/it, loss=0.918]
Epoch 1/4:  79%|███████▉  | 985/1250 [21:52<05:04,  1.15s/it, loss=0.918]
Epoch 1/4:  79%|███████▉  | 985/1250 [21:53<05:04,  1.15s/it, loss=1.11] 
Epoch 1/4:  79%|███████▉  | 986/1250 [21:53<05:02,  1.15s/it, loss=1.11]
Epoch 1/4:  79%|███████▉  | 986/1250 [21:56<05:02,  1.15s/it, loss=0.707]
Epoch 1/4:  79%|███████▉  | 987/1250 [21:56<06:55,  1.58s/it, loss=0.707]
Epoch 1/4:  79%|███████▉  | 987/1250 [21:58<06:55,  1.58s/it, loss=0.944]
Epoch 1/4:  79%|███████▉  | 988/1250 [21:58<07:45,  1.78s/it, loss=0.944]
Epoch 1/4:  79%|███████▉  | 988/1250 [21:59<07:45,  1.78s/it, loss=1]    
Epoch 1/4:  79%|███████▉  | 989/1250 [21:59<07:02,  1.62s/it, loss=1]
Epoch 1/4:  79%|███████▉  | 989/1250 [22:01<07:02,  1.62s/it, loss=0.771]
Epoch 1/4:  79%|███████▉  | 990/1250 [22:01<06:34,  1.52s/it, loss=0.771]
Epoch 1/4:  79%|███████▉  | 990/1250 [22:02<06:34,  1.52s/it, loss=0.676]
Epoch 1/4:  79%|███████▉  | 991/1250 [22:02<06:24,  1.49s/it, loss=0.676]
Epoch 1/4:  79%|███████▉  | 991/1250 [22:03<06:24,  1.49s/it, loss=0.91] 
Epoch 1/4:  79%|███████▉  | 992/1250 [22:03<05:57,  1.38s/it, loss=0.91]
Epoch 1/4:  79%|███████▉  | 992/1250 [22:04<05:57,  1.38s/it, loss=0.626]
Epoch 1/4:  79%|███████▉  | 993/1250 [22:04<05:33,  1.30s/it, loss=0.626]
Epoch 1/4:  79%|███████▉  | 993/1250 [22:06<05:33,  1.30s/it, loss=1.72] 
Epoch 1/4:  80%|███████▉  | 994/1250 [22:06<05:27,  1.28s/it, loss=1.72]
Epoch 1/4:  80%|███████▉  | 994/1250 [22:07<05:27,  1.28s/it, loss=1.02]
Epoch 1/4:  80%|███████▉  | 995/1250 [22:07<05:23,  1.27s/it, loss=1.02]
Epoch 1/4:  80%|███████▉  | 995/1250 [22:08<05:23,  1.27s/it, loss=0.966]
Epoch 1/4:  80%|███████▉  | 996/1250 [22:08<05:19,  1.26s/it, loss=0.966]
Epoch 1/4:  80%|███████▉  | 996/1250 [22:09<05:19,  1.26s/it, loss=0.736]
Epoch 1/4:  80%|███████▉  | 997/1250 [22:09<05:16,  1.25s/it, loss=0.736]
Epoch 1/4:  80%|███████▉  | 997/1250 [22:10<05:16,  1.25s/it, loss=1.08] 
Epoch 1/4:  80%|███████▉  | 998/1250 [22:10<05:02,  1.20s/it, loss=1.08]
Epoch 1/4:  80%|███████▉  | 998/1250 [22:12<05:02,  1.20s/it, loss=1.05]
Epoch 1/4:  80%|███████▉  | 999/1250 [22:12<05:05,  1.22s/it, loss=1.05]
Epoch 1/4:  80%|███████▉  | 999/1250 [22:13<05:05,  1.22s/it, loss=0.621]
Epoch 1/4:  80%|████████  | 1000/1250 [22:13<05:41,  1.36s/it, loss=0.621]
Epoch 1/4:  80%|████████  | 1000/1250 [22:15<05:41,  1.36s/it, loss=0.744]
Epoch 1/4:  80%|████████  | 1001/1250 [22:15<05:39,  1.36s/it, loss=0.744]
Epoch 1/4:  80%|████████  | 1001/1250 [22:16<05:39,  1.36s/it, loss=0.69] 
Epoch 1/4:  80%|████████  | 1002/1250 [22:16<05:16,  1.28s/it, loss=0.69]
Epoch 1/4:  80%|████████  | 1002/1250 [22:17<05:16,  1.28s/it, loss=0.722]
Epoch 1/4:  80%|████████  | 1003/1250 [22:17<05:04,  1.23s/it, loss=0.722]
Epoch 1/4:  80%|████████  | 1003/1250 [22:18<05:04,  1.23s/it, loss=0.466]
Epoch 1/4:  80%|████████  | 1004/1250 [22:18<04:47,  1.17s/it, loss=0.466]
Epoch 1/4:  80%|████████  | 1004/1250 [22:19<04:47,  1.17s/it, loss=1.37] 
Epoch 1/4:  80%|████████  | 1005/1250 [22:19<04:33,  1.11s/it, loss=1.37]
Epoch 1/4:  80%|████████  | 1005/1250 [22:20<04:33,  1.11s/it, loss=0.714]
Epoch 1/4:  80%|████████  | 1006/1250 [22:20<04:27,  1.10s/it, loss=0.714]
Epoch 1/4:  80%|████████  | 1006/1250 [22:21<04:27,  1.10s/it, loss=0.595]
Epoch 1/4:  81%|████████  | 1007/1250 [22:21<04:23,  1.08s/it, loss=0.595]
Epoch 1/4:  81%|████████  | 1007/1250 [22:22<04:23,  1.08s/it, loss=1.15] 
Epoch 1/4:  81%|████████  | 1008/1250 [22:22<04:16,  1.06s/it, loss=1.15]
Epoch 1/4:  81%|████████  | 1008/1250 [22:23<04:16,  1.06s/it, loss=0.655]
Epoch 1/4:  81%|████████  | 1009/1250 [22:23<04:17,  1.07s/it, loss=0.655]
Epoch 1/4:  81%|████████  | 1009/1250 [22:25<04:17,  1.07s/it, loss=0.959]
Epoch 1/4:  81%|████████  | 1010/1250 [22:25<05:35,  1.40s/it, loss=0.959]
Epoch 1/4:  81%|████████  | 1010/1250 [22:28<05:35,  1.40s/it, loss=0.815]
Epoch 1/4:  81%|████████  | 1011/1250 [22:28<06:50,  1.72s/it, loss=0.815]
Epoch 1/4:  81%|████████  | 1011/1250 [22:29<06:50,  1.72s/it, loss=0.85] 
Epoch 1/4:  81%|████████  | 1012/1250 [22:29<06:02,  1.52s/it, loss=0.85]
Epoch 1/4:  81%|████████  | 1012/1250 [22:30<06:02,  1.52s/it, loss=0.657]
Epoch 1/4:  81%|████████  | 1013/1250 [22:30<05:53,  1.49s/it, loss=0.657]
Epoch 1/4:  81%|████████  | 1013/1250 [22:33<05:53,  1.49s/it, loss=0.911]
Epoch 1/4:  81%|████████  | 1014/1250 [22:33<06:43,  1.71s/it, loss=0.911]
Epoch 1/4:  81%|████████  | 1014/1250 [22:34<06:43,  1.71s/it, loss=0.682]
Epoch 1/4:  81%|████████  | 1015/1250 [22:34<06:16,  1.60s/it, loss=0.682]
Epoch 1/4:  81%|████████  | 1015/1250 [22:35<06:16,  1.60s/it, loss=0.587]
Epoch 1/4:  81%|████████▏ | 1016/1250 [22:35<06:02,  1.55s/it, loss=0.587]
Epoch 1/4:  81%|████████▏ | 1016/1250 [22:39<06:02,  1.55s/it, loss=0.751]
Epoch 1/4:  81%|████████▏ | 1017/1250 [22:39<07:59,  2.06s/it, loss=0.751]
Epoch 1/4:  81%|████████▏ | 1017/1250 [22:40<07:59,  2.06s/it, loss=0.702]
Epoch 1/4:  81%|████████▏ | 1018/1250 [22:40<07:02,  1.82s/it, loss=0.702]
Epoch 1/4:  81%|████████▏ | 1018/1250 [22:41<07:02,  1.82s/it, loss=0.567]
Epoch 1/4:  82%|████████▏ | 1019/1250 [22:41<06:25,  1.67s/it, loss=0.567]
Epoch 1/4:  82%|████████▏ | 1019/1250 [22:44<06:25,  1.67s/it, loss=0.763]
Epoch 1/4:  82%|████████▏ | 1020/1250 [22:44<07:22,  1.92s/it, loss=0.763]
Epoch 1/4:  82%|████████▏ | 1020/1250 [22:44<07:22,  1.92s/it, loss=0.767]
Epoch 1/4:  82%|████████▏ | 1021/1250 [22:44<05:59,  1.57s/it, loss=0.767]
Epoch 1/4:  82%|████████▏ | 1021/1250 [22:45<05:59,  1.57s/it, loss=0.65] 
Epoch 1/4:  82%|████████▏ | 1022/1250 [22:45<05:09,  1.36s/it, loss=0.65]
Epoch 1/4:  82%|████████▏ | 1022/1250 [22:47<05:09,  1.36s/it, loss=0.64]
Epoch 1/4:  82%|████████▏ | 1023/1250 [22:47<05:15,  1.39s/it, loss=0.64]
Epoch 1/4:  82%|████████▏ | 1023/1250 [22:48<05:15,  1.39s/it, loss=1.13]
Epoch 1/4:  82%|████████▏ | 1024/1250 [22:48<05:15,  1.39s/it, loss=1.13]
Epoch 1/4:  82%|████████▏ | 1024/1250 [22:49<05:15,  1.39s/it, loss=1.41]
Epoch 1/4:  82%|████████▏ | 1025/1250 [22:49<04:48,  1.28s/it, loss=1.41]
Epoch 1/4:  82%|████████▏ | 1025/1250 [22:50<04:48,  1.28s/it, loss=0.75]
Epoch 1/4:  82%|████████▏ | 1026/1250 [22:50<04:32,  1.22s/it, loss=0.75]
Epoch 1/4:  82%|████████▏ | 1026/1250 [22:51<04:32,  1.22s/it, loss=1.03]
Epoch 1/4:  82%|████████▏ | 1027/1250 [22:51<04:22,  1.18s/it, loss=1.03]
Epoch 1/4:  82%|████████▏ | 1027/1250 [22:53<04:22,  1.18s/it, loss=0.707]
Epoch 1/4:  82%|████████▏ | 1028/1250 [22:53<04:25,  1.20s/it, loss=0.707]
Epoch 1/4:  82%|████████▏ | 1028/1250 [22:54<04:25,  1.20s/it, loss=0.959]
Epoch 1/4:  82%|████████▏ | 1029/1250 [22:54<04:58,  1.35s/it, loss=0.959]
Epoch 1/4:  82%|████████▏ | 1029/1250 [22:55<04:58,  1.35s/it, loss=0.666]
Epoch 1/4:  82%|████████▏ | 1030/1250 [22:55<04:48,  1.31s/it, loss=0.666]
Epoch 1/4:  82%|████████▏ | 1030/1250 [22:57<04:48,  1.31s/it, loss=0.594]
Epoch 1/4:  82%|████████▏ | 1031/1250 [22:57<04:54,  1.35s/it, loss=0.594]
Epoch 1/4:  82%|████████▏ | 1031/1250 [22:59<04:54,  1.35s/it, loss=1.05] 
Epoch 1/4:  83%|████████▎ | 1032/1250 [22:59<05:18,  1.46s/it, loss=1.05]
Epoch 1/4:  83%|████████▎ | 1032/1250 [23:00<05:18,  1.46s/it, loss=0.649]
Epoch 1/4:  83%|████████▎ | 1033/1250 [23:00<04:43,  1.31s/it, loss=0.649]
Epoch 1/4:  83%|████████▎ | 1033/1250 [23:00<04:43,  1.31s/it, loss=0.959]
Epoch 1/4:  83%|████████▎ | 1034/1250 [23:00<04:16,  1.19s/it, loss=0.959]
Epoch 1/4:  83%|████████▎ | 1034/1250 [23:02<04:16,  1.19s/it, loss=0.967]
Epoch 1/4:  83%|████████▎ | 1035/1250 [23:02<04:20,  1.21s/it, loss=0.967]
Epoch 1/4:  83%|████████▎ | 1035/1250 [23:03<04:20,  1.21s/it, loss=0.684]
Epoch 1/4:  83%|████████▎ | 1036/1250 [23:03<04:38,  1.30s/it, loss=0.684]
Epoch 1/4:  83%|████████▎ | 1036/1250 [23:04<04:38,  1.30s/it, loss=0.713]
Epoch 1/4:  83%|████████▎ | 1037/1250 [23:04<04:25,  1.24s/it, loss=0.713]
Epoch 1/4:  83%|████████▎ | 1037/1250 [23:06<04:25,  1.24s/it, loss=0.96] 
Epoch 1/4:  83%|████████▎ | 1038/1250 [23:06<04:21,  1.23s/it, loss=0.96]
Epoch 1/4:  83%|████████▎ | 1038/1250 [23:07<04:21,  1.23s/it, loss=1.15]
Epoch 1/4:  83%|████████▎ | 1039/1250 [23:07<04:24,  1.25s/it, loss=1.15]
Epoch 1/4:  83%|████████▎ | 1039/1250 [23:08<04:24,  1.25s/it, loss=0.755]
Epoch 1/4:  83%|████████▎ | 1040/1250 [23:08<04:29,  1.28s/it, loss=0.755]
Epoch 1/4:  83%|████████▎ | 1040/1250 [23:09<04:29,  1.28s/it, loss=1.39] 
Epoch 1/4:  83%|████████▎ | 1041/1250 [23:09<04:16,  1.23s/it, loss=1.39]
Epoch 1/4:  83%|████████▎ | 1041/1250 [23:10<04:16,  1.23s/it, loss=1.2] 
Epoch 1/4:  83%|████████▎ | 1042/1250 [23:10<03:51,  1.11s/it, loss=1.2]
Epoch 1/4:  83%|████████▎ | 1042/1250 [23:11<03:51,  1.11s/it, loss=0.8]
Epoch 1/4:  83%|████████▎ | 1043/1250 [23:11<03:31,  1.02s/it, loss=0.8]
Epoch 1/4:  83%|████████▎ | 1043/1250 [23:12<03:31,  1.02s/it, loss=1.1]
Epoch 1/4:  84%|████████▎ | 1044/1250 [23:12<03:25,  1.00it/s, loss=1.1]
Epoch 1/4:  84%|████████▎ | 1044/1250 [23:13<03:25,  1.00it/s, loss=0.407]
Epoch 1/4:  84%|████████▎ | 1045/1250 [23:13<03:31,  1.03s/it, loss=0.407]
Epoch 1/4:  84%|████████▎ | 1045/1250 [23:14<03:31,  1.03s/it, loss=0.955]
Epoch 1/4:  84%|████████▎ | 1046/1250 [23:14<03:26,  1.01s/it, loss=0.955]
Epoch 1/4:  84%|████████▎ | 1046/1250 [23:15<03:26,  1.01s/it, loss=1.19] 
Epoch 1/4:  84%|████████▍ | 1047/1250 [23:15<03:14,  1.05it/s, loss=1.19]
Epoch 1/4:  84%|████████▍ | 1047/1250 [23:16<03:14,  1.05it/s, loss=0.916]
Epoch 1/4:  84%|████████▍ | 1048/1250 [23:16<03:03,  1.10it/s, loss=0.916]
Epoch 1/4:  84%|████████▍ | 1048/1250 [23:17<03:03,  1.10it/s, loss=1.14] 
Epoch 1/4:  84%|████████▍ | 1049/1250 [23:17<03:21,  1.00s/it, loss=1.14]
Epoch 1/4:  84%|████████▍ | 1049/1250 [23:18<03:21,  1.00s/it, loss=1.42]
Epoch 1/4:  84%|████████▍ | 1050/1250 [23:18<03:51,  1.16s/it, loss=1.42]
Epoch 1/4:  84%|████████▍ | 1050/1250 [23:19<03:51,  1.16s/it, loss=0.865]
Epoch 1/4:  84%|████████▍ | 1051/1250 [23:19<03:46,  1.14s/it, loss=0.865]
Epoch 1/4:  84%|████████▍ | 1051/1250 [23:20<03:46,  1.14s/it, loss=1.12] 
Epoch 1/4:  84%|████████▍ | 1052/1250 [23:20<03:34,  1.08s/it, loss=1.12]
Epoch 1/4:  84%|████████▍ | 1052/1250 [23:21<03:34,  1.08s/it, loss=1.02]
Epoch 1/4:  84%|████████▍ | 1053/1250 [23:21<03:23,  1.03s/it, loss=1.02]
Epoch 1/4:  84%|████████▍ | 1053/1250 [23:22<03:23,  1.03s/it, loss=0.803]
Epoch 1/4:  84%|████████▍ | 1054/1250 [23:22<03:18,  1.01s/it, loss=0.803]
Epoch 1/4:  84%|████████▍ | 1054/1250 [23:23<03:18,  1.01s/it, loss=0.766]
Epoch 1/4:  84%|████████▍ | 1055/1250 [23:23<03:19,  1.02s/it, loss=0.766]
Epoch 1/4:  84%|████████▍ | 1055/1250 [23:24<03:19,  1.02s/it, loss=0.478]
Epoch 1/4:  84%|████████▍ | 1056/1250 [23:24<03:21,  1.04s/it, loss=0.478]
Epoch 1/4:  84%|████████▍ | 1056/1250 [23:25<03:21,  1.04s/it, loss=0.66] 
Epoch 1/4:  85%|████████▍ | 1057/1250 [23:25<03:17,  1.02s/it, loss=0.66]
Epoch 1/4:  85%|████████▍ | 1057/1250 [23:26<03:17,  1.02s/it, loss=1.12]
Epoch 1/4:  85%|████████▍ | 1058/1250 [23:26<03:18,  1.03s/it, loss=1.12]
Epoch 1/4:  85%|████████▍ | 1058/1250 [23:28<03:18,  1.03s/it, loss=1.34]
Epoch 1/4:  85%|████████▍ | 1059/1250 [23:28<03:22,  1.06s/it, loss=1.34]
Epoch 1/4:  85%|████████▍ | 1059/1250 [23:29<03:22,  1.06s/it, loss=0.965]
Epoch 1/4:  85%|████████▍ | 1060/1250 [23:29<03:30,  1.11s/it, loss=0.965]
Epoch 1/4:  85%|████████▍ | 1060/1250 [23:30<03:30,  1.11s/it, loss=0.793]
Epoch 1/4:  85%|████████▍ | 1061/1250 [23:30<03:29,  1.11s/it, loss=0.793]
Epoch 1/4:  85%|████████▍ | 1061/1250 [23:31<03:29,  1.11s/it, loss=0.729]
Epoch 1/4:  85%|████████▍ | 1062/1250 [23:31<03:21,  1.07s/it, loss=0.729]
Epoch 1/4:  85%|████████▍ | 1062/1250 [23:32<03:21,  1.07s/it, loss=1.14] 
Epoch 1/4:  85%|████████▌ | 1063/1250 [23:32<03:15,  1.05s/it, loss=1.14]
Epoch 1/4:  85%|████████▌ | 1063/1250 [23:33<03:15,  1.05s/it, loss=1.23]
Epoch 1/4:  85%|████████▌ | 1064/1250 [23:33<03:14,  1.04s/it, loss=1.23]
Epoch 1/4:  85%|████████▌ | 1064/1250 [23:34<03:14,  1.04s/it, loss=1.07]
Epoch 1/4:  85%|████████▌ | 1065/1250 [23:34<03:12,  1.04s/it, loss=1.07]
Epoch 1/4:  85%|████████▌ | 1065/1250 [23:35<03:12,  1.04s/it, loss=1.23]
Epoch 1/4:  85%|████████▌ | 1066/1250 [23:35<03:09,  1.03s/it, loss=1.23]
Epoch 1/4:  85%|████████▌ | 1066/1250 [23:36<03:09,  1.03s/it, loss=0.863]
Epoch 1/4:  85%|████████▌ | 1067/1250 [23:36<03:05,  1.02s/it, loss=0.863]
Epoch 1/4:  85%|████████▌ | 1067/1250 [23:37<03:05,  1.02s/it, loss=0.679]
Epoch 1/4:  85%|████████▌ | 1068/1250 [23:37<03:08,  1.04s/it, loss=0.679]
Epoch 1/4:  85%|████████▌ | 1068/1250 [23:38<03:08,  1.04s/it, loss=1.12] 
Epoch 1/4:  86%|████████▌ | 1069/1250 [23:38<03:09,  1.05s/it, loss=1.12]
Epoch 1/4:  86%|████████▌ | 1069/1250 [23:39<03:09,  1.05s/it, loss=0.997]
Epoch 1/4:  86%|████████▌ | 1070/1250 [23:39<03:10,  1.06s/it, loss=0.997]
Epoch 1/4:  86%|████████▌ | 1070/1250 [23:40<03:10,  1.06s/it, loss=1.02] 
Epoch 1/4:  86%|████████▌ | 1071/1250 [23:40<03:07,  1.05s/it, loss=1.02]
Epoch 1/4:  86%|████████▌ | 1071/1250 [23:41<03:07,  1.05s/it, loss=0.973]
Epoch 1/4:  86%|████████▌ | 1072/1250 [23:41<03:00,  1.02s/it, loss=0.973]
Epoch 1/4:  86%|████████▌ | 1072/1250 [23:42<03:00,  1.02s/it, loss=0.687]
Epoch 1/4:  86%|████████▌ | 1073/1250 [23:42<02:54,  1.01it/s, loss=0.687]
Epoch 1/4:  86%|████████▌ | 1073/1250 [23:43<02:54,  1.01it/s, loss=1.31] 
Epoch 1/4:  86%|████████▌ | 1074/1250 [23:43<02:55,  1.00it/s, loss=1.31]
Epoch 1/4:  86%|████████▌ | 1074/1250 [23:44<02:55,  1.00it/s, loss=1.01]
Epoch 1/4:  86%|████████▌ | 1075/1250 [23:44<02:52,  1.01it/s, loss=1.01]
Epoch 1/4:  86%|████████▌ | 1075/1250 [23:45<02:52,  1.01it/s, loss=0.898]
Epoch 1/4:  86%|████████▌ | 1076/1250 [23:45<02:47,  1.04it/s, loss=0.898]
Epoch 1/4:  86%|████████▌ | 1076/1250 [23:46<02:47,  1.04it/s, loss=0.564]
Epoch 1/4:  86%|████████▌ | 1077/1250 [23:46<02:43,  1.06it/s, loss=0.564]
Epoch 1/4:  86%|████████▌ | 1077/1250 [23:47<02:43,  1.06it/s, loss=1.15] 
Epoch 1/4:  86%|████████▌ | 1078/1250 [23:47<02:47,  1.03it/s, loss=1.15]
Epoch 1/4:  86%|████████▌ | 1078/1250 [23:48<02:47,  1.03it/s, loss=0.834]
Epoch 1/4:  86%|████████▋ | 1079/1250 [23:48<02:52,  1.01s/it, loss=0.834]
Epoch 1/4:  86%|████████▋ | 1079/1250 [23:49<02:52,  1.01s/it, loss=1.03] 
Epoch 1/4:  86%|████████▋ | 1080/1250 [23:49<02:50,  1.00s/it, loss=1.03]
Epoch 1/4:  86%|████████▋ | 1080/1250 [23:50<02:50,  1.00s/it, loss=0.597]
Epoch 1/4:  86%|████████▋ | 1081/1250 [23:50<02:44,  1.03it/s, loss=0.597]
Epoch 1/4:  86%|████████▋ | 1081/1250 [23:51<02:44,  1.03it/s, loss=1.15] 
Epoch 1/4:  87%|████████▋ | 1082/1250 [23:51<02:39,  1.06it/s, loss=1.15]
Epoch 1/4:  87%|████████▋ | 1082/1250 [23:52<02:39,  1.06it/s, loss=0.611]
Epoch 1/4:  87%|████████▋ | 1083/1250 [23:52<02:34,  1.08it/s, loss=0.611]
Epoch 1/4:  87%|████████▋ | 1083/1250 [23:53<02:34,  1.08it/s, loss=0.753]
Epoch 1/4:  87%|████████▋ | 1084/1250 [23:53<02:38,  1.05it/s, loss=0.753]
Epoch 1/4:  87%|████████▋ | 1084/1250 [23:54<02:38,  1.05it/s, loss=0.805]
Epoch 1/4:  87%|████████▋ | 1085/1250 [23:54<02:40,  1.03it/s, loss=0.805]
Epoch 1/4:  87%|████████▋ | 1085/1250 [23:55<02:40,  1.03it/s, loss=0.692]
Epoch 1/4:  87%|████████▋ | 1086/1250 [23:55<02:39,  1.03it/s, loss=0.692]
Epoch 1/4:  87%|████████▋ | 1086/1250 [23:56<02:39,  1.03it/s, loss=1.03] 
Epoch 1/4:  87%|████████▋ | 1087/1250 [23:56<02:45,  1.01s/it, loss=1.03]
Epoch 1/4:  87%|████████▋ | 1087/1250 [23:57<02:45,  1.01s/it, loss=1.1] 
Epoch 1/4:  87%|████████▋ | 1088/1250 [23:57<02:56,  1.09s/it, loss=1.1]
Epoch 1/4:  87%|████████▋ | 1088/1250 [23:58<02:56,  1.09s/it, loss=1.1]
Epoch 1/4:  87%|████████▋ | 1089/1250 [23:58<03:03,  1.14s/it, loss=1.1]
Epoch 1/4:  87%|████████▋ | 1089/1250 [23:59<03:03,  1.14s/it, loss=0.814]
Epoch 1/4:  87%|████████▋ | 1090/1250 [23:59<03:01,  1.13s/it, loss=0.814]
Epoch 1/4:  87%|████████▋ | 1090/1250 [24:00<03:01,  1.13s/it, loss=1.08] 
Epoch 1/4:  87%|████████▋ | 1091/1250 [24:00<02:53,  1.09s/it, loss=1.08]
Epoch 1/4:  87%|████████▋ | 1091/1250 [24:01<02:53,  1.09s/it, loss=1.29]
Epoch 1/4:  87%|████████▋ | 1092/1250 [24:01<02:46,  1.06s/it, loss=1.29]
Epoch 1/4:  87%|████████▋ | 1092/1250 [24:02<02:46,  1.06s/it, loss=0.86]
Epoch 1/4:  87%|████████▋ | 1093/1250 [24:02<02:43,  1.04s/it, loss=0.86]
Epoch 1/4:  87%|████████▋ | 1093/1250 [24:03<02:43,  1.04s/it, loss=0.831]
Epoch 1/4:  88%|████████▊ | 1094/1250 [24:03<02:42,  1.04s/it, loss=0.831]
Epoch 1/4:  88%|████████▊ | 1094/1250 [24:04<02:42,  1.04s/it, loss=0.622]
Epoch 1/4:  88%|████████▊ | 1095/1250 [24:04<02:41,  1.04s/it, loss=0.622]
Epoch 1/4:  88%|████████▊ | 1095/1250 [24:05<02:41,  1.04s/it, loss=0.853]
Epoch 1/4:  88%|████████▊ | 1096/1250 [24:05<02:38,  1.03s/it, loss=0.853]
Epoch 1/4:  88%|████████▊ | 1096/1250 [24:07<02:38,  1.03s/it, loss=0.768]
Epoch 1/4:  88%|████████▊ | 1097/1250 [24:07<02:39,  1.04s/it, loss=0.768]
Epoch 1/4:  88%|████████▊ | 1097/1250 [24:08<02:39,  1.04s/it, loss=0.626]
Epoch 1/4:  88%|████████▊ | 1098/1250 [24:08<02:39,  1.05s/it, loss=0.626]
Epoch 1/4:  88%|████████▊ | 1098/1250 [24:09<02:39,  1.05s/it, loss=0.912]
Epoch 1/4:  88%|████████▊ | 1099/1250 [24:09<02:42,  1.07s/it, loss=0.912]
Epoch 1/4:  88%|████████▊ | 1099/1250 [24:10<02:42,  1.07s/it, loss=0.984]
Epoch 1/4:  88%|████████▊ | 1100/1250 [24:10<02:41,  1.08s/it, loss=0.984]
Epoch 1/4:  88%|████████▊ | 1100/1250 [24:11<02:41,  1.08s/it, loss=0.655]
Epoch 1/4:  88%|████████▊ | 1101/1250 [24:11<02:37,  1.05s/it, loss=0.655]
Epoch 1/4:  88%|████████▊ | 1101/1250 [24:12<02:37,  1.05s/it, loss=0.824]
Epoch 1/4:  88%|████████▊ | 1102/1250 [24:12<02:34,  1.04s/it, loss=0.824]
Epoch 1/4:  88%|████████▊ | 1102/1250 [24:13<02:34,  1.04s/it, loss=0.632]
Epoch 1/4:  88%|████████▊ | 1103/1250 [24:13<02:32,  1.04s/it, loss=0.632]
Epoch 1/4:  88%|████████▊ | 1103/1250 [24:14<02:32,  1.04s/it, loss=0.838]
Epoch 1/4:  88%|████████▊ | 1104/1250 [24:14<02:31,  1.04s/it, loss=0.838]
Epoch 1/4:  88%|████████▊ | 1104/1250 [24:15<02:31,  1.04s/it, loss=0.876]
Epoch 1/4:  88%|████████▊ | 1105/1250 [24:15<02:30,  1.04s/it, loss=0.876]
Epoch 1/4:  88%|████████▊ | 1105/1250 [24:16<02:30,  1.04s/it, loss=0.585]
Epoch 1/4:  88%|████████▊ | 1106/1250 [24:16<02:27,  1.03s/it, loss=0.585]
Epoch 1/4:  88%|████████▊ | 1106/1250 [24:17<02:27,  1.03s/it, loss=0.967]
Epoch 1/4:  89%|████████▊ | 1107/1250 [24:17<02:29,  1.04s/it, loss=0.967]
Epoch 1/4:  89%|████████▊ | 1107/1250 [24:18<02:29,  1.04s/it, loss=1.37] 
Epoch 1/4:  89%|████████▊ | 1108/1250 [24:18<02:28,  1.04s/it, loss=1.37]
Epoch 1/4:  89%|████████▊ | 1108/1250 [24:19<02:28,  1.04s/it, loss=1.13]
Epoch 1/4:  89%|████████▊ | 1109/1250 [24:19<02:29,  1.06s/it, loss=1.13]
Epoch 1/4:  89%|████████▊ | 1109/1250 [24:20<02:29,  1.06s/it, loss=0.771]
Epoch 1/4:  89%|████████▉ | 1110/1250 [24:20<02:30,  1.08s/it, loss=0.771]
Epoch 1/4:  89%|████████▉ | 1110/1250 [24:21<02:30,  1.08s/it, loss=0.81] 
Epoch 1/4:  89%|████████▉ | 1111/1250 [24:21<02:26,  1.06s/it, loss=0.81]
Epoch 1/4:  89%|████████▉ | 1111/1250 [24:22<02:26,  1.06s/it, loss=0.846]
Epoch 1/4:  89%|████████▉ | 1112/1250 [24:22<02:23,  1.04s/it, loss=0.846]
Epoch 1/4:  89%|████████▉ | 1112/1250 [24:23<02:23,  1.04s/it, loss=0.825]
Epoch 1/4:  89%|████████▉ | 1113/1250 [24:23<02:19,  1.02s/it, loss=0.825]
Epoch 1/4:  89%|████████▉ | 1113/1250 [24:24<02:19,  1.02s/it, loss=0.888]
Epoch 1/4:  89%|████████▉ | 1114/1250 [24:24<02:14,  1.01it/s, loss=0.888]
Epoch 1/4:  89%|████████▉ | 1114/1250 [24:25<02:14,  1.01it/s, loss=0.956]
Epoch 1/4:  89%|████████▉ | 1115/1250 [24:25<02:09,  1.04it/s, loss=0.956]
Epoch 1/4:  89%|████████▉ | 1115/1250 [24:26<02:09,  1.04it/s, loss=0.82] 
Epoch 1/4:  89%|████████▉ | 1116/1250 [24:26<02:04,  1.08it/s, loss=0.82]
Epoch 1/4:  89%|████████▉ | 1116/1250 [24:27<02:04,  1.08it/s, loss=0.917]
Epoch 1/4:  89%|████████▉ | 1117/1250 [24:27<02:07,  1.04it/s, loss=0.917]
Epoch 1/4:  89%|████████▉ | 1117/1250 [24:28<02:07,  1.04it/s, loss=0.689]
Epoch 1/4:  89%|████████▉ | 1118/1250 [24:28<02:15,  1.03s/it, loss=0.689]
Epoch 1/4:  89%|████████▉ | 1118/1250 [24:29<02:15,  1.03s/it, loss=0.821]
Epoch 1/4:  90%|████████▉ | 1119/1250 [24:29<02:17,  1.05s/it, loss=0.821]
Epoch 1/4:  90%|████████▉ | 1119/1250 [24:30<02:17,  1.05s/it, loss=0.812]
Epoch 1/4:  90%|████████▉ | 1120/1250 [24:30<02:10,  1.00s/it, loss=0.812]
Epoch 1/4:  90%|████████▉ | 1120/1250 [24:31<02:10,  1.00s/it, loss=0.577]
Epoch 1/4:  90%|████████▉ | 1121/1250 [24:31<02:04,  1.03it/s, loss=0.577]
Epoch 1/4:  90%|████████▉ | 1121/1250 [24:32<02:04,  1.03it/s, loss=0.555]
Epoch 1/4:  90%|████████▉ | 1122/1250 [24:32<02:03,  1.04it/s, loss=0.555]
Epoch 1/4:  90%|████████▉ | 1122/1250 [24:33<02:03,  1.04it/s, loss=0.485]
Epoch 1/4:  90%|████████▉ | 1123/1250 [24:33<02:04,  1.02it/s, loss=0.485]
Epoch 1/4:  90%|████████▉ | 1123/1250 [24:34<02:04,  1.02it/s, loss=1.06] 
Epoch 1/4:  90%|████████▉ | 1124/1250 [24:34<02:08,  1.02s/it, loss=1.06]
Epoch 1/4:  90%|████████▉ | 1124/1250 [24:35<02:08,  1.02s/it, loss=0.867]
Epoch 1/4:  90%|█████████ | 1125/1250 [24:35<02:03,  1.01it/s, loss=0.867]
Epoch 1/4:  90%|█████████ | 1125/1250 [24:36<02:03,  1.01it/s, loss=0.707]
Epoch 1/4:  90%|█████████ | 1126/1250 [24:36<01:58,  1.05it/s, loss=0.707]
Epoch 1/4:  90%|█████████ | 1126/1250 [24:37<01:58,  1.05it/s, loss=1.08] 
Epoch 1/4:  90%|█████████ | 1127/1250 [24:37<01:57,  1.04it/s, loss=1.08]
Epoch 1/4:  90%|█████████ | 1127/1250 [24:38<01:57,  1.04it/s, loss=0.784]
Epoch 1/4:  90%|█████████ | 1128/1250 [24:38<02:00,  1.01it/s, loss=0.784]
Epoch 1/4:  90%|█████████ | 1128/1250 [24:39<02:00,  1.01it/s, loss=1.05] 
Epoch 1/4:  90%|█████████ | 1129/1250 [24:39<02:02,  1.01s/it, loss=1.05]
Epoch 1/4:  90%|█████████ | 1129/1250 [24:40<02:02,  1.01s/it, loss=0.818]
Epoch 1/4:  90%|█████████ | 1130/1250 [24:40<01:56,  1.03it/s, loss=0.818]
Epoch 1/4:  90%|█████████ | 1130/1250 [24:41<01:56,  1.03it/s, loss=0.69] 
Epoch 1/4:  90%|█████████ | 1131/1250 [24:41<01:51,  1.07it/s, loss=0.69]
Epoch 1/4:  90%|█████████ | 1131/1250 [24:42<01:51,  1.07it/s, loss=1.2] 
Epoch 1/4:  91%|█████████ | 1132/1250 [24:42<01:54,  1.03it/s, loss=1.2]
Epoch 1/4:  91%|█████████ | 1132/1250 [24:43<01:54,  1.03it/s, loss=0.539]
Epoch 1/4:  91%|█████████ | 1133/1250 [24:43<01:59,  1.02s/it, loss=0.539]
Epoch 1/4:  91%|█████████ | 1133/1250 [24:44<01:59,  1.02s/it, loss=0.992]
Epoch 1/4:  91%|█████████ | 1134/1250 [24:44<02:00,  1.04s/it, loss=0.992]
Epoch 1/4:  91%|█████████ | 1134/1250 [24:45<02:00,  1.04s/it, loss=0.583]
Epoch 1/4:  91%|█████████ | 1135/1250 [24:45<01:54,  1.00it/s, loss=0.583]
Epoch 1/4:  91%|█████████ | 1135/1250 [24:46<01:54,  1.00it/s, loss=1.13] 
Epoch 1/4:  91%|█████████ | 1136/1250 [24:46<01:49,  1.04it/s, loss=1.13]
Epoch 1/4:  91%|█████████ | 1136/1250 [24:47<01:49,  1.04it/s, loss=0.67]
Epoch 1/4:  91%|█████████ | 1137/1250 [24:47<01:49,  1.03it/s, loss=0.67]
Epoch 1/4:  91%|█████████ | 1137/1250 [24:48<01:49,  1.03it/s, loss=0.913]
Epoch 1/4:  91%|█████████ | 1138/1250 [24:48<01:48,  1.03it/s, loss=0.913]
Epoch 1/4:  91%|█████████ | 1138/1250 [24:49<01:48,  1.03it/s, loss=0.7]  
Epoch 1/4:  91%|█████████ | 1139/1250 [24:49<01:51,  1.01s/it, loss=0.7]
Epoch 1/4:  91%|█████████ | 1139/1250 [24:50<01:51,  1.01s/it, loss=1.02]
Epoch 1/4:  91%|█████████ | 1140/1250 [24:50<01:49,  1.01it/s, loss=1.02]
Epoch 1/4:  91%|█████████ | 1140/1250 [24:51<01:49,  1.01it/s, loss=1.22]
Epoch 1/4:  91%|█████████▏| 1141/1250 [24:51<01:44,  1.05it/s, loss=1.22]
Epoch 1/4:  91%|█████████▏| 1141/1250 [24:52<01:44,  1.05it/s, loss=0.685]
Epoch 1/4:  91%|█████████▏| 1142/1250 [24:52<01:40,  1.08it/s, loss=0.685]
Epoch 1/4:  91%|█████████▏| 1142/1250 [24:52<01:40,  1.08it/s, loss=0.932]
Epoch 1/4:  91%|█████████▏| 1143/1250 [24:52<01:37,  1.09it/s, loss=0.932]
Epoch 1/4:  91%|█████████▏| 1143/1250 [24:54<01:37,  1.09it/s, loss=0.572]
Epoch 1/4:  92%|█████████▏| 1144/1250 [24:54<01:43,  1.02it/s, loss=0.572]
Epoch 1/4:  92%|█████████▏| 1144/1250 [24:55<01:43,  1.02it/s, loss=1.02] 
Epoch 1/4:  92%|█████████▏| 1145/1250 [24:55<01:49,  1.04s/it, loss=1.02]
Epoch 1/4:  92%|█████████▏| 1145/1250 [24:56<01:49,  1.04s/it, loss=0.838]
Epoch 1/4:  92%|█████████▏| 1146/1250 [24:56<01:52,  1.08s/it, loss=0.838]
Epoch 1/4:  92%|█████████▏| 1146/1250 [24:57<01:52,  1.08s/it, loss=0.792]
Epoch 1/4:  92%|█████████▏| 1147/1250 [24:57<01:53,  1.10s/it, loss=0.792]
Epoch 1/4:  92%|█████████▏| 1147/1250 [24:58<01:53,  1.10s/it, loss=0.657]
Epoch 1/4:  92%|█████████▏| 1148/1250 [24:58<01:51,  1.09s/it, loss=0.657]
Epoch 1/4:  92%|█████████▏| 1148/1250 [24:59<01:51,  1.09s/it, loss=1.25] 
Epoch 1/4:  92%|█████████▏| 1149/1250 [24:59<01:49,  1.09s/it, loss=1.25]
Epoch 1/4:  92%|█████████▏| 1149/1250 [25:00<01:49,  1.09s/it, loss=0.858]
Epoch 1/4:  92%|█████████▏| 1150/1250 [25:00<01:47,  1.07s/it, loss=0.858]
Epoch 1/4:  92%|█████████▏| 1150/1250 [25:01<01:47,  1.07s/it, loss=0.722]
Epoch 1/4:  92%|█████████▏| 1151/1250 [25:01<01:43,  1.05s/it, loss=0.722]
Epoch 1/4:  92%|█████████▏| 1151/1250 [25:02<01:43,  1.05s/it, loss=0.581]
Epoch 1/4:  92%|█████████▏| 1152/1250 [25:02<01:41,  1.04s/it, loss=0.581]
Epoch 1/4:  92%|█████████▏| 1152/1250 [25:03<01:41,  1.04s/it, loss=0.419]
Epoch 1/4:  92%|█████████▏| 1153/1250 [25:03<01:41,  1.05s/it, loss=0.419]
Epoch 1/4:  92%|█████████▏| 1153/1250 [25:04<01:41,  1.05s/it, loss=1.19] 
Epoch 1/4:  92%|█████████▏| 1154/1250 [25:04<01:41,  1.06s/it, loss=1.19]
Epoch 1/4:  92%|█████████▏| 1154/1250 [25:05<01:41,  1.06s/it, loss=1.18]
Epoch 1/4:  92%|█████████▏| 1155/1250 [25:05<01:39,  1.04s/it, loss=1.18]
Epoch 1/4:  92%|█████████▏| 1155/1250 [25:06<01:39,  1.04s/it, loss=0.614]
Epoch 1/4:  92%|█████████▏| 1156/1250 [25:06<01:37,  1.04s/it, loss=0.614]
Epoch 1/4:  92%|█████████▏| 1156/1250 [25:07<01:37,  1.04s/it, loss=0.94] 
Epoch 1/4:  93%|█████████▎| 1157/1250 [25:07<01:36,  1.03s/it, loss=0.94]
Epoch 1/4:  93%|█████████▎| 1157/1250 [25:09<01:36,  1.03s/it, loss=0.659]
Epoch 1/4:  93%|█████████▎| 1158/1250 [25:09<01:36,  1.05s/it, loss=0.659]
Epoch 1/4:  93%|█████████▎| 1158/1250 [25:10<01:36,  1.05s/it, loss=0.793]
Epoch 1/4:  93%|█████████▎| 1159/1250 [25:10<01:37,  1.07s/it, loss=0.793]
Epoch 1/4:  93%|█████████▎| 1159/1250 [25:11<01:37,  1.07s/it, loss=1.01] 
Epoch 1/4:  93%|█████████▎| 1160/1250 [25:11<01:35,  1.06s/it, loss=1.01]
Epoch 1/4:  93%|█████████▎| 1160/1250 [25:12<01:35,  1.06s/it, loss=0.89]
Epoch 1/4:  93%|█████████▎| 1161/1250 [25:12<01:32,  1.04s/it, loss=0.89]
Epoch 1/4:  93%|█████████▎| 1161/1250 [25:13<01:32,  1.04s/it, loss=0.79]
Epoch 1/4:  93%|█████████▎| 1162/1250 [25:13<01:29,  1.02s/it, loss=0.79]
Epoch 1/4:  93%|█████████▎| 1162/1250 [25:14<01:29,  1.02s/it, loss=0.999]
Epoch 1/4:  93%|█████████▎| 1163/1250 [25:14<01:29,  1.03s/it, loss=0.999]
Epoch 1/4:  93%|█████████▎| 1163/1250 [25:15<01:29,  1.03s/it, loss=0.893]
Epoch 1/4:  93%|█████████▎| 1164/1250 [25:15<01:30,  1.05s/it, loss=0.893]
Epoch 1/4:  93%|█████████▎| 1164/1250 [25:16<01:30,  1.05s/it, loss=0.963]
Epoch 1/4:  93%|█████████▎| 1165/1250 [25:16<01:30,  1.06s/it, loss=0.963]
Epoch 1/4:  93%|█████████▎| 1165/1250 [25:17<01:30,  1.06s/it, loss=0.966]
Epoch 1/4:  93%|█████████▎| 1166/1250 [25:17<01:30,  1.07s/it, loss=0.966]
Epoch 1/4:  93%|█████████▎| 1166/1250 [25:18<01:30,  1.07s/it, loss=0.86] 
Epoch 1/4:  93%|█████████▎| 1167/1250 [25:18<01:28,  1.07s/it, loss=0.86]
Epoch 1/4:  93%|█████████▎| 1167/1250 [25:19<01:28,  1.07s/it, loss=0.843]
Epoch 1/4:  93%|█████████▎| 1168/1250 [25:19<01:28,  1.08s/it, loss=0.843]
Epoch 1/4:  93%|█████████▎| 1168/1250 [25:20<01:28,  1.08s/it, loss=1.02] 
Epoch 1/4:  94%|█████████▎| 1169/1250 [25:20<01:26,  1.07s/it, loss=1.02]
Epoch 1/4:  94%|█████████▎| 1169/1250 [25:21<01:26,  1.07s/it, loss=0.72]
Epoch 1/4:  94%|█████████▎| 1170/1250 [25:21<01:20,  1.00s/it, loss=0.72]
Epoch 1/4:  94%|█████████▎| 1170/1250 [25:22<01:20,  1.00s/it, loss=0.761]
Epoch 1/4:  94%|█████████▎| 1171/1250 [25:22<01:14,  1.06it/s, loss=0.761]
Epoch 1/4:  94%|█████████▎| 1171/1250 [25:23<01:14,  1.06it/s, loss=1.36] 
Epoch 1/4:  94%|█████████▍| 1172/1250 [25:23<01:11,  1.09it/s, loss=1.36]
Epoch 1/4:  94%|█████████▍| 1172/1250 [25:24<01:11,  1.09it/s, loss=0.551]
Epoch 1/4:  94%|█████████▍| 1173/1250 [25:24<01:13,  1.05it/s, loss=0.551]
Epoch 1/4:  94%|█████████▍| 1173/1250 [25:25<01:13,  1.05it/s, loss=0.755]
Epoch 1/4:  94%|█████████▍| 1174/1250 [25:25<01:14,  1.01it/s, loss=0.755]
Epoch 1/4:  94%|█████████▍| 1174/1250 [25:26<01:14,  1.01it/s, loss=0.767]
Epoch 1/4:  94%|█████████▍| 1175/1250 [25:26<01:12,  1.04it/s, loss=0.767]
Epoch 1/4:  94%|█████████▍| 1175/1250 [25:27<01:12,  1.04it/s, loss=0.736]
Epoch 1/4:  94%|█████████▍| 1176/1250 [25:27<01:11,  1.03it/s, loss=0.736]
Epoch 1/4:  94%|█████████▍| 1176/1250 [25:28<01:11,  1.03it/s, loss=0.67] 
Epoch 1/4:  94%|█████████▍| 1177/1250 [25:28<01:10,  1.03it/s, loss=0.67]
Epoch 1/4:  94%|█████████▍| 1177/1250 [25:29<01:10,  1.03it/s, loss=1.11]
Epoch 1/4:  94%|█████████▍| 1178/1250 [25:29<01:14,  1.03s/it, loss=1.11]
Epoch 1/4:  94%|█████████▍| 1178/1250 [25:30<01:14,  1.03s/it, loss=1.01]
Epoch 1/4:  94%|█████████▍| 1179/1250 [25:30<01:24,  1.19s/it, loss=1.01]
Epoch 1/4:  94%|█████████▍| 1179/1250 [25:32<01:24,  1.19s/it, loss=1.09]
Epoch 1/4:  94%|█████████▍| 1180/1250 [25:32<01:40,  1.43s/it, loss=1.09]
Epoch 1/4:  94%|█████████▍| 1180/1250 [25:34<01:40,  1.43s/it, loss=0.75]
Epoch 1/4:  94%|█████████▍| 1181/1250 [25:34<01:34,  1.37s/it, loss=0.75]
Epoch 1/4:  94%|█████████▍| 1181/1250 [25:35<01:34,  1.37s/it, loss=0.967]
Epoch 1/4:  95%|█████████▍| 1182/1250 [25:35<01:27,  1.28s/it, loss=0.967]
Epoch 1/4:  95%|█████████▍| 1182/1250 [25:36<01:27,  1.28s/it, loss=0.956]
Epoch 1/4:  95%|█████████▍| 1183/1250 [25:36<01:29,  1.33s/it, loss=0.956]
Epoch 1/4:  95%|█████████▍| 1183/1250 [25:37<01:29,  1.33s/it, loss=0.899]
Epoch 1/4:  95%|█████████▍| 1184/1250 [25:37<01:22,  1.25s/it, loss=0.899]
Epoch 1/4:  95%|█████████▍| 1184/1250 [25:38<01:22,  1.25s/it, loss=1.64] 
Epoch 1/4:  95%|█████████▍| 1185/1250 [25:38<01:19,  1.23s/it, loss=1.64]
Epoch 1/4:  95%|█████████▍| 1185/1250 [25:40<01:19,  1.23s/it, loss=0.807]
Epoch 1/4:  95%|█████████▍| 1186/1250 [25:40<01:17,  1.21s/it, loss=0.807]
Epoch 1/4:  95%|█████████▍| 1186/1250 [25:41<01:17,  1.21s/it, loss=0.732]
Epoch 1/4:  95%|█████████▍| 1187/1250 [25:41<01:14,  1.18s/it, loss=0.732]
Epoch 1/4:  95%|█████████▍| 1187/1250 [25:42<01:14,  1.18s/it, loss=0.779]
Epoch 1/4:  95%|█████████▌| 1188/1250 [25:42<01:11,  1.15s/it, loss=0.779]
Epoch 1/4:  95%|█████████▌| 1188/1250 [25:43<01:11,  1.15s/it, loss=0.678]
Epoch 1/4:  95%|█████████▌| 1189/1250 [25:43<01:08,  1.12s/it, loss=0.678]
Epoch 1/4:  95%|█████████▌| 1189/1250 [25:45<01:08,  1.12s/it, loss=0.843]
Epoch 1/4:  95%|█████████▌| 1190/1250 [25:45<01:18,  1.31s/it, loss=0.843]
Epoch 1/4:  95%|█████████▌| 1190/1250 [25:46<01:18,  1.31s/it, loss=0.925]
Epoch 1/4:  95%|█████████▌| 1191/1250 [25:46<01:11,  1.21s/it, loss=0.925]
Epoch 1/4:  95%|█████████▌| 1191/1250 [25:47<01:11,  1.21s/it, loss=0.625]
Epoch 1/4:  95%|█████████▌| 1192/1250 [25:47<01:09,  1.19s/it, loss=0.625]
Epoch 1/4:  95%|█████████▌| 1192/1250 [25:48<01:09,  1.19s/it, loss=0.514]
Epoch 1/4:  95%|█████████▌| 1193/1250 [25:48<01:05,  1.15s/it, loss=0.514]
Epoch 1/4:  95%|█████████▌| 1193/1250 [25:49<01:05,  1.15s/it, loss=1.05] 
Epoch 1/4:  96%|█████████▌| 1194/1250 [25:49<01:09,  1.24s/it, loss=1.05]
Epoch 1/4:  96%|█████████▌| 1194/1250 [25:50<01:09,  1.24s/it, loss=0.637]
Epoch 1/4:  96%|█████████▌| 1195/1250 [25:50<01:08,  1.24s/it, loss=0.637]
Epoch 1/4:  96%|█████████▌| 1195/1250 [25:52<01:08,  1.24s/it, loss=0.73] 
Epoch 1/4:  96%|█████████▌| 1196/1250 [25:52<01:05,  1.21s/it, loss=0.73]
Epoch 1/4:  96%|█████████▌| 1196/1250 [25:53<01:05,  1.21s/it, loss=0.582]
Epoch 1/4:  96%|█████████▌| 1197/1250 [25:53<01:01,  1.16s/it, loss=0.582]
Epoch 1/4:  96%|█████████▌| 1197/1250 [25:54<01:01,  1.16s/it, loss=0.666]
Epoch 1/4:  96%|█████████▌| 1198/1250 [25:54<01:00,  1.16s/it, loss=0.666]
Epoch 1/4:  96%|█████████▌| 1198/1250 [25:55<01:00,  1.16s/it, loss=0.959]
Epoch 1/4:  96%|█████████▌| 1199/1250 [25:55<01:00,  1.19s/it, loss=0.959]
Epoch 1/4:  96%|█████████▌| 1199/1250 [25:56<01:00,  1.19s/it, loss=1.18] 
Epoch 1/4:  96%|█████████▌| 1200/1250 [25:56<00:58,  1.17s/it, loss=1.18]
Epoch 1/4:  96%|█████████▌| 1200/1250 [25:57<00:58,  1.17s/it, loss=1.38]
Epoch 1/4:  96%|█████████▌| 1201/1250 [25:57<00:57,  1.16s/it, loss=1.38]
Epoch 1/4:  96%|█████████▌| 1201/1250 [25:58<00:57,  1.16s/it, loss=0.941]
Epoch 1/4:  96%|█████████▌| 1202/1250 [25:58<00:53,  1.11s/it, loss=0.941]
Epoch 1/4:  96%|█████████▌| 1202/1250 [26:00<00:53,  1.11s/it, loss=1.33] 
Epoch 1/4:  96%|█████████▌| 1203/1250 [26:00<00:56,  1.19s/it, loss=1.33]
Epoch 1/4:  96%|█████████▌| 1203/1250 [26:01<00:56,  1.19s/it, loss=1.25]
Epoch 1/4:  96%|█████████▋| 1204/1250 [26:01<00:54,  1.18s/it, loss=1.25]
Epoch 1/4:  96%|█████████▋| 1204/1250 [26:02<00:54,  1.18s/it, loss=1.11]
Epoch 1/4:  96%|█████████▋| 1205/1250 [26:02<00:51,  1.14s/it, loss=1.11]
Epoch 1/4:  96%|█████████▋| 1205/1250 [26:03<00:51,  1.14s/it, loss=1.06]
Epoch 1/4:  96%|█████████▋| 1206/1250 [26:03<00:47,  1.09s/it, loss=1.06]
Epoch 1/4:  96%|█████████▋| 1206/1250 [26:04<00:47,  1.09s/it, loss=0.763]
Epoch 1/4:  97%|█████████▋| 1207/1250 [26:04<00:46,  1.07s/it, loss=0.763]
Epoch 1/4:  97%|█████████▋| 1207/1250 [26:05<00:46,  1.07s/it, loss=1.08] 
Epoch 1/4:  97%|█████████▋| 1208/1250 [26:05<00:47,  1.14s/it, loss=1.08]
Epoch 1/4:  97%|█████████▋| 1208/1250 [26:06<00:47,  1.14s/it, loss=0.662]
Epoch 1/4:  97%|█████████▋| 1209/1250 [26:06<00:48,  1.17s/it, loss=0.662]
Epoch 1/4:  97%|█████████▋| 1209/1250 [26:07<00:48,  1.17s/it, loss=0.713]
Epoch 1/4:  97%|█████████▋| 1210/1250 [26:07<00:45,  1.14s/it, loss=0.713]
Epoch 1/4:  97%|█████████▋| 1210/1250 [26:08<00:45,  1.14s/it, loss=0.812]
Epoch 1/4:  97%|█████████▋| 1211/1250 [26:08<00:42,  1.08s/it, loss=0.812]
Epoch 1/4:  97%|█████████▋| 1211/1250 [26:10<00:42,  1.08s/it, loss=0.773]
Epoch 1/4:  97%|█████████▋| 1212/1250 [26:10<00:43,  1.13s/it, loss=0.773]
Epoch 1/4:  97%|█████████▋| 1212/1250 [26:11<00:43,  1.13s/it, loss=0.639]
Epoch 1/4:  97%|█████████▋| 1213/1250 [26:11<00:42,  1.14s/it, loss=0.639]
Epoch 1/4:  97%|█████████▋| 1213/1250 [26:12<00:42,  1.14s/it, loss=0.714]
Epoch 1/4:  97%|█████████▋| 1214/1250 [26:12<00:39,  1.10s/it, loss=0.714]
Epoch 1/4:  97%|█████████▋| 1214/1250 [26:13<00:39,  1.10s/it, loss=0.857]
Epoch 1/4:  97%|█████████▋| 1215/1250 [26:13<00:36,  1.04s/it, loss=0.857]
Epoch 1/4:  97%|█████████▋| 1215/1250 [26:14<00:36,  1.04s/it, loss=0.606]
Epoch 1/4:  97%|█████████▋| 1216/1250 [26:14<00:34,  1.01s/it, loss=0.606]
Epoch 1/4:  97%|█████████▋| 1216/1250 [26:15<00:34,  1.01s/it, loss=0.747]
Epoch 1/4:  97%|█████████▋| 1217/1250 [26:15<00:36,  1.09s/it, loss=0.747]
Epoch 1/4:  97%|█████████▋| 1217/1250 [26:16<00:36,  1.09s/it, loss=0.607]
Epoch 1/4:  97%|█████████▋| 1218/1250 [26:16<00:38,  1.19s/it, loss=0.607]
Epoch 1/4:  97%|█████████▋| 1218/1250 [26:18<00:38,  1.19s/it, loss=1.08] 
Epoch 1/4:  98%|█████████▊| 1219/1250 [26:18<00:36,  1.17s/it, loss=1.08]
Epoch 1/4:  98%|█████████▊| 1219/1250 [26:18<00:36,  1.17s/it, loss=0.784]
Epoch 1/4:  98%|█████████▊| 1220/1250 [26:18<00:32,  1.09s/it, loss=0.784]
Epoch 1/4:  98%|█████████▊| 1220/1250 [26:19<00:32,  1.09s/it, loss=0.581]
Epoch 1/4:  98%|█████████▊| 1221/1250 [26:19<00:30,  1.06s/it, loss=0.581]
Epoch 1/4:  98%|█████████▊| 1221/1250 [26:21<00:30,  1.06s/it, loss=0.548]
Epoch 1/4:  98%|█████████▊| 1222/1250 [26:21<00:30,  1.09s/it, loss=0.548]
Epoch 1/4:  98%|█████████▊| 1222/1250 [26:21<00:30,  1.09s/it, loss=0.819]
Epoch 1/4:  98%|█████████▊| 1223/1250 [26:21<00:27,  1.03s/it, loss=0.819]
Epoch 1/4:  98%|█████████▊| 1223/1250 [26:22<00:27,  1.03s/it, loss=1.59] 
Epoch 1/4:  98%|█████████▊| 1224/1250 [26:22<00:25,  1.02it/s, loss=1.59]
Epoch 1/4:  98%|█████████▊| 1224/1250 [26:23<00:25,  1.02it/s, loss=0.609]
Epoch 1/4:  98%|█████████▊| 1225/1250 [26:23<00:23,  1.08it/s, loss=0.609]
Epoch 1/4:  98%|█████████▊| 1225/1250 [26:24<00:23,  1.08it/s, loss=0.818]
Epoch 1/4:  98%|█████████▊| 1226/1250 [26:24<00:22,  1.09it/s, loss=0.818]
Epoch 1/4:  98%|█████████▊| 1226/1250 [26:25<00:22,  1.09it/s, loss=0.39] 
Epoch 1/4:  98%|█████████▊| 1227/1250 [26:25<00:23,  1.04s/it, loss=0.39]
Epoch 1/4:  98%|█████████▊| 1227/1250 [26:27<00:23,  1.04s/it, loss=0.6] 
Epoch 1/4:  98%|█████████▊| 1228/1250 [26:27<00:23,  1.08s/it, loss=0.6]
Epoch 1/4:  98%|█████████▊| 1228/1250 [26:28<00:23,  1.08s/it, loss=1.2]
Epoch 1/4:  98%|█████████▊| 1229/1250 [26:28<00:22,  1.06s/it, loss=1.2]
Epoch 1/4:  98%|█████████▊| 1229/1250 [26:28<00:22,  1.06s/it, loss=1.1]
Epoch 1/4:  98%|█████████▊| 1230/1250 [26:28<00:20,  1.02s/it, loss=1.1]
Epoch 1/4:  98%|█████████▊| 1230/1250 [26:29<00:20,  1.02s/it, loss=0.753]
Epoch 1/4:  98%|█████████▊| 1231/1250 [26:30<00:19,  1.03s/it, loss=0.753]
Epoch 1/4:  98%|█████████▊| 1231/1250 [26:31<00:19,  1.03s/it, loss=0.974]
Epoch 1/4:  99%|█████████▊| 1232/1250 [26:31<00:19,  1.08s/it, loss=0.974]
Epoch 1/4:  99%|█████████▊| 1232/1250 [26:32<00:19,  1.08s/it, loss=1.06] 
Epoch 1/4:  99%|█████████▊| 1233/1250 [26:32<00:18,  1.06s/it, loss=1.06]
Epoch 1/4:  99%|█████████▊| 1233/1250 [26:33<00:18,  1.06s/it, loss=1.04]
Epoch 1/4:  99%|█████████▊| 1234/1250 [26:33<00:15,  1.01it/s, loss=1.04]
Epoch 1/4:  99%|█████████▊| 1234/1250 [26:33<00:15,  1.01it/s, loss=0.636]
Epoch 1/4:  99%|█████████▉| 1235/1250 [26:33<00:14,  1.05it/s, loss=0.636]
Epoch 1/4:  99%|█████████▉| 1235/1250 [26:34<00:14,  1.05it/s, loss=1.09] 
Epoch 1/4:  99%|█████████▉| 1236/1250 [26:34<00:13,  1.03it/s, loss=1.09]
Epoch 1/4:  99%|█████████▉| 1236/1250 [26:36<00:13,  1.03it/s, loss=0.824]
Epoch 1/4:  99%|█████████▉| 1237/1250 [26:36<00:14,  1.09s/it, loss=0.824]
Epoch 1/4:  99%|█████████▉| 1237/1250 [26:37<00:14,  1.09s/it, loss=0.929]
Epoch 1/4:  99%|█████████▉| 1238/1250 [26:37<00:13,  1.10s/it, loss=0.929]
Epoch 1/4:  99%|█████████▉| 1238/1250 [26:38<00:13,  1.10s/it, loss=0.842]
Epoch 1/4:  99%|█████████▉| 1239/1250 [26:38<00:11,  1.03s/it, loss=0.842]
Epoch 1/4:  99%|█████████▉| 1239/1250 [26:39<00:11,  1.03s/it, loss=0.83] 
Epoch 1/4:  99%|█████████▉| 1240/1250 [26:39<00:09,  1.01it/s, loss=0.83]
Epoch 1/4:  99%|█████████▉| 1240/1250 [26:40<00:09,  1.01it/s, loss=1.03]
Epoch 1/4:  99%|█████████▉| 1241/1250 [26:40<00:08,  1.01it/s, loss=1.03]
Epoch 1/4:  99%|█████████▉| 1241/1250 [26:41<00:08,  1.01it/s, loss=0.628]
Epoch 1/4:  99%|█████████▉| 1242/1250 [26:41<00:08,  1.06s/it, loss=0.628]
Epoch 1/4:  99%|█████████▉| 1242/1250 [26:42<00:08,  1.06s/it, loss=0.72] 
Epoch 1/4:  99%|█████████▉| 1243/1250 [26:42<00:06,  1.00it/s, loss=0.72]
Epoch 1/4:  99%|█████████▉| 1243/1250 [26:43<00:06,  1.00it/s, loss=0.863]
Epoch 1/4: 100%|█████████▉| 1244/1250 [26:43<00:05,  1.05it/s, loss=0.863]
Epoch 1/4: 100%|█████████▉| 1244/1250 [26:43<00:05,  1.05it/s, loss=0.618]
Epoch 1/4: 100%|█████████▉| 1245/1250 [26:43<00:04,  1.08it/s, loss=0.618]
Epoch 1/4: 100%|█████████▉| 1245/1250 [26:44<00:04,  1.08it/s, loss=0.661]
Epoch 1/4: 100%|█████████▉| 1246/1250 [26:44<00:03,  1.04it/s, loss=0.661]
Epoch 1/4: 100%|█████████▉| 1246/1250 [26:46<00:03,  1.04it/s, loss=0.818]
Epoch 1/4: 100%|█████████▉| 1247/1250 [26:46<00:03,  1.26s/it, loss=0.818]
Epoch 1/4: 100%|█████████▉| 1247/1250 [26:48<00:03,  1.26s/it, loss=0.616]
Epoch 1/4: 100%|█████████▉| 1248/1250 [26:48<00:02,  1.20s/it, loss=0.616]
Epoch 1/4: 100%|█████████▉| 1248/1250 [26:48<00:02,  1.20s/it, loss=0.999]
Epoch 1/4: 100%|█████████▉| 1249/1250 [26:48<00:01,  1.12s/it, loss=0.999]
Epoch 1/4: 100%|█████████▉| 1249/1250 [26:49<00:01,  1.12s/it, loss=0.841]
Epoch 1/4: 100%|██████████| 1250/1250 [26:49<00:00,  1.06s/it, loss=0.841]
Epoch 1/4: 100%|██████████| 1250/1250 [26:49<00:00,  1.29s/it, loss=0.841]
Epoch 1 Avg Loss: 1.4184

Epoch 2/4:   0%|          | 0/1250 [00:00<?, ?it/s]
Epoch 2/4:   0%|          | 0/1250 [00:01<?, ?it/s, loss=0.573]
Epoch 2/4:   0%|          | 1/1250 [00:01<21:54,  1.05s/it, loss=0.573]
Epoch 2/4:   0%|          | 1/1250 [00:02<21:54,  1.05s/it, loss=0.941]
Epoch 2/4:   0%|          | 2/1250 [00:02<23:59,  1.15s/it, loss=0.941]
Epoch 2/4:   0%|          | 2/1250 [00:03<23:59,  1.15s/it, loss=0.859]
Epoch 2/4:   0%|          | 3/1250 [00:03<22:26,  1.08s/it, loss=0.859]
Epoch 2/4:   0%|          | 3/1250 [00:04<22:26,  1.08s/it, loss=0.429]
Epoch 2/4:   0%|          | 4/1250 [00:04<21:27,  1.03s/it, loss=0.429]
Epoch 2/4:   0%|          | 4/1250 [00:05<21:27,  1.03s/it, loss=1.04] 
Epoch 2/4:   0%|          | 5/1250 [00:05<21:13,  1.02s/it, loss=1.04]
Epoch 2/4:   0%|          | 5/1250 [00:06<21:13,  1.02s/it, loss=1]   
Epoch 2/4:   0%|          | 6/1250 [00:06<21:56,  1.06s/it, loss=1]
Epoch 2/4:   0%|          | 6/1250 [00:07<21:56,  1.06s/it, loss=0.51]
Epoch 2/4:   1%|          | 7/1250 [00:07<23:40,  1.14s/it, loss=0.51]
Epoch 2/4:   1%|          | 7/1250 [00:08<23:40,  1.14s/it, loss=0.99]
Epoch 2/4:   1%|          | 8/1250 [00:08<22:45,  1.10s/it, loss=0.99]
Epoch 2/4:   1%|          | 8/1250 [00:09<22:45,  1.10s/it, loss=0.735]
Epoch 2/4:   1%|          | 9/1250 [00:09<22:02,  1.07s/it, loss=0.735]
Epoch 2/4:   1%|          | 9/1250 [00:10<22:02,  1.07s/it, loss=0.712]
Epoch 2/4:   1%|          | 10/1250 [00:10<21:43,  1.05s/it, loss=0.712]
Epoch 2/4:   1%|          | 10/1250 [00:11<21:43,  1.05s/it, loss=0.506]
Epoch 2/4:   1%|          | 11/1250 [00:11<22:44,  1.10s/it, loss=0.506]
Epoch 2/4:   1%|          | 11/1250 [00:13<22:44,  1.10s/it, loss=0.921]
Epoch 2/4:   1%|          | 12/1250 [00:13<23:03,  1.12s/it, loss=0.921]
Epoch 2/4:   1%|          | 12/1250 [00:14<23:03,  1.12s/it, loss=0.516]
Epoch 2/4:   1%|          | 13/1250 [00:14<24:24,  1.18s/it, loss=0.516]
Epoch 2/4:   1%|          | 13/1250 [00:15<24:24,  1.18s/it, loss=0.538]
Epoch 2/4:   1%|          | 14/1250 [00:15<23:34,  1.14s/it, loss=0.538]
Epoch 2/4:   1%|          | 14/1250 [00:16<23:34,  1.14s/it, loss=0.594]
Epoch 2/4:   1%|          | 15/1250 [00:16<23:04,  1.12s/it, loss=0.594]
Epoch 2/4:   1%|          | 15/1250 [00:17<23:04,  1.12s/it, loss=0.837]
Epoch 2/4:   1%|▏         | 16/1250 [00:17<24:36,  1.20s/it, loss=0.837]
Epoch 2/4:   1%|▏         | 16/1250 [00:18<24:36,  1.20s/it, loss=0.946]
Epoch 2/4:   1%|▏         | 17/1250 [00:18<23:17,  1.13s/it, loss=0.946]
Epoch 2/4:   1%|▏         | 17/1250 [00:19<23:17,  1.13s/it, loss=0.697]
Epoch 2/4:   1%|▏         | 18/1250 [00:19<22:05,  1.08s/it, loss=0.697]
Epoch 2/4:   1%|▏         | 18/1250 [00:20<22:05,  1.08s/it, loss=0.538]
Epoch 2/4:   2%|▏         | 19/1250 [00:20<21:21,  1.04s/it, loss=0.538]
Epoch 2/4:   2%|▏         | 19/1250 [00:21<21:21,  1.04s/it, loss=0.61] 
Epoch 2/4:   2%|▏         | 20/1250 [00:21<21:21,  1.04s/it, loss=0.61]
Epoch 2/4:   2%|▏         | 20/1250 [00:23<21:21,  1.04s/it, loss=0.743]
Epoch 2/4:   2%|▏         | 21/1250 [00:23<23:10,  1.13s/it, loss=0.743]
Epoch 2/4:   2%|▏         | 21/1250 [00:24<23:10,  1.13s/it, loss=0.684]
Epoch 2/4:   2%|▏         | 22/1250 [00:24<21:20,  1.04s/it, loss=0.684]
Epoch 2/4:   2%|▏         | 22/1250 [00:24<21:20,  1.04s/it, loss=0.791]
Epoch 2/4:   2%|▏         | 23/1250 [00:24<20:16,  1.01it/s, loss=0.791]
Epoch 2/4:   2%|▏         | 23/1250 [00:25<20:16,  1.01it/s, loss=0.574]
Epoch 2/4:   2%|▏         | 24/1250 [00:25<19:47,  1.03it/s, loss=0.574]
Epoch 2/4:   2%|▏         | 24/1250 [00:26<19:47,  1.03it/s, loss=0.544]
Epoch 2/4:   2%|▏         | 25/1250 [00:26<20:28,  1.00s/it, loss=0.544]
Epoch 2/4:   2%|▏         | 25/1250 [00:28<20:28,  1.00s/it, loss=0.551]
Epoch 2/4:   2%|▏         | 26/1250 [00:28<22:22,  1.10s/it, loss=0.551]
Epoch 2/4:   2%|▏         | 26/1250 [00:29<22:22,  1.10s/it, loss=0.656]
Epoch 2/4:   2%|▏         | 27/1250 [00:29<20:51,  1.02s/it, loss=0.656]
Epoch 2/4:   2%|▏         | 27/1250 [00:30<20:51,  1.02s/it, loss=0.652]
Epoch 2/4:   2%|▏         | 28/1250 [00:30<21:17,  1.05s/it, loss=0.652]
Epoch 2/4:   2%|▏         | 28/1250 [00:31<21:17,  1.05s/it, loss=0.669]
Epoch 2/4:   2%|▏         | 29/1250 [00:31<21:46,  1.07s/it, loss=0.669]
Epoch 2/4:   2%|▏         | 29/1250 [00:32<21:46,  1.07s/it, loss=0.719]
Epoch 2/4:   2%|▏         | 30/1250 [00:32<20:46,  1.02s/it, loss=0.719]
Epoch 2/4:   2%|▏         | 30/1250 [00:33<20:46,  1.02s/it, loss=0.546]
Epoch 2/4:   2%|▏         | 31/1250 [00:33<21:30,  1.06s/it, loss=0.546]
Epoch 2/4:   2%|▏         | 31/1250 [00:34<21:30,  1.06s/it, loss=1.01] 
Epoch 2/4:   3%|▎         | 32/1250 [00:34<20:35,  1.01s/it, loss=1.01]
Epoch 2/4:   3%|▎         | 32/1250 [00:35<20:35,  1.01s/it, loss=0.801]
Epoch 2/4:   3%|▎         | 33/1250 [00:35<20:08,  1.01it/s, loss=0.801]
Epoch 2/4:   3%|▎         | 33/1250 [00:36<20:08,  1.01it/s, loss=0.608]
Epoch 2/4:   3%|▎         | 34/1250 [00:36<20:10,  1.00it/s, loss=0.608]
Epoch 2/4:   3%|▎         | 34/1250 [00:37<20:10,  1.00it/s, loss=0.623]
Epoch 2/4:   3%|▎         | 35/1250 [00:37<22:54,  1.13s/it, loss=0.623]
Epoch 2/4:   3%|▎         | 35/1250 [00:38<22:54,  1.13s/it, loss=0.403]
Epoch 2/4:   3%|▎         | 36/1250 [00:38<23:31,  1.16s/it, loss=0.403]
Epoch 2/4:   3%|▎         | 36/1250 [00:39<23:31,  1.16s/it, loss=0.554]
Epoch 2/4:   3%|▎         | 37/1250 [00:39<22:03,  1.09s/it, loss=0.554]
Epoch 2/4:   3%|▎         | 37/1250 [00:40<22:03,  1.09s/it, loss=0.896]
Epoch 2/4:   3%|▎         | 38/1250 [00:40<21:24,  1.06s/it, loss=0.896]
Epoch 2/4:   3%|▎         | 38/1250 [00:41<21:24,  1.06s/it, loss=0.628]
Epoch 2/4:   3%|▎         | 39/1250 [00:41<21:06,  1.05s/it, loss=0.628]
Epoch 2/4:   3%|▎         | 39/1250 [00:42<21:06,  1.05s/it, loss=0.632]
Epoch 2/4:   3%|▎         | 40/1250 [00:42<20:29,  1.02s/it, loss=0.632]
Epoch 2/4:   3%|▎         | 40/1250 [00:43<20:29,  1.02s/it, loss=0.661]
Epoch 2/4:   3%|▎         | 41/1250 [00:43<21:08,  1.05s/it, loss=0.661]
Epoch 2/4:   3%|▎         | 41/1250 [00:44<21:08,  1.05s/it, loss=1.14] 
Epoch 2/4:   3%|▎         | 42/1250 [00:44<20:43,  1.03s/it, loss=1.14]
Epoch 2/4:   3%|▎         | 42/1250 [00:45<20:43,  1.03s/it, loss=0.68]
Epoch 2/4:   3%|▎         | 43/1250 [00:45<21:22,  1.06s/it, loss=0.68]
Epoch 2/4:   3%|▎         | 43/1250 [00:47<21:22,  1.06s/it, loss=0.499]
Epoch 2/4:   4%|▎         | 44/1250 [00:47<23:18,  1.16s/it, loss=0.499]
Epoch 2/4:   4%|▎         | 44/1250 [00:48<23:18,  1.16s/it, loss=0.724]
Epoch 2/4:   4%|▎         | 45/1250 [00:48<23:25,  1.17s/it, loss=0.724]
Epoch 2/4:   4%|▎         | 45/1250 [00:49<23:25,  1.17s/it, loss=0.818]
Epoch 2/4:   4%|▎         | 46/1250 [00:49<23:10,  1.15s/it, loss=0.818]
Epoch 2/4:   4%|▎         | 46/1250 [00:50<23:10,  1.15s/it, loss=0.529]
Epoch 2/4:   4%|▍         | 47/1250 [00:50<22:08,  1.10s/it, loss=0.529]
Epoch 2/4:   4%|▍         | 47/1250 [00:51<22:08,  1.10s/it, loss=0.576]
Epoch 2/4:   4%|▍         | 48/1250 [00:51<21:32,  1.07s/it, loss=0.576]
Epoch 2/4:   4%|▍         | 48/1250 [00:52<21:32,  1.07s/it, loss=0.886]
Epoch 2/4:   4%|▍         | 49/1250 [00:52<21:05,  1.05s/it, loss=0.886]
Epoch 2/4:   4%|▍         | 49/1250 [00:53<21:05,  1.05s/it, loss=1.15] 
Epoch 2/4:   4%|▍         | 50/1250 [00:53<21:23,  1.07s/it, loss=1.15]
Epoch 2/4:   4%|▍         | 50/1250 [00:54<21:23,  1.07s/it, loss=0.739]
Epoch 2/4:   4%|▍         | 51/1250 [00:54<21:50,  1.09s/it, loss=0.739]
Epoch 2/4:   4%|▍         | 51/1250 [00:55<21:50,  1.09s/it, loss=0.839]
Epoch 2/4:   4%|▍         | 52/1250 [00:55<21:26,  1.07s/it, loss=0.839]
Epoch 2/4:   4%|▍         | 52/1250 [00:57<21:26,  1.07s/it, loss=0.721]
Epoch 2/4:   4%|▍         | 53/1250 [00:57<22:27,  1.13s/it, loss=0.721]
Epoch 2/4:   4%|▍         | 53/1250 [00:58<22:27,  1.13s/it, loss=0.379]
Epoch 2/4:   4%|▍         | 54/1250 [00:58<22:30,  1.13s/it, loss=0.379]
Epoch 2/4:   4%|▍         | 54/1250 [00:59<22:30,  1.13s/it, loss=0.576]
Epoch 2/4:   4%|▍         | 55/1250 [00:59<22:53,  1.15s/it, loss=0.576]
Epoch 2/4:   4%|▍         | 55/1250 [01:00<22:53,  1.15s/it, loss=0.678]
Epoch 2/4:   4%|▍         | 56/1250 [01:00<21:56,  1.10s/it, loss=0.678]
Epoch 2/4:   4%|▍         | 56/1250 [01:01<21:56,  1.10s/it, loss=0.722]
Epoch 2/4:   5%|▍         | 57/1250 [01:01<21:22,  1.07s/it, loss=0.722]
Epoch 2/4:   5%|▍         | 57/1250 [01:02<21:22,  1.07s/it, loss=0.834]
Epoch 2/4:   5%|▍         | 58/1250 [01:02<21:28,  1.08s/it, loss=0.834]
Epoch 2/4:   5%|▍         | 58/1250 [01:03<21:28,  1.08s/it, loss=0.629]
Epoch 2/4:   5%|▍         | 59/1250 [01:03<21:10,  1.07s/it, loss=0.629]
Epoch 2/4:   5%|▍         | 59/1250 [01:04<21:10,  1.07s/it, loss=0.434]
Epoch 2/4:   5%|▍         | 60/1250 [01:04<21:31,  1.09s/it, loss=0.434]
Epoch 2/4:   5%|▍         | 60/1250 [01:05<21:31,  1.09s/it, loss=0.816]
Epoch 2/4:   5%|▍         | 61/1250 [01:05<21:23,  1.08s/it, loss=0.816]
Epoch 2/4:   5%|▍         | 61/1250 [01:07<21:23,  1.08s/it, loss=0.544]
Epoch 2/4:   5%|▍         | 62/1250 [01:07<22:21,  1.13s/it, loss=0.544]
Epoch 2/4:   5%|▍         | 62/1250 [01:08<22:21,  1.13s/it, loss=0.644]
Epoch 2/4:   5%|▌         | 63/1250 [01:08<22:01,  1.11s/it, loss=0.644]
Epoch 2/4:   5%|▌         | 63/1250 [01:09<22:01,  1.11s/it, loss=0.496]
Epoch 2/4:   5%|▌         | 64/1250 [01:09<21:31,  1.09s/it, loss=0.496]
Epoch 2/4:   5%|▌         | 64/1250 [01:10<21:31,  1.09s/it, loss=0.7]  
Epoch 2/4:   5%|▌         | 65/1250 [01:10<22:00,  1.11s/it, loss=0.7]
Epoch 2/4:   5%|▌         | 65/1250 [01:11<22:00,  1.11s/it, loss=0.899]
Epoch 2/4:   5%|▌         | 66/1250 [01:11<21:25,  1.09s/it, loss=0.899]
Epoch 2/4:   5%|▌         | 66/1250 [01:12<21:25,  1.09s/it, loss=1.15] 
Epoch 2/4:   5%|▌         | 67/1250 [01:12<21:07,  1.07s/it, loss=1.15]
Epoch 2/4:   5%|▌         | 67/1250 [01:13<21:07,  1.07s/it, loss=0.667]
Epoch 2/4:   5%|▌         | 68/1250 [01:13<20:20,  1.03s/it, loss=0.667]
Epoch 2/4:   5%|▌         | 68/1250 [01:14<20:20,  1.03s/it, loss=0.768]
Epoch 2/4:   6%|▌         | 69/1250 [01:14<19:21,  1.02it/s, loss=0.768]
Epoch 2/4:   6%|▌         | 69/1250 [01:15<19:21,  1.02it/s, loss=0.56] 
Epoch 2/4:   6%|▌         | 70/1250 [01:15<19:13,  1.02it/s, loss=0.56]
Epoch 2/4:   6%|▌         | 70/1250 [01:16<19:13,  1.02it/s, loss=0.599]
Epoch 2/4:   6%|▌         | 71/1250 [01:16<19:04,  1.03it/s, loss=0.599]
Epoch 2/4:   6%|▌         | 71/1250 [01:17<19:04,  1.03it/s, loss=0.768]
Epoch 2/4:   6%|▌         | 72/1250 [01:17<20:40,  1.05s/it, loss=0.768]
Epoch 2/4:   6%|▌         | 72/1250 [01:18<20:40,  1.05s/it, loss=0.746]
Epoch 2/4:   6%|▌         | 73/1250 [01:18<20:43,  1.06s/it, loss=0.746]
Epoch 2/4:   6%|▌         | 73/1250 [01:19<20:43,  1.06s/it, loss=0.685]
Epoch 2/4:   6%|▌         | 74/1250 [01:19<20:11,  1.03s/it, loss=0.685]
Epoch 2/4:   6%|▌         | 74/1250 [01:20<20:11,  1.03s/it, loss=0.48] 
Epoch 2/4:   6%|▌         | 75/1250 [01:20<21:02,  1.07s/it, loss=0.48]
Epoch 2/4:   6%|▌         | 75/1250 [01:21<21:02,  1.07s/it, loss=0.516]
Epoch 2/4:   6%|▌         | 76/1250 [01:21<20:52,  1.07s/it, loss=0.516]
Epoch 2/4:   6%|▌         | 76/1250 [01:22<20:52,  1.07s/it, loss=0.771]
Epoch 2/4:   6%|▌         | 77/1250 [01:22<19:39,  1.01s/it, loss=0.771]
Epoch 2/4:   6%|▌         | 77/1250 [01:23<19:39,  1.01s/it, loss=0.586]
Epoch 2/4:   6%|▌         | 78/1250 [01:23<19:01,  1.03it/s, loss=0.586]
Epoch 2/4:   6%|▌         | 78/1250 [01:24<19:01,  1.03it/s, loss=0.683]
Epoch 2/4:   6%|▋         | 79/1250 [01:24<18:27,  1.06it/s, loss=0.683]
Epoch 2/4:   6%|▋         | 79/1250 [01:25<18:27,  1.06it/s, loss=0.791]
Epoch 2/4:   6%|▋         | 80/1250 [01:25<19:29,  1.00it/s, loss=0.791]
Epoch 2/4:   6%|▋         | 80/1250 [01:26<19:29,  1.00it/s, loss=0.609]
Epoch 2/4:   6%|▋         | 81/1250 [01:26<20:30,  1.05s/it, loss=0.609]
Epoch 2/4:   6%|▋         | 81/1250 [01:27<20:30,  1.05s/it, loss=0.621]
Epoch 2/4:   7%|▋         | 82/1250 [01:27<21:25,  1.10s/it, loss=0.621]
Epoch 2/4:   7%|▋         | 82/1250 [01:28<21:25,  1.10s/it, loss=0.923]
Epoch 2/4:   7%|▋         | 83/1250 [01:28<19:59,  1.03s/it, loss=0.923]
Epoch 2/4:   7%|▋         | 83/1250 [01:29<19:59,  1.03s/it, loss=0.507]
Epoch 2/4:   7%|▋         | 84/1250 [01:29<18:30,  1.05it/s, loss=0.507]
Epoch 2/4:   7%|▋         | 84/1250 [01:30<18:30,  1.05it/s, loss=0.609]
Epoch 2/4:   7%|▋         | 85/1250 [01:30<19:26,  1.00s/it, loss=0.609]
Epoch 2/4:   7%|▋         | 85/1250 [01:31<19:26,  1.00s/it, loss=0.545]
Epoch 2/4:   7%|▋         | 86/1250 [01:31<20:23,  1.05s/it, loss=0.545]
Epoch 2/4:   7%|▋         | 86/1250 [01:32<20:23,  1.05s/it, loss=0.429]
Epoch 2/4:   7%|▋         | 87/1250 [01:32<19:57,  1.03s/it, loss=0.429]
Epoch 2/4:   7%|▋         | 87/1250 [01:33<19:57,  1.03s/it, loss=0.55] 
Epoch 2/4:   7%|▋         | 88/1250 [01:33<19:27,  1.00s/it, loss=0.55]
Epoch 2/4:   7%|▋         | 88/1250 [01:34<19:27,  1.00s/it, loss=0.738]
Epoch 2/4:   7%|▋         | 89/1250 [01:34<18:27,  1.05it/s, loss=0.738]
Epoch 2/4:   7%|▋         | 89/1250 [01:35<18:27,  1.05it/s, loss=0.538]
Epoch 2/4:   7%|▋         | 90/1250 [01:35<19:21,  1.00s/it, loss=0.538]
Epoch 2/4:   7%|▋         | 90/1250 [01:37<19:21,  1.00s/it, loss=0.508]
Epoch 2/4:   7%|▋         | 91/1250 [01:37<24:51,  1.29s/it, loss=0.508]
Epoch 2/4:   7%|▋         | 91/1250 [01:38<24:51,  1.29s/it, loss=0.571]
Epoch 2/4:   7%|▋         | 92/1250 [01:38<22:25,  1.16s/it, loss=0.571]
Epoch 2/4:   7%|▋         | 92/1250 [01:39<22:25,  1.16s/it, loss=0.721]
Epoch 2/4:   7%|▋         | 93/1250 [01:39<20:36,  1.07s/it, loss=0.721]
Epoch 2/4:   7%|▋         | 93/1250 [01:40<20:36,  1.07s/it, loss=0.582]
Epoch 2/4:   8%|▊         | 94/1250 [01:40<19:05,  1.01it/s, loss=0.582]
Epoch 2/4:   8%|▊         | 94/1250 [01:41<19:05,  1.01it/s, loss=0.65] 
Epoch 2/4:   8%|▊         | 95/1250 [01:41<20:08,  1.05s/it, loss=0.65]
Epoch 2/4:   8%|▊         | 95/1250 [01:42<20:08,  1.05s/it, loss=0.818]
Epoch 2/4:   8%|▊         | 96/1250 [01:42<20:29,  1.07s/it, loss=0.818]
Epoch 2/4:   8%|▊         | 96/1250 [01:43<20:29,  1.07s/it, loss=0.575]
Epoch 2/4:   8%|▊         | 97/1250 [01:43<20:18,  1.06s/it, loss=0.575]
Epoch 2/4:   8%|▊         | 97/1250 [01:44<20:18,  1.06s/it, loss=0.936]
Epoch 2/4:   8%|▊         | 98/1250 [01:44<20:13,  1.05s/it, loss=0.936]
Epoch 2/4:   8%|▊         | 98/1250 [01:45<20:13,  1.05s/it, loss=0.77] 
Epoch 2/4:   8%|▊         | 99/1250 [01:45<20:39,  1.08s/it, loss=0.77]
Epoch 2/4:   8%|▊         | 99/1250 [01:47<20:39,  1.08s/it, loss=0.618]
Epoch 2/4:   8%|▊         | 100/1250 [01:47<25:38,  1.34s/it, loss=0.618]
Epoch 2/4:   8%|▊         | 100/1250 [01:48<25:38,  1.34s/it, loss=0.506]
Epoch 2/4:   8%|▊         | 101/1250 [01:48<23:12,  1.21s/it, loss=0.506]
Epoch 2/4:   8%|▊         | 101/1250 [01:49<23:12,  1.21s/it, loss=0.569]
Epoch 2/4:   8%|▊         | 102/1250 [01:49<21:58,  1.15s/it, loss=0.569]
Epoch 2/4:   8%|▊         | 102/1250 [01:50<21:58,  1.15s/it, loss=0.536]
Epoch 2/4:   8%|▊         | 103/1250 [01:50<24:00,  1.26s/it, loss=0.536]
Epoch 2/4:   8%|▊         | 103/1250 [01:52<24:00,  1.26s/it, loss=0.522]
Epoch 2/4:   8%|▊         | 104/1250 [01:52<24:49,  1.30s/it, loss=0.522]
Epoch 2/4:   8%|▊         | 104/1250 [01:53<24:49,  1.30s/it, loss=0.57] 
Epoch 2/4:   8%|▊         | 105/1250 [01:53<24:09,  1.27s/it, loss=0.57]
Epoch 2/4:   8%|▊         | 105/1250 [01:54<24:09,  1.27s/it, loss=0.659]
Epoch 2/4:   8%|▊         | 106/1250 [01:54<23:37,  1.24s/it, loss=0.659]
Epoch 2/4:   8%|▊         | 106/1250 [01:55<23:37,  1.24s/it, loss=0.624]
Epoch 2/4:   9%|▊         | 107/1250 [01:55<22:18,  1.17s/it, loss=0.624]
Epoch 2/4:   9%|▊         | 107/1250 [01:57<22:18,  1.17s/it, loss=0.765]
Epoch 2/4:   9%|▊         | 108/1250 [01:57<24:18,  1.28s/it, loss=0.765]
Epoch 2/4:   9%|▊         | 108/1250 [01:58<24:18,  1.28s/it, loss=0.595]
Epoch 2/4:   9%|▊         | 109/1250 [01:58<25:36,  1.35s/it, loss=0.595]
Epoch 2/4:   9%|▊         | 109/1250 [01:59<25:36,  1.35s/it, loss=0.516]
Epoch 2/4:   9%|▉         | 110/1250 [01:59<23:25,  1.23s/it, loss=0.516]
Epoch 2/4:   9%|▉         | 110/1250 [02:02<23:25,  1.23s/it, loss=0.658]
Epoch 2/4:   9%|▉         | 111/1250 [02:02<33:53,  1.79s/it, loss=0.658]
Epoch 2/4:   9%|▉         | 111/1250 [02:03<33:53,  1.79s/it, loss=0.731]
Epoch 2/4:   9%|▉         | 112/1250 [02:03<29:31,  1.56s/it, loss=0.731]
Epoch 2/4:   9%|▉         | 112/1250 [02:04<29:31,  1.56s/it, loss=0.84] 
Epoch 2/4:   9%|▉         | 113/1250 [02:04<25:21,  1.34s/it, loss=0.84]
Epoch 2/4:   9%|▉         | 113/1250 [02:05<25:21,  1.34s/it, loss=0.888]
Epoch 2/4:   9%|▉         | 114/1250 [02:05<23:48,  1.26s/it, loss=0.888]
Epoch 2/4:   9%|▉         | 114/1250 [02:07<23:48,  1.26s/it, loss=0.669]
Epoch 2/4:   9%|▉         | 115/1250 [02:07<23:58,  1.27s/it, loss=0.669]
Epoch 2/4:   9%|▉         | 115/1250 [02:08<23:58,  1.27s/it, loss=1.28] 
Epoch 2/4:   9%|▉         | 116/1250 [02:08<24:43,  1.31s/it, loss=1.28]
Epoch 2/4:   9%|▉         | 116/1250 [02:10<24:43,  1.31s/it, loss=0.453]
Epoch 2/4:   9%|▉         | 117/1250 [02:10<26:06,  1.38s/it, loss=0.453]
Epoch 2/4:   9%|▉         | 117/1250 [02:11<26:06,  1.38s/it, loss=0.844]
Epoch 2/4:   9%|▉         | 118/1250 [02:11<23:56,  1.27s/it, loss=0.844]
Epoch 2/4:   9%|▉         | 118/1250 [02:12<23:56,  1.27s/it, loss=0.62] 
Epoch 2/4:  10%|▉         | 119/1250 [02:12<27:13,  1.44s/it, loss=0.62]
Epoch 2/4:  10%|▉         | 119/1250 [02:14<27:13,  1.44s/it, loss=0.532]
Epoch 2/4:  10%|▉         | 120/1250 [02:14<25:34,  1.36s/it, loss=0.532]
Epoch 2/4:  10%|▉         | 120/1250 [02:15<25:34,  1.36s/it, loss=0.737]
Epoch 2/4:  10%|▉         | 121/1250 [02:15<24:36,  1.31s/it, loss=0.737]
Epoch 2/4:  10%|▉         | 121/1250 [02:16<24:36,  1.31s/it, loss=0.834]
Epoch 2/4:  10%|▉         | 122/1250 [02:16<23:09,  1.23s/it, loss=0.834]
Epoch 2/4:  10%|▉         | 122/1250 [02:17<23:09,  1.23s/it, loss=0.517]
Epoch 2/4:  10%|▉         | 123/1250 [02:17<22:35,  1.20s/it, loss=0.517]
Epoch 2/4:  10%|▉         | 123/1250 [02:18<22:35,  1.20s/it, loss=0.547]
Epoch 2/4:  10%|▉         | 124/1250 [02:18<21:52,  1.17s/it, loss=0.547]
Epoch 2/4:  10%|▉         | 124/1250 [02:19<21:52,  1.17s/it, loss=0.525]
Epoch 2/4:  10%|█         | 125/1250 [02:19<21:02,  1.12s/it, loss=0.525]
Epoch 2/4:  10%|█         | 125/1250 [02:20<21:02,  1.12s/it, loss=0.711]
Epoch 2/4:  10%|█         | 126/1250 [02:20<20:14,  1.08s/it, loss=0.711]
Epoch 2/4:  10%|█         | 126/1250 [02:21<20:14,  1.08s/it, loss=0.563]
Epoch 2/4:  10%|█         | 127/1250 [02:21<19:47,  1.06s/it, loss=0.563]
Epoch 2/4:  10%|█         | 127/1250 [02:22<19:47,  1.06s/it, loss=0.799]
Epoch 2/4:  10%|█         | 128/1250 [02:22<21:26,  1.15s/it, loss=0.799]
Epoch 2/4:  10%|█         | 128/1250 [02:24<21:26,  1.15s/it, loss=0.794]
Epoch 2/4:  10%|█         | 129/1250 [02:24<22:50,  1.22s/it, loss=0.794]
Epoch 2/4:  10%|█         | 129/1250 [02:25<22:50,  1.22s/it, loss=0.549]
Epoch 2/4:  10%|█         | 130/1250 [02:25<23:04,  1.24s/it, loss=0.549]
Epoch 2/4:  10%|█         | 130/1250 [02:26<23:04,  1.24s/it, loss=0.524]
Epoch 2/4:  10%|█         | 131/1250 [02:26<22:42,  1.22s/it, loss=0.524]
Epoch 2/4:  10%|█         | 131/1250 [02:28<22:42,  1.22s/it, loss=0.686]
Epoch 2/4:  11%|█         | 132/1250 [02:28<23:40,  1.27s/it, loss=0.686]
Epoch 2/4:  11%|█         | 132/1250 [02:29<23:40,  1.27s/it, loss=0.775]
Epoch 2/4:  11%|█         | 133/1250 [02:29<24:42,  1.33s/it, loss=0.775]
Epoch 2/4:  11%|█         | 133/1250 [02:30<24:42,  1.33s/it, loss=0.672]
Epoch 2/4:  11%|█         | 134/1250 [02:30<24:28,  1.32s/it, loss=0.672]
Epoch 2/4:  11%|█         | 134/1250 [02:32<24:28,  1.32s/it, loss=0.609]
Epoch 2/4:  11%|█         | 135/1250 [02:32<23:59,  1.29s/it, loss=0.609]
Epoch 2/4:  11%|█         | 135/1250 [02:33<23:59,  1.29s/it, loss=0.607]
Epoch 2/4:  11%|█         | 136/1250 [02:33<23:52,  1.29s/it, loss=0.607]
Epoch 2/4:  11%|█         | 136/1250 [02:34<23:52,  1.29s/it, loss=0.531]
Epoch 2/4:  11%|█         | 137/1250 [02:34<24:19,  1.31s/it, loss=0.531]
Epoch 2/4:  11%|█         | 137/1250 [02:35<24:19,  1.31s/it, loss=0.458]
Epoch 2/4:  11%|█         | 138/1250 [02:35<23:20,  1.26s/it, loss=0.458]
Epoch 2/4:  11%|█         | 138/1250 [02:37<23:20,  1.26s/it, loss=0.334]
Epoch 2/4:  11%|█         | 139/1250 [02:37<23:18,  1.26s/it, loss=0.334]
Epoch 2/4:  11%|█         | 139/1250 [02:38<23:18,  1.26s/it, loss=0.528]
Epoch 2/4:  11%|█         | 140/1250 [02:38<23:53,  1.29s/it, loss=0.528]
Epoch 2/4:  11%|█         | 140/1250 [02:39<23:53,  1.29s/it, loss=0.688]
Epoch 2/4:  11%|█▏        | 141/1250 [02:39<23:17,  1.26s/it, loss=0.688]
Epoch 2/4:  11%|█▏        | 141/1250 [02:40<23:17,  1.26s/it, loss=0.595]
Epoch 2/4:  11%|█▏        | 142/1250 [02:40<22:47,  1.23s/it, loss=0.595]
Epoch 2/4:  11%|█▏        | 142/1250 [02:41<22:47,  1.23s/it, loss=0.854]
Epoch 2/4:  11%|█▏        | 143/1250 [02:41<21:42,  1.18s/it, loss=0.854]
Epoch 2/4:  11%|█▏        | 143/1250 [02:43<21:42,  1.18s/it, loss=0.662]
Epoch 2/4:  12%|█▏        | 144/1250 [02:43<21:43,  1.18s/it, loss=0.662]
Epoch 2/4:  12%|█▏        | 144/1250 [02:44<21:43,  1.18s/it, loss=0.534]
Epoch 2/4:  12%|█▏        | 145/1250 [02:44<21:54,  1.19s/it, loss=0.534]
Epoch 2/4:  12%|█▏        | 145/1250 [02:45<21:54,  1.19s/it, loss=0.877]
Epoch 2/4:  12%|█▏        | 146/1250 [02:45<21:42,  1.18s/it, loss=0.877]
Epoch 2/4:  12%|█▏        | 146/1250 [02:46<21:42,  1.18s/it, loss=0.465]
Epoch 2/4:  12%|█▏        | 147/1250 [02:46<22:17,  1.21s/it, loss=0.465]
Epoch 2/4:  12%|█▏        | 147/1250 [02:47<22:17,  1.21s/it, loss=0.372]
Epoch 2/4:  12%|█▏        | 148/1250 [02:47<22:26,  1.22s/it, loss=0.372]
Epoch 2/4:  12%|█▏        | 148/1250 [02:49<22:26,  1.22s/it, loss=0.94] 
Epoch 2/4:  12%|█▏        | 149/1250 [02:49<22:24,  1.22s/it, loss=0.94]
Epoch 2/4:  12%|█▏        | 149/1250 [02:50<22:24,  1.22s/it, loss=0.864]
Epoch 2/4:  12%|█▏        | 150/1250 [02:50<22:00,  1.20s/it, loss=0.864]
Epoch 2/4:  12%|█▏        | 150/1250 [02:51<22:00,  1.20s/it, loss=0.629]
Epoch 2/4:  12%|█▏        | 151/1250 [02:51<21:17,  1.16s/it, loss=0.629]
Epoch 2/4:  12%|█▏        | 151/1250 [02:52<21:17,  1.16s/it, loss=0.394]
Epoch 2/4:  12%|█▏        | 152/1250 [02:52<20:51,  1.14s/it, loss=0.394]
Epoch 2/4:  12%|█▏        | 152/1250 [02:53<20:51,  1.14s/it, loss=0.459]
Epoch 2/4:  12%|█▏        | 153/1250 [02:53<21:24,  1.17s/it, loss=0.459]
Epoch 2/4:  12%|█▏        | 153/1250 [02:54<21:24,  1.17s/it, loss=0.504]
Epoch 2/4:  12%|█▏        | 154/1250 [02:54<21:16,  1.16s/it, loss=0.504]
Epoch 2/4:  12%|█▏        | 154/1250 [02:55<21:16,  1.16s/it, loss=0.66] 
Epoch 2/4:  12%|█▏        | 155/1250 [02:55<20:41,  1.13s/it, loss=0.66]
Epoch 2/4:  12%|█▏        | 155/1250 [02:57<20:41,  1.13s/it, loss=0.49]
Epoch 2/4:  12%|█▏        | 156/1250 [02:57<20:22,  1.12s/it, loss=0.49]
Epoch 2/4:  12%|█▏        | 156/1250 [02:58<20:22,  1.12s/it, loss=0.906]
Epoch 2/4:  13%|█▎        | 157/1250 [02:58<20:36,  1.13s/it, loss=0.906]
Epoch 2/4:  13%|█▎        | 157/1250 [02:59<20:36,  1.13s/it, loss=0.617]
Epoch 2/4:  13%|█▎        | 158/1250 [02:59<20:12,  1.11s/it, loss=0.617]
Epoch 2/4:  13%|█▎        | 158/1250 [03:00<20:12,  1.11s/it, loss=0.814]
Epoch 2/4:  13%|█▎        | 159/1250 [03:00<19:52,  1.09s/it, loss=0.814]
Epoch 2/4:  13%|█▎        | 159/1250 [03:01<19:52,  1.09s/it, loss=0.828]
Epoch 2/4:  13%|█▎        | 160/1250 [03:01<18:27,  1.02s/it, loss=0.828]
Epoch 2/4:  13%|█▎        | 160/1250 [03:01<18:27,  1.02s/it, loss=1.05] 
Epoch 2/4:  13%|█▎        | 161/1250 [03:01<17:15,  1.05it/s, loss=1.05]
Epoch 2/4:  13%|█▎        | 161/1250 [03:02<17:15,  1.05it/s, loss=0.527]
Epoch 2/4:  13%|█▎        | 162/1250 [03:02<17:05,  1.06it/s, loss=0.527]
Epoch 2/4:  13%|█▎        | 162/1250 [03:03<17:05,  1.06it/s, loss=0.647]
Epoch 2/4:  13%|█▎        | 163/1250 [03:03<18:01,  1.01it/s, loss=0.647]
Epoch 2/4:  13%|█▎        | 163/1250 [03:04<18:01,  1.01it/s, loss=0.424]
Epoch 2/4:  13%|█▎        | 164/1250 [03:04<17:48,  1.02it/s, loss=0.424]
Epoch 2/4:  13%|█▎        | 164/1250 [03:05<17:48,  1.02it/s, loss=0.554]
Epoch 2/4:  13%|█▎        | 165/1250 [03:05<17:19,  1.04it/s, loss=0.554]
Epoch 2/4:  13%|█▎        | 165/1250 [03:06<17:19,  1.04it/s, loss=0.618]
Epoch 2/4:  13%|█▎        | 166/1250 [03:06<16:38,  1.09it/s, loss=0.618]
Epoch 2/4:  13%|█▎        | 166/1250 [03:07<16:38,  1.09it/s, loss=0.696]
Epoch 2/4:  13%|█▎        | 167/1250 [03:07<16:59,  1.06it/s, loss=0.696]
Epoch 2/4:  13%|█▎        | 167/1250 [03:08<16:59,  1.06it/s, loss=0.73] 
Epoch 2/4:  13%|█▎        | 168/1250 [03:08<17:44,  1.02it/s, loss=0.73]
Epoch 2/4:  13%|█▎        | 168/1250 [03:09<17:44,  1.02it/s, loss=0.64]
Epoch 2/4:  14%|█▎        | 169/1250 [03:09<17:14,  1.04it/s, loss=0.64]
Epoch 2/4:  14%|█▎        | 169/1250 [03:10<17:14,  1.04it/s, loss=0.723]
Epoch 2/4:  14%|█▎        | 170/1250 [03:10<17:05,  1.05it/s, loss=0.723]
Epoch 2/4:  14%|█▎        | 170/1250 [03:11<17:05,  1.05it/s, loss=0.691]
Epoch 2/4:  14%|█▎        | 171/1250 [03:11<17:55,  1.00it/s, loss=0.691]
Epoch 2/4:  14%|█▎        | 171/1250 [03:13<17:55,  1.00it/s, loss=0.631]
Epoch 2/4:  14%|█▍        | 172/1250 [03:13<19:43,  1.10s/it, loss=0.631]
Epoch 2/4:  14%|█▍        | 172/1250 [03:14<19:43,  1.10s/it, loss=0.703]
Epoch 2/4:  14%|█▍        | 173/1250 [03:14<20:41,  1.15s/it, loss=0.703]
Epoch 2/4:  14%|█▍        | 173/1250 [03:15<20:41,  1.15s/it, loss=0.59] 
Epoch 2/4:  14%|█▍        | 174/1250 [03:15<21:44,  1.21s/it, loss=0.59]
Epoch 2/4:  14%|█▍        | 174/1250 [03:17<21:44,  1.21s/it, loss=0.681]
Epoch 2/4:  14%|█▍        | 175/1250 [03:17<22:52,  1.28s/it, loss=0.681]
Epoch 2/4:  14%|█▍        | 175/1250 [03:19<22:52,  1.28s/it, loss=0.95] 
Epoch 2/4:  14%|█▍        | 176/1250 [03:19<27:39,  1.54s/it, loss=0.95]
Epoch 2/4:  14%|█▍        | 176/1250 [03:20<27:39,  1.54s/it, loss=0.441]
Epoch 2/4:  14%|█▍        | 177/1250 [03:20<27:41,  1.55s/it, loss=0.441]
Epoch 2/4:  14%|█▍        | 177/1250 [03:21<27:41,  1.55s/it, loss=0.655]
Epoch 2/4:  14%|█▍        | 178/1250 [03:21<25:31,  1.43s/it, loss=0.655]
Epoch 2/4:  14%|█▍        | 178/1250 [03:23<25:31,  1.43s/it, loss=0.65] 
Epoch 2/4:  14%|█▍        | 179/1250 [03:23<25:11,  1.41s/it, loss=0.65]
Epoch 2/4:  14%|█▍        | 179/1250 [03:25<25:11,  1.41s/it, loss=0.658]
Epoch 2/4:  14%|█▍        | 180/1250 [03:25<29:50,  1.67s/it, loss=0.658]
Epoch 2/4:  14%|█▍        | 180/1250 [03:27<29:50,  1.67s/it, loss=0.486]
Epoch 2/4:  14%|█▍        | 181/1250 [03:27<31:28,  1.77s/it, loss=0.486]
Epoch 2/4:  14%|█▍        | 181/1250 [03:30<31:28,  1.77s/it, loss=0.63] 
Epoch 2/4:  15%|█▍        | 182/1250 [03:30<38:40,  2.17s/it, loss=0.63]
Epoch 2/4:  15%|█▍        | 182/1250 [03:31<38:40,  2.17s/it, loss=0.739]
Epoch 2/4:  15%|█▍        | 183/1250 [03:31<32:12,  1.81s/it, loss=0.739]
Epoch 2/4:  15%|█▍        | 183/1250 [03:32<32:12,  1.81s/it, loss=0.449]
Epoch 2/4:  15%|█▍        | 184/1250 [03:32<28:23,  1.60s/it, loss=0.449]
Epoch 2/4:  15%|█▍        | 184/1250 [03:34<28:23,  1.60s/it, loss=0.69] 
Epoch 2/4:  15%|█▍        | 185/1250 [03:34<30:02,  1.69s/it, loss=0.69]
Epoch 2/4:  15%|█▍        | 185/1250 [03:35<30:02,  1.69s/it, loss=0.632]
Epoch 2/4:  15%|█▍        | 186/1250 [03:35<26:45,  1.51s/it, loss=0.632]
Epoch 2/4:  15%|█▍        | 186/1250 [03:36<26:45,  1.51s/it, loss=0.478]
Epoch 2/4:  15%|█▍        | 187/1250 [03:36<24:52,  1.40s/it, loss=0.478]
Epoch 2/4:  15%|█▍        | 187/1250 [03:38<24:52,  1.40s/it, loss=0.724]
Epoch 2/4:  15%|█▌        | 188/1250 [03:38<25:43,  1.45s/it, loss=0.724]
Epoch 2/4:  15%|█▌        | 188/1250 [03:39<25:43,  1.45s/it, loss=0.765]
Epoch 2/4:  15%|█▌        | 189/1250 [03:39<24:37,  1.39s/it, loss=0.765]
Epoch 2/4:  15%|█▌        | 189/1250 [03:40<24:37,  1.39s/it, loss=0.579]
Epoch 2/4:  15%|█▌        | 190/1250 [03:40<22:59,  1.30s/it, loss=0.579]
Epoch 2/4:  15%|█▌        | 190/1250 [03:42<22:59,  1.30s/it, loss=0.797]
Epoch 2/4:  15%|█▌        | 191/1250 [03:42<23:42,  1.34s/it, loss=0.797]
Epoch 2/4:  15%|█▌        | 191/1250 [03:43<23:42,  1.34s/it, loss=0.463]
Epoch 2/4:  15%|█▌        | 192/1250 [03:43<22:42,  1.29s/it, loss=0.463]
Epoch 2/4:  15%|█▌        | 192/1250 [03:44<22:42,  1.29s/it, loss=0.728]
Epoch 2/4:  15%|█▌        | 193/1250 [03:44<21:14,  1.21s/it, loss=0.728]
Epoch 2/4:  15%|█▌        | 193/1250 [03:45<21:14,  1.21s/it, loss=0.589]
Epoch 2/4:  16%|█▌        | 194/1250 [03:45<20:18,  1.15s/it, loss=0.589]
Epoch 2/4:  16%|█▌        | 194/1250 [03:46<20:18,  1.15s/it, loss=1.1]  
Epoch 2/4:  16%|█▌        | 195/1250 [03:46<20:33,  1.17s/it, loss=1.1]
Epoch 2/4:  16%|█▌        | 195/1250 [03:47<20:33,  1.17s/it, loss=0.771]
Epoch 2/4:  16%|█▌        | 196/1250 [03:47<20:28,  1.17s/it, loss=0.771]
Epoch 2/4:  16%|█▌        | 196/1250 [03:49<20:28,  1.17s/it, loss=0.552]
Epoch 2/4:  16%|█▌        | 197/1250 [03:49<20:34,  1.17s/it, loss=0.552]
Epoch 2/4:  16%|█▌        | 197/1250 [03:50<20:34,  1.17s/it, loss=0.633]
Epoch 2/4:  16%|█▌        | 198/1250 [03:50<20:28,  1.17s/it, loss=0.633]
Epoch 2/4:  16%|█▌        | 198/1250 [03:51<20:28,  1.17s/it, loss=0.581]
Epoch 2/4:  16%|█▌        | 199/1250 [03:51<20:03,  1.14s/it, loss=0.581]
Epoch 2/4:  16%|█▌        | 199/1250 [03:52<20:03,  1.14s/it, loss=0.515]
Epoch 2/4:  16%|█▌        | 200/1250 [03:52<19:56,  1.14s/it, loss=0.515]
Epoch 2/4:  16%|█▌        | 200/1250 [03:53<19:56,  1.14s/it, loss=0.909]
Epoch 2/4:  16%|█▌        | 201/1250 [03:53<21:23,  1.22s/it, loss=0.909]
Epoch 2/4:  16%|█▌        | 201/1250 [03:54<21:23,  1.22s/it, loss=0.602]
Epoch 2/4:  16%|█▌        | 202/1250 [03:54<20:54,  1.20s/it, loss=0.602]
Epoch 2/4:  16%|█▌        | 202/1250 [03:56<20:54,  1.20s/it, loss=0.495]
Epoch 2/4:  16%|█▌        | 203/1250 [03:56<22:46,  1.31s/it, loss=0.495]
Epoch 2/4:  16%|█▌        | 203/1250 [03:57<22:46,  1.31s/it, loss=0.724]
Epoch 2/4:  16%|█▋        | 204/1250 [03:57<22:28,  1.29s/it, loss=0.724]
Epoch 2/4:  16%|█▋        | 204/1250 [04:00<22:28,  1.29s/it, loss=0.447]
Epoch 2/4:  16%|█▋        | 205/1250 [04:00<30:37,  1.76s/it, loss=0.447]
Epoch 2/4:  16%|█▋        | 205/1250 [04:03<30:37,  1.76s/it, loss=0.615]
Epoch 2/4:  16%|█▋        | 206/1250 [04:03<35:48,  2.06s/it, loss=0.615]
Epoch 2/4:  16%|█▋        | 206/1250 [04:04<35:48,  2.06s/it, loss=0.666]
Epoch 2/4:  17%|█▋        | 207/1250 [04:04<29:18,  1.69s/it, loss=0.666]
Epoch 2/4:  17%|█▋        | 207/1250 [04:05<29:18,  1.69s/it, loss=0.647]
Epoch 2/4:  17%|█▋        | 208/1250 [04:05<25:31,  1.47s/it, loss=0.647]
Epoch 2/4:  17%|█▋        | 208/1250 [04:07<25:31,  1.47s/it, loss=0.927]
Epoch 2/4:  17%|█▋        | 209/1250 [04:07<28:14,  1.63s/it, loss=0.927]
Epoch 2/4:  17%|█▋        | 209/1250 [04:08<28:14,  1.63s/it, loss=0.679]
Epoch 2/4:  17%|█▋        | 210/1250 [04:08<25:15,  1.46s/it, loss=0.679]
Epoch 2/4:  17%|█▋        | 210/1250 [04:09<25:15,  1.46s/it, loss=1.17] 
Epoch 2/4:  17%|█▋        | 211/1250 [04:09<22:52,  1.32s/it, loss=1.17]
Epoch 2/4:  17%|█▋        | 211/1250 [04:10<22:52,  1.32s/it, loss=0.447]
Epoch 2/4:  17%|█▋        | 212/1250 [04:10<21:28,  1.24s/it, loss=0.447]
Epoch 2/4:  17%|█▋        | 212/1250 [04:11<21:28,  1.24s/it, loss=0.528]
Epoch 2/4:  17%|█▋        | 213/1250 [04:11<20:51,  1.21s/it, loss=0.528]
Epoch 2/4:  17%|█▋        | 213/1250 [04:12<20:51,  1.21s/it, loss=0.803]
Epoch 2/4:  17%|█▋        | 214/1250 [04:12<20:27,  1.18s/it, loss=0.803]
Epoch 2/4:  17%|█▋        | 214/1250 [04:13<20:27,  1.18s/it, loss=0.667]
Epoch 2/4:  17%|█▋        | 215/1250 [04:13<19:57,  1.16s/it, loss=0.667]
Epoch 2/4:  17%|█▋        | 215/1250 [04:14<19:57,  1.16s/it, loss=0.703]
Epoch 2/4:  17%|█▋        | 216/1250 [04:14<19:42,  1.14s/it, loss=0.703]
Epoch 2/4:  17%|█▋        | 216/1250 [04:16<19:42,  1.14s/it, loss=0.598]
Epoch 2/4:  17%|█▋        | 217/1250 [04:16<25:08,  1.46s/it, loss=0.598]
Epoch 2/4:  17%|█▋        | 217/1250 [04:17<25:08,  1.46s/it, loss=0.759]
Epoch 2/4:  17%|█▋        | 218/1250 [04:17<22:05,  1.28s/it, loss=0.759]
Epoch 2/4:  17%|█▋        | 218/1250 [04:19<22:05,  1.28s/it, loss=0.614]
Epoch 2/4:  18%|█▊        | 219/1250 [04:19<21:32,  1.25s/it, loss=0.614]
Epoch 2/4:  18%|█▊        | 219/1250 [04:20<21:32,  1.25s/it, loss=0.69] 
Epoch 2/4:  18%|█▊        | 220/1250 [04:20<22:42,  1.32s/it, loss=0.69]
Epoch 2/4:  18%|█▊        | 220/1250 [04:21<22:42,  1.32s/it, loss=0.676]
Epoch 2/4:  18%|█▊        | 221/1250 [04:21<22:47,  1.33s/it, loss=0.676]
Epoch 2/4:  18%|█▊        | 221/1250 [04:23<22:47,  1.33s/it, loss=0.797]
Epoch 2/4:  18%|█▊        | 222/1250 [04:23<22:15,  1.30s/it, loss=0.797]
Epoch 2/4:  18%|█▊        | 222/1250 [04:24<22:15,  1.30s/it, loss=0.809]
Epoch 2/4:  18%|█▊        | 223/1250 [04:24<21:21,  1.25s/it, loss=0.809]
Epoch 2/4:  18%|█▊        | 223/1250 [04:25<21:21,  1.25s/it, loss=0.729]
Epoch 2/4:  18%|█▊        | 224/1250 [04:25<21:45,  1.27s/it, loss=0.729]
Epoch 2/4:  18%|█▊        | 224/1250 [04:27<21:45,  1.27s/it, loss=0.578]
Epoch 2/4:  18%|█▊        | 225/1250 [04:27<22:58,  1.34s/it, loss=0.578]
Epoch 2/4:  18%|█▊        | 225/1250 [04:28<22:58,  1.34s/it, loss=0.499]
Epoch 2/4:  18%|█▊        | 226/1250 [04:28<22:40,  1.33s/it, loss=0.499]
Epoch 2/4:  18%|█▊        | 226/1250 [04:29<22:40,  1.33s/it, loss=0.707]
Epoch 2/4:  18%|█▊        | 227/1250 [04:29<22:12,  1.30s/it, loss=0.707]
Epoch 2/4:  18%|█▊        | 227/1250 [04:31<22:12,  1.30s/it, loss=0.653]
Epoch 2/4:  18%|█▊        | 228/1250 [04:31<24:29,  1.44s/it, loss=0.653]
Epoch 2/4:  18%|█▊        | 228/1250 [04:32<24:29,  1.44s/it, loss=1.05] 
Epoch 2/4:  18%|█▊        | 229/1250 [04:32<23:34,  1.39s/it, loss=1.05]
Epoch 2/4:  18%|█▊        | 229/1250 [04:33<23:34,  1.39s/it, loss=0.498]
Epoch 2/4:  18%|█▊        | 230/1250 [04:33<22:58,  1.35s/it, loss=0.498]
Epoch 2/4:  18%|█▊        | 230/1250 [04:35<22:58,  1.35s/it, loss=0.599]
Epoch 2/4:  18%|█▊        | 231/1250 [04:35<22:33,  1.33s/it, loss=0.599]
Epoch 2/4:  18%|█▊        | 231/1250 [04:36<22:33,  1.33s/it, loss=0.564]
Epoch 2/4:  19%|█▊        | 232/1250 [04:36<23:11,  1.37s/it, loss=0.564]
Epoch 2/4:  19%|█▊        | 232/1250 [04:39<23:11,  1.37s/it, loss=0.587]
Epoch 2/4:  19%|█▊        | 233/1250 [04:39<29:32,  1.74s/it, loss=0.587]
Epoch 2/4:  19%|█▊        | 233/1250 [04:41<29:32,  1.74s/it, loss=1.04] 
Epoch 2/4:  19%|█▊        | 234/1250 [04:41<33:33,  1.98s/it, loss=1.04]
Epoch 2/4:  19%|█▊        | 234/1250 [04:43<33:33,  1.98s/it, loss=0.592]
Epoch 2/4:  19%|█▉        | 235/1250 [04:43<32:33,  1.92s/it, loss=0.592]
Epoch 2/4:  19%|█▉        | 235/1250 [04:45<32:33,  1.92s/it, loss=0.681]
Epoch 2/4:  19%|█▉        | 236/1250 [04:45<34:19,  2.03s/it, loss=0.681]
Epoch 2/4:  19%|█▉        | 236/1250 [04:47<34:19,  2.03s/it, loss=0.494]
Epoch 2/4:  19%|█▉        | 237/1250 [04:47<30:27,  1.80s/it, loss=0.494]
Epoch 2/4:  19%|█▉        | 237/1250 [04:48<30:27,  1.80s/it, loss=0.64] 
Epoch 2/4:  19%|█▉        | 238/1250 [04:48<28:55,  1.71s/it, loss=0.64]
Epoch 2/4:  19%|█▉        | 238/1250 [04:49<28:55,  1.71s/it, loss=0.762]
Epoch 2/4:  19%|█▉        | 239/1250 [04:49<27:08,  1.61s/it, loss=0.762]
Epoch 2/4:  19%|█▉        | 239/1250 [04:51<27:08,  1.61s/it, loss=0.552]
Epoch 2/4:  19%|█▉        | 240/1250 [04:51<25:30,  1.52s/it, loss=0.552]
Epoch 2/4:  19%|█▉        | 240/1250 [04:53<25:30,  1.52s/it, loss=0.706]
Epoch 2/4:  19%|█▉        | 241/1250 [04:53<28:09,  1.67s/it, loss=0.706]
Epoch 2/4:  19%|█▉        | 241/1250 [04:55<28:09,  1.67s/it, loss=0.447]
Epoch 2/4:  19%|█▉        | 242/1250 [04:55<30:56,  1.84s/it, loss=0.447]
Epoch 2/4:  19%|█▉        | 242/1250 [04:57<30:56,  1.84s/it, loss=0.843]
Epoch 2/4:  19%|█▉        | 243/1250 [04:57<31:24,  1.87s/it, loss=0.843]
Epoch 2/4:  19%|█▉        | 243/1250 [04:59<31:24,  1.87s/it, loss=0.725]
Epoch 2/4:  20%|█▉        | 244/1250 [04:59<29:50,  1.78s/it, loss=0.725]
Epoch 2/4:  20%|█▉        | 244/1250 [05:00<29:50,  1.78s/it, loss=0.883]
Epoch 2/4:  20%|█▉        | 245/1250 [05:00<28:48,  1.72s/it, loss=0.883]
Epoch 2/4:  20%|█▉        | 245/1250 [05:02<28:48,  1.72s/it, loss=0.782]
Epoch 2/4:  20%|█▉        | 246/1250 [05:02<29:48,  1.78s/it, loss=0.782]
Epoch 2/4:  20%|█▉        | 246/1250 [05:04<29:48,  1.78s/it, loss=0.787]
Epoch 2/4:  20%|█▉        | 247/1250 [05:04<29:43,  1.78s/it, loss=0.787]
Epoch 2/4:  20%|█▉        | 247/1250 [05:05<29:43,  1.78s/it, loss=1.55] 
Epoch 2/4:  20%|█▉        | 248/1250 [05:05<28:34,  1.71s/it, loss=1.55]
Epoch 2/4:  20%|█▉        | 248/1250 [05:07<28:34,  1.71s/it, loss=0.489]
Epoch 2/4:  20%|█▉        | 249/1250 [05:07<30:17,  1.82s/it, loss=0.489]
Epoch 2/4:  20%|█▉        | 249/1250 [05:09<30:17,  1.82s/it, loss=0.399]
Epoch 2/4:  20%|██        | 250/1250 [05:09<27:08,  1.63s/it, loss=0.399]
Epoch 2/4:  20%|██        | 250/1250 [05:10<27:08,  1.63s/it, loss=0.551]
Epoch 2/4:  20%|██        | 251/1250 [05:10<26:31,  1.59s/it, loss=0.551]
Epoch 2/4:  20%|██        | 251/1250 [05:12<26:31,  1.59s/it, loss=0.59] 
Epoch 2/4:  20%|██        | 252/1250 [05:12<28:16,  1.70s/it, loss=0.59]
Epoch 2/4:  20%|██        | 252/1250 [05:13<28:16,  1.70s/it, loss=0.761]
Epoch 2/4:  20%|██        | 253/1250 [05:13<26:47,  1.61s/it, loss=0.761]
Epoch 2/4:  20%|██        | 253/1250 [05:15<26:47,  1.61s/it, loss=0.586]
Epoch 2/4:  20%|██        | 254/1250 [05:15<25:29,  1.54s/it, loss=0.586]
Epoch 2/4:  20%|██        | 254/1250 [05:17<25:29,  1.54s/it, loss=0.731]
Epoch 2/4:  20%|██        | 255/1250 [05:17<29:12,  1.76s/it, loss=0.731]
Epoch 2/4:  20%|██        | 255/1250 [05:19<29:12,  1.76s/it, loss=0.349]
Epoch 2/4:  20%|██        | 256/1250 [05:19<30:51,  1.86s/it, loss=0.349]
Epoch 2/4:  20%|██        | 256/1250 [05:21<30:51,  1.86s/it, loss=0.813]
Epoch 2/4:  21%|██        | 257/1250 [05:21<32:49,  1.98s/it, loss=0.813]
Epoch 2/4:  21%|██        | 257/1250 [05:24<32:49,  1.98s/it, loss=1.01] 
Epoch 2/4:  21%|██        | 258/1250 [05:24<33:45,  2.04s/it, loss=1.01]
Epoch 2/4:  21%|██        | 258/1250 [05:26<33:45,  2.04s/it, loss=0.976]
Epoch 2/4:  21%|██        | 259/1250 [05:26<33:24,  2.02s/it, loss=0.976]
Epoch 2/4:  21%|██        | 259/1250 [05:28<33:24,  2.02s/it, loss=0.644]
Epoch 2/4:  21%|██        | 260/1250 [05:28<33:48,  2.05s/it, loss=0.644]
Epoch 2/4:  21%|██        | 260/1250 [05:29<33:48,  2.05s/it, loss=0.676]
Epoch 2/4:  21%|██        | 261/1250 [05:29<31:13,  1.89s/it, loss=0.676]
Epoch 2/4:  21%|██        | 261/1250 [05:31<31:13,  1.89s/it, loss=0.67] 
Epoch 2/4:  21%|██        | 262/1250 [05:31<29:08,  1.77s/it, loss=0.67]
Epoch 2/4:  21%|██        | 262/1250 [05:32<29:08,  1.77s/it, loss=0.62]
Epoch 2/4:  21%|██        | 263/1250 [05:32<27:39,  1.68s/it, loss=0.62]
Epoch 2/4:  21%|██        | 263/1250 [05:34<27:39,  1.68s/it, loss=0.72]
Epoch 2/4:  21%|██        | 264/1250 [05:34<26:22,  1.61s/it, loss=0.72]
Epoch 2/4:  21%|██        | 264/1250 [05:36<26:22,  1.61s/it, loss=0.948]
Epoch 2/4:  21%|██        | 265/1250 [05:36<28:38,  1.75s/it, loss=0.948]
Epoch 2/4:  21%|██        | 265/1250 [05:38<28:38,  1.75s/it, loss=0.711]
Epoch 2/4:  21%|██▏       | 266/1250 [05:38<29:36,  1.81s/it, loss=0.711]
Epoch 2/4:  21%|██▏       | 266/1250 [05:39<29:36,  1.81s/it, loss=0.723]
Epoch 2/4:  21%|██▏       | 267/1250 [05:39<26:13,  1.60s/it, loss=0.723]
Epoch 2/4:  21%|██▏       | 267/1250 [05:41<26:13,  1.60s/it, loss=0.983]
Epoch 2/4:  21%|██▏       | 268/1250 [05:41<30:01,  1.83s/it, loss=0.983]
Epoch 2/4:  21%|██▏       | 268/1250 [05:45<30:01,  1.83s/it, loss=0.693]
Epoch 2/4:  22%|██▏       | 269/1250 [05:45<38:43,  2.37s/it, loss=0.693]
Epoch 2/4:  22%|██▏       | 269/1250 [05:46<38:43,  2.37s/it, loss=1.04] 
Epoch 2/4:  22%|██▏       | 270/1250 [05:46<33:39,  2.06s/it, loss=1.04]
Epoch 2/4:  22%|██▏       | 270/1250 [05:48<33:39,  2.06s/it, loss=0.579]
Epoch 2/4:  22%|██▏       | 271/1250 [05:48<30:48,  1.89s/it, loss=0.579]
Epoch 2/4:  22%|██▏       | 271/1250 [05:49<30:48,  1.89s/it, loss=0.835]
Epoch 2/4:  22%|██▏       | 272/1250 [05:49<28:29,  1.75s/it, loss=0.835]
Epoch 2/4:  22%|██▏       | 272/1250 [05:52<28:29,  1.75s/it, loss=0.619]
Epoch 2/4:  22%|██▏       | 273/1250 [05:52<32:05,  1.97s/it, loss=0.619]
Epoch 2/4:  22%|██▏       | 273/1250 [05:54<32:05,  1.97s/it, loss=0.787]
Epoch 2/4:  22%|██▏       | 274/1250 [05:54<34:01,  2.09s/it, loss=0.787]
Epoch 2/4:  22%|██▏       | 274/1250 [05:56<34:01,  2.09s/it, loss=0.693]
Epoch 2/4:  22%|██▏       | 275/1250 [05:56<35:51,  2.21s/it, loss=0.693]
Epoch 2/4:  22%|██▏       | 275/1250 [05:58<35:51,  2.21s/it, loss=0.78] 
Epoch 2/4:  22%|██▏       | 276/1250 [05:58<34:05,  2.10s/it, loss=0.78]
Epoch 2/4:  22%|██▏       | 276/1250 [05:59<34:05,  2.10s/it, loss=0.58]
Epoch 2/4:  22%|██▏       | 277/1250 [05:59<28:28,  1.76s/it, loss=0.58]
Epoch 2/4:  22%|██▏       | 277/1250 [06:00<28:28,  1.76s/it, loss=0.926]
Epoch 2/4:  22%|██▏       | 278/1250 [06:00<25:45,  1.59s/it, loss=0.926]
Epoch 2/4:  22%|██▏       | 278/1250 [06:02<25:45,  1.59s/it, loss=0.784]
Epoch 2/4:  22%|██▏       | 279/1250 [06:02<24:36,  1.52s/it, loss=0.784]
Epoch 2/4:  22%|██▏       | 279/1250 [06:03<24:36,  1.52s/it, loss=0.593]
Epoch 2/4:  22%|██▏       | 280/1250 [06:03<23:11,  1.43s/it, loss=0.593]
Epoch 2/4:  22%|██▏       | 280/1250 [06:04<23:11,  1.43s/it, loss=0.688]
Epoch 2/4:  22%|██▏       | 281/1250 [06:04<21:58,  1.36s/it, loss=0.688]
Epoch 2/4:  22%|██▏       | 281/1250 [06:06<21:58,  1.36s/it, loss=0.681]
Epoch 2/4:  23%|██▎       | 282/1250 [06:06<23:50,  1.48s/it, loss=0.681]
Epoch 2/4:  23%|██▎       | 282/1250 [06:07<23:50,  1.48s/it, loss=0.89] 
Epoch 2/4:  23%|██▎       | 283/1250 [06:07<21:24,  1.33s/it, loss=0.89]
Epoch 2/4:  23%|██▎       | 283/1250 [06:08<21:24,  1.33s/it, loss=0.861]
Epoch 2/4:  23%|██▎       | 284/1250 [06:08<20:09,  1.25s/it, loss=0.861]
Epoch 2/4:  23%|██▎       | 284/1250 [06:10<20:09,  1.25s/it, loss=0.633]
Epoch 2/4:  23%|██▎       | 285/1250 [06:10<22:07,  1.38s/it, loss=0.633]
Epoch 2/4:  23%|██▎       | 285/1250 [06:11<22:07,  1.38s/it, loss=0.764]
Epoch 2/4:  23%|██▎       | 286/1250 [06:11<21:09,  1.32s/it, loss=0.764]
Epoch 2/4:  23%|██▎       | 286/1250 [06:12<21:09,  1.32s/it, loss=0.619]
Epoch 2/4:  23%|██▎       | 287/1250 [06:12<22:07,  1.38s/it, loss=0.619]
Epoch 2/4:  23%|██▎       | 287/1250 [06:14<22:07,  1.38s/it, loss=0.904]
Epoch 2/4:  23%|██▎       | 288/1250 [06:14<21:33,  1.34s/it, loss=0.904]
Epoch 2/4:  23%|██▎       | 288/1250 [06:15<21:33,  1.34s/it, loss=0.733]
Epoch 2/4:  23%|██▎       | 289/1250 [06:15<21:56,  1.37s/it, loss=0.733]
Epoch 2/4:  23%|██▎       | 289/1250 [06:16<21:56,  1.37s/it, loss=1.08] 
Epoch 2/4:  23%|██▎       | 290/1250 [06:16<21:16,  1.33s/it, loss=1.08]
Epoch 2/4:  23%|██▎       | 290/1250 [06:18<21:16,  1.33s/it, loss=0.536]
Epoch 2/4:  23%|██▎       | 291/1250 [06:18<22:02,  1.38s/it, loss=0.536]
Epoch 2/4:  23%|██▎       | 291/1250 [06:19<22:02,  1.38s/it, loss=0.621]
Epoch 2/4:  23%|██▎       | 292/1250 [06:19<21:41,  1.36s/it, loss=0.621]
Epoch 2/4:  23%|██▎       | 292/1250 [06:21<21:41,  1.36s/it, loss=0.513]
Epoch 2/4:  23%|██▎       | 293/1250 [06:21<22:32,  1.41s/it, loss=0.513]
Epoch 2/4:  23%|██▎       | 293/1250 [06:23<22:32,  1.41s/it, loss=0.669]
Epoch 2/4:  24%|██▎       | 294/1250 [06:23<27:26,  1.72s/it, loss=0.669]
Epoch 2/4:  24%|██▎       | 294/1250 [06:26<27:26,  1.72s/it, loss=0.775]
Epoch 2/4:  24%|██▎       | 295/1250 [06:26<33:40,  2.12s/it, loss=0.775]
Epoch 2/4:  24%|██▎       | 295/1250 [06:27<33:40,  2.12s/it, loss=0.747]
Epoch 2/4:  24%|██▎       | 296/1250 [06:27<27:54,  1.76s/it, loss=0.747]
Epoch 2/4:  24%|██▎       | 296/1250 [06:29<27:54,  1.76s/it, loss=0.924]
Epoch 2/4:  24%|██▍       | 297/1250 [06:29<29:56,  1.89s/it, loss=0.924]
Epoch 2/4:  24%|██▍       | 297/1250 [06:31<29:56,  1.89s/it, loss=0.52] 
Epoch 2/4:  24%|██▍       | 298/1250 [06:31<30:47,  1.94s/it, loss=0.52]
Epoch 2/4:  24%|██▍       | 298/1250 [06:33<30:47,  1.94s/it, loss=0.646]
Epoch 2/4:  24%|██▍       | 299/1250 [06:33<30:16,  1.91s/it, loss=0.646]
Epoch 2/4:  24%|██▍       | 299/1250 [06:34<30:16,  1.91s/it, loss=0.984]
Epoch 2/4:  24%|██▍       | 300/1250 [06:34<26:06,  1.65s/it, loss=0.984]
Epoch 2/4:  24%|██▍       | 300/1250 [06:39<26:06,  1.65s/it, loss=0.681]
Epoch 2/4:  24%|██▍       | 301/1250 [06:39<40:51,  2.58s/it, loss=0.681]
Epoch 2/4:  24%|██▍       | 301/1250 [06:40<40:51,  2.58s/it, loss=0.54] 
Epoch 2/4:  24%|██▍       | 302/1250 [06:40<32:32,  2.06s/it, loss=0.54]
Epoch 2/4:  24%|██▍       | 302/1250 [06:41<32:32,  2.06s/it, loss=0.88]
Epoch 2/4:  24%|██▍       | 303/1250 [06:41<27:20,  1.73s/it, loss=0.88]
Epoch 2/4:  24%|██▍       | 303/1250 [06:43<27:20,  1.73s/it, loss=0.617]
Epoch 2/4:  24%|██▍       | 304/1250 [06:43<28:51,  1.83s/it, loss=0.617]
Epoch 2/4:  24%|██▍       | 304/1250 [06:44<28:51,  1.83s/it, loss=0.746]
Epoch 2/4:  24%|██▍       | 305/1250 [06:44<25:14,  1.60s/it, loss=0.746]
Epoch 2/4:  24%|██▍       | 305/1250 [06:45<25:14,  1.60s/it, loss=0.593]
Epoch 2/4:  24%|██▍       | 306/1250 [06:45<22:46,  1.45s/it, loss=0.593]
Epoch 2/4:  24%|██▍       | 306/1250 [06:46<22:46,  1.45s/it, loss=0.599]
Epoch 2/4:  25%|██▍       | 307/1250 [06:46<22:37,  1.44s/it, loss=0.599]
Epoch 2/4:  25%|██▍       | 307/1250 [06:48<22:37,  1.44s/it, loss=0.592]
Epoch 2/4:  25%|██▍       | 308/1250 [06:48<22:26,  1.43s/it, loss=0.592]
Epoch 2/4:  25%|██▍       | 308/1250 [06:49<22:26,  1.43s/it, loss=0.481]
Epoch 2/4:  25%|██▍       | 309/1250 [06:49<21:44,  1.39s/it, loss=0.481]
Epoch 2/4:  25%|██▍       | 309/1250 [06:50<21:44,  1.39s/it, loss=0.401]
Epoch 2/4:  25%|██▍       | 310/1250 [06:50<21:23,  1.37s/it, loss=0.401]
Epoch 2/4:  25%|██▍       | 310/1250 [06:52<21:23,  1.37s/it, loss=1.08] 
Epoch 2/4:  25%|██▍       | 311/1250 [06:52<21:53,  1.40s/it, loss=1.08]
Epoch 2/4:  25%|██▍       | 311/1250 [06:53<21:53,  1.40s/it, loss=0.512]
Epoch 2/4:  25%|██▍       | 312/1250 [06:53<22:10,  1.42s/it, loss=0.512]
Epoch 2/4:  25%|██▍       | 312/1250 [06:55<22:10,  1.42s/it, loss=0.638]
Epoch 2/4:  25%|██▌       | 313/1250 [06:55<22:03,  1.41s/it, loss=0.638]
Epoch 2/4:  25%|██▌       | 313/1250 [06:56<22:03,  1.41s/it, loss=0.97] 
Epoch 2/4:  25%|██▌       | 314/1250 [06:56<21:17,  1.36s/it, loss=0.97]
Epoch 2/4:  25%|██▌       | 314/1250 [06:57<21:17,  1.36s/it, loss=0.636]
Epoch 2/4:  25%|██▌       | 315/1250 [06:57<21:17,  1.37s/it, loss=0.636]
Epoch 2/4:  25%|██▌       | 315/1250 [06:59<21:17,  1.37s/it, loss=0.646]
Epoch 2/4:  25%|██▌       | 316/1250 [06:59<21:44,  1.40s/it, loss=0.646]
Epoch 2/4:  25%|██▌       | 316/1250 [07:00<21:44,  1.40s/it, loss=0.681]
Epoch 2/4:  25%|██▌       | 317/1250 [07:00<21:20,  1.37s/it, loss=0.681]
Epoch 2/4:  25%|██▌       | 317/1250 [07:02<21:20,  1.37s/it, loss=0.669]
Epoch 2/4:  25%|██▌       | 318/1250 [07:02<21:52,  1.41s/it, loss=0.669]
Epoch 2/4:  25%|██▌       | 318/1250 [07:03<21:52,  1.41s/it, loss=0.788]
Epoch 2/4:  26%|██▌       | 319/1250 [07:03<22:40,  1.46s/it, loss=0.788]
Epoch 2/4:  26%|██▌       | 319/1250 [07:05<22:40,  1.46s/it, loss=0.792]
Epoch 2/4:  26%|██▌       | 320/1250 [07:05<22:07,  1.43s/it, loss=0.792]
Epoch 2/4:  26%|██▌       | 320/1250 [07:06<22:07,  1.43s/it, loss=0.724]
Epoch 2/4:  26%|██▌       | 321/1250 [07:06<20:58,  1.36s/it, loss=0.724]
Epoch 2/4:  26%|██▌       | 321/1250 [07:07<20:58,  1.36s/it, loss=0.65] 
Epoch 2/4:  26%|██▌       | 322/1250 [07:07<22:31,  1.46s/it, loss=0.65]
Epoch 2/4:  26%|██▌       | 322/1250 [07:09<22:31,  1.46s/it, loss=0.522]
Epoch 2/4:  26%|██▌       | 323/1250 [07:09<21:36,  1.40s/it, loss=0.522]
Epoch 2/4:  26%|██▌       | 323/1250 [07:10<21:36,  1.40s/it, loss=0.557]
Epoch 2/4:  26%|██▌       | 324/1250 [07:10<20:56,  1.36s/it, loss=0.557]
Epoch 2/4:  26%|██▌       | 324/1250 [07:11<20:56,  1.36s/it, loss=0.434]
Epoch 2/4:  26%|██▌       | 325/1250 [07:11<21:05,  1.37s/it, loss=0.434]
Epoch 2/4:  26%|██▌       | 325/1250 [07:13<21:05,  1.37s/it, loss=0.594]
Epoch 2/4:  26%|██▌       | 326/1250 [07:13<22:09,  1.44s/it, loss=0.594]
Epoch 2/4:  26%|██▌       | 326/1250 [07:14<22:09,  1.44s/it, loss=0.388]
Epoch 2/4:  26%|██▌       | 327/1250 [07:14<21:46,  1.42s/it, loss=0.388]
Epoch 2/4:  26%|██▌       | 327/1250 [07:16<21:46,  1.42s/it, loss=0.578]
Epoch 2/4:  26%|██▌       | 328/1250 [07:16<21:08,  1.38s/it, loss=0.578]
Epoch 2/4:  26%|██▌       | 328/1250 [07:17<21:08,  1.38s/it, loss=0.734]
Epoch 2/4:  26%|██▋       | 329/1250 [07:17<21:42,  1.41s/it, loss=0.734]
Epoch 2/4:  26%|██▋       | 329/1250 [07:19<21:42,  1.41s/it, loss=0.477]
Epoch 2/4:  26%|██▋       | 330/1250 [07:19<21:51,  1.43s/it, loss=0.477]
Epoch 2/4:  26%|██▋       | 330/1250 [07:20<21:51,  1.43s/it, loss=0.663]
Epoch 2/4:  26%|██▋       | 331/1250 [07:20<21:14,  1.39s/it, loss=0.663]
Epoch 2/4:  26%|██▋       | 331/1250 [07:21<21:14,  1.39s/it, loss=0.363]
Epoch 2/4:  27%|██▋       | 332/1250 [07:21<21:01,  1.37s/it, loss=0.363]
Epoch 2/4:  27%|██▋       | 332/1250 [07:23<21:01,  1.37s/it, loss=0.646]
Epoch 2/4:  27%|██▋       | 333/1250 [07:23<21:38,  1.42s/it, loss=0.646]
Epoch 2/4:  27%|██▋       | 333/1250 [07:24<21:38,  1.42s/it, loss=0.968]
Epoch 2/4:  27%|██▋       | 334/1250 [07:24<20:55,  1.37s/it, loss=0.968]
Epoch 2/4:  27%|██▋       | 334/1250 [07:25<20:55,  1.37s/it, loss=0.743]
Epoch 2/4:  27%|██▋       | 335/1250 [07:25<19:56,  1.31s/it, loss=0.743]
Epoch 2/4:  27%|██▋       | 335/1250 [07:26<19:56,  1.31s/it, loss=1.09] 
Epoch 2/4:  27%|██▋       | 336/1250 [07:26<19:28,  1.28s/it, loss=1.09]
Epoch 2/4:  27%|██▋       | 336/1250 [07:28<19:28,  1.28s/it, loss=0.822]
Epoch 2/4:  27%|██▋       | 337/1250 [07:28<20:18,  1.33s/it, loss=0.822]
Epoch 2/4:  27%|██▋       | 337/1250 [07:29<20:18,  1.33s/it, loss=0.692]
Epoch 2/4:  27%|██▋       | 338/1250 [07:29<19:37,  1.29s/it, loss=0.692]
Epoch 2/4:  27%|██▋       | 338/1250 [07:30<19:37,  1.29s/it, loss=0.461]
Epoch 2/4:  27%|██▋       | 339/1250 [07:30<18:41,  1.23s/it, loss=0.461]
Epoch 2/4:  27%|██▋       | 339/1250 [07:32<18:41,  1.23s/it, loss=0.607]
Epoch 2/4:  27%|██▋       | 340/1250 [07:32<19:40,  1.30s/it, loss=0.607]
Epoch 2/4:  27%|██▋       | 340/1250 [07:33<19:40,  1.30s/it, loss=0.799]
Epoch 2/4:  27%|██▋       | 341/1250 [07:33<21:26,  1.42s/it, loss=0.799]
Epoch 2/4:  27%|██▋       | 341/1250 [07:34<21:26,  1.42s/it, loss=0.717]
Epoch 2/4:  27%|██▋       | 342/1250 [07:34<20:21,  1.35s/it, loss=0.717]
Epoch 2/4:  27%|██▋       | 342/1250 [07:36<20:21,  1.35s/it, loss=0.841]
Epoch 2/4:  27%|██▋       | 343/1250 [07:36<20:50,  1.38s/it, loss=0.841]
Epoch 2/4:  27%|██▋       | 343/1250 [07:38<20:50,  1.38s/it, loss=0.809]
Epoch 2/4:  28%|██▊       | 344/1250 [07:38<23:48,  1.58s/it, loss=0.809]
Epoch 2/4:  28%|██▊       | 344/1250 [07:39<23:48,  1.58s/it, loss=0.662]
Epoch 2/4:  28%|██▊       | 345/1250 [07:39<22:42,  1.51s/it, loss=0.662]
Epoch 2/4:  28%|██▊       | 345/1250 [07:40<22:42,  1.51s/it, loss=0.536]
Epoch 2/4:  28%|██▊       | 346/1250 [07:40<21:29,  1.43s/it, loss=0.536]
Epoch 2/4:  28%|██▊       | 346/1250 [07:42<21:29,  1.43s/it, loss=0.601]
Epoch 2/4:  28%|██▊       | 347/1250 [07:42<21:16,  1.41s/it, loss=0.601]
Epoch 2/4:  28%|██▊       | 347/1250 [07:43<21:16,  1.41s/it, loss=0.938]
Epoch 2/4:  28%|██▊       | 348/1250 [07:43<21:38,  1.44s/it, loss=0.938]
Epoch 2/4:  28%|██▊       | 348/1250 [07:45<21:38,  1.44s/it, loss=0.941]
Epoch 2/4:  28%|██▊       | 349/1250 [07:45<21:09,  1.41s/it, loss=0.941]
Epoch 2/4:  28%|██▊       | 349/1250 [07:47<21:09,  1.41s/it, loss=1.02] 
Epoch 2/4:  28%|██▊       | 350/1250 [07:47<26:30,  1.77s/it, loss=1.02]
Epoch 2/4:  28%|██▊       | 350/1250 [07:49<26:30,  1.77s/it, loss=0.929]
Epoch 2/4:  28%|██▊       | 351/1250 [07:49<28:16,  1.89s/it, loss=0.929]
Epoch 2/4:  28%|██▊       | 351/1250 [07:52<28:16,  1.89s/it, loss=0.776]
Epoch 2/4:  28%|██▊       | 352/1250 [07:52<30:17,  2.02s/it, loss=0.776]
Epoch 2/4:  28%|██▊       | 352/1250 [07:54<30:17,  2.02s/it, loss=0.891]
Epoch 2/4:  28%|██▊       | 353/1250 [07:54<29:48,  1.99s/it, loss=0.891]
Epoch 2/4:  28%|██▊       | 353/1250 [07:56<29:48,  1.99s/it, loss=0.88] 
Epoch 2/4:  28%|██▊       | 354/1250 [07:56<30:00,  2.01s/it, loss=0.88]
Epoch 2/4:  28%|██▊       | 354/1250 [07:58<30:00,  2.01s/it, loss=0.679]
Epoch 2/4:  28%|██▊       | 355/1250 [07:58<30:07,  2.02s/it, loss=0.679]
Epoch 2/4:  28%|██▊       | 355/1250 [08:00<30:07,  2.02s/it, loss=0.854]
Epoch 2/4:  28%|██▊       | 356/1250 [08:00<30:21,  2.04s/it, loss=0.854]
Epoch 2/4:  28%|██▊       | 356/1250 [08:02<30:21,  2.04s/it, loss=0.501]
Epoch 2/4:  29%|██▊       | 357/1250 [08:02<32:35,  2.19s/it, loss=0.501]
Epoch 2/4:  29%|██▊       | 357/1250 [08:04<32:35,  2.19s/it, loss=0.672]
Epoch 2/4:  29%|██▊       | 358/1250 [08:04<29:41,  2.00s/it, loss=0.672]
Epoch 2/4:  29%|██▊       | 358/1250 [08:06<29:41,  2.00s/it, loss=0.753]
Epoch 2/4:  29%|██▊       | 359/1250 [08:06<30:47,  2.07s/it, loss=0.753]
Epoch 2/4:  29%|██▊       | 359/1250 [08:08<30:47,  2.07s/it, loss=0.577]
Epoch 2/4:  29%|██▉       | 360/1250 [08:08<29:11,  1.97s/it, loss=0.577]
Epoch 2/4:  29%|██▉       | 360/1250 [08:10<29:11,  1.97s/it, loss=0.489]
Epoch 2/4:  29%|██▉       | 361/1250 [08:10<28:23,  1.92s/it, loss=0.489]
Epoch 2/4:  29%|██▉       | 361/1250 [08:12<28:23,  1.92s/it, loss=0.888]
Epoch 2/4:  29%|██▉       | 362/1250 [08:12<27:33,  1.86s/it, loss=0.888]
Epoch 2/4:  29%|██▉       | 362/1250 [08:13<27:33,  1.86s/it, loss=0.615]
Epoch 2/4:  29%|██▉       | 363/1250 [08:13<26:52,  1.82s/it, loss=0.615]
Epoch 2/4:  29%|██▉       | 363/1250 [08:15<26:52,  1.82s/it, loss=0.592]
Epoch 2/4:  29%|██▉       | 364/1250 [08:15<26:28,  1.79s/it, loss=0.592]
Epoch 2/4:  29%|██▉       | 364/1250 [08:16<26:28,  1.79s/it, loss=0.778]
Epoch 2/4:  29%|██▉       | 365/1250 [08:16<25:01,  1.70s/it, loss=0.778]
Epoch 2/4:  29%|██▉       | 365/1250 [08:18<25:01,  1.70s/it, loss=0.841]
Epoch 2/4:  29%|██▉       | 366/1250 [08:18<23:56,  1.63s/it, loss=0.841]
Epoch 2/4:  29%|██▉       | 366/1250 [08:20<23:56,  1.63s/it, loss=0.892]
Epoch 2/4:  29%|██▉       | 367/1250 [08:20<25:26,  1.73s/it, loss=0.892]
Epoch 2/4:  29%|██▉       | 367/1250 [08:21<25:26,  1.73s/it, loss=0.939]
Epoch 2/4:  29%|██▉       | 368/1250 [08:21<24:26,  1.66s/it, loss=0.939]
Epoch 2/4:  29%|██▉       | 368/1250 [08:24<24:26,  1.66s/it, loss=0.758]
Epoch 2/4:  30%|██▉       | 369/1250 [08:24<27:16,  1.86s/it, loss=0.758]
Epoch 2/4:  30%|██▉       | 369/1250 [08:26<27:16,  1.86s/it, loss=0.691]
Epoch 2/4:  30%|██▉       | 370/1250 [08:26<28:28,  1.94s/it, loss=0.691]
Epoch 2/4:  30%|██▉       | 370/1250 [08:27<28:28,  1.94s/it, loss=0.576]
Epoch 2/4:  30%|██▉       | 371/1250 [08:27<26:26,  1.80s/it, loss=0.576]
Epoch 2/4:  30%|██▉       | 371/1250 [08:29<26:26,  1.80s/it, loss=0.785]
Epoch 2/4:  30%|██▉       | 372/1250 [08:29<25:44,  1.76s/it, loss=0.785]
Epoch 2/4:  30%|██▉       | 372/1250 [08:31<25:44,  1.76s/it, loss=0.903]
Epoch 2/4:  30%|██▉       | 373/1250 [08:31<26:12,  1.79s/it, loss=0.903]
Epoch 2/4:  30%|██▉       | 373/1250 [08:33<26:12,  1.79s/it, loss=0.452]
Epoch 2/4:  30%|██▉       | 374/1250 [08:33<26:42,  1.83s/it, loss=0.452]
Epoch 2/4:  30%|██▉       | 374/1250 [08:35<26:42,  1.83s/it, loss=0.599]
Epoch 2/4:  30%|███       | 375/1250 [08:35<26:31,  1.82s/it, loss=0.599]
Epoch 2/4:  30%|███       | 375/1250 [08:36<26:31,  1.82s/it, loss=0.592]
Epoch 2/4:  30%|███       | 376/1250 [08:36<25:49,  1.77s/it, loss=0.592]
Epoch 2/4:  30%|███       | 376/1250 [08:38<25:49,  1.77s/it, loss=0.598]
Epoch 2/4:  30%|███       | 377/1250 [08:38<25:09,  1.73s/it, loss=0.598]
Epoch 2/4:  30%|███       | 377/1250 [08:40<25:09,  1.73s/it, loss=0.804]
Epoch 2/4:  30%|███       | 378/1250 [08:40<26:07,  1.80s/it, loss=0.804]
Epoch 2/4:  30%|███       | 378/1250 [08:41<26:07,  1.80s/it, loss=0.579]
Epoch 2/4:  30%|███       | 379/1250 [08:41<25:17,  1.74s/it, loss=0.579]
Epoch 2/4:  30%|███       | 379/1250 [08:43<25:17,  1.74s/it, loss=0.637]
Epoch 2/4:  30%|███       | 380/1250 [08:43<26:13,  1.81s/it, loss=0.637]
Epoch 2/4:  30%|███       | 380/1250 [08:45<26:13,  1.81s/it, loss=0.503]
Epoch 2/4:  30%|███       | 381/1250 [08:45<25:29,  1.76s/it, loss=0.503]
Epoch 2/4:  30%|███       | 381/1250 [08:47<25:29,  1.76s/it, loss=0.735]
Epoch 2/4:  31%|███       | 382/1250 [08:47<24:53,  1.72s/it, loss=0.735]
Epoch 2/4:  31%|███       | 382/1250 [08:48<24:53,  1.72s/it, loss=0.728]
Epoch 2/4:  31%|███       | 383/1250 [08:48<24:13,  1.68s/it, loss=0.728]
Epoch 2/4:  31%|███       | 383/1250 [08:50<24:13,  1.68s/it, loss=0.672]
Epoch 2/4:  31%|███       | 384/1250 [08:50<23:50,  1.65s/it, loss=0.672]
Epoch 2/4:  31%|███       | 384/1250 [08:51<23:50,  1.65s/it, loss=0.49] 
Epoch 2/4:  31%|███       | 385/1250 [08:51<22:59,  1.60s/it, loss=0.49]
Epoch 2/4:  31%|███       | 385/1250 [08:53<22:59,  1.60s/it, loss=1.03]
Epoch 2/4:  31%|███       | 386/1250 [08:53<22:32,  1.57s/it, loss=1.03]
Epoch 2/4:  31%|███       | 386/1250 [08:55<22:32,  1.57s/it, loss=0.506]
Epoch 2/4:  31%|███       | 387/1250 [08:55<25:09,  1.75s/it, loss=0.506]
Epoch 2/4:  31%|███       | 387/1250 [08:56<25:09,  1.75s/it, loss=0.616]
Epoch 2/4:  31%|███       | 388/1250 [08:56<24:02,  1.67s/it, loss=0.616]
Epoch 2/4:  31%|███       | 388/1250 [08:58<24:02,  1.67s/it, loss=0.864]
Epoch 2/4:  31%|███       | 389/1250 [08:58<22:48,  1.59s/it, loss=0.864]
Epoch 2/4:  31%|███       | 389/1250 [09:00<22:48,  1.59s/it, loss=1.07] 
Epoch 2/4:  31%|███       | 390/1250 [09:00<23:51,  1.67s/it, loss=1.07]
Epoch 2/4:  31%|███       | 390/1250 [09:01<23:51,  1.67s/it, loss=0.642]
Epoch 2/4:  31%|███▏      | 391/1250 [09:01<23:04,  1.61s/it, loss=0.642]
Epoch 2/4:  31%|███▏      | 391/1250 [09:04<23:04,  1.61s/it, loss=0.776]
Epoch 2/4:  31%|███▏      | 392/1250 [09:04<28:48,  2.01s/it, loss=0.776]
Epoch 2/4:  31%|███▏      | 392/1250 [09:06<28:48,  2.01s/it, loss=0.542]
Epoch 2/4:  31%|███▏      | 393/1250 [09:06<28:30,  2.00s/it, loss=0.542]
Epoch 2/4:  31%|███▏      | 393/1250 [09:08<28:30,  2.00s/it, loss=0.777]
Epoch 2/4:  32%|███▏      | 394/1250 [09:08<26:31,  1.86s/it, loss=0.777]
Epoch 2/4:  32%|███▏      | 394/1250 [09:09<26:31,  1.86s/it, loss=0.433]
Epoch 2/4:  32%|███▏      | 395/1250 [09:09<26:08,  1.83s/it, loss=0.433]
Epoch 2/4:  32%|███▏      | 395/1250 [09:11<26:08,  1.83s/it, loss=0.667]
Epoch 2/4:  32%|███▏      | 396/1250 [09:11<26:20,  1.85s/it, loss=0.667]
Epoch 2/4:  32%|███▏      | 396/1250 [09:13<26:20,  1.85s/it, loss=0.534]
Epoch 2/4:  32%|███▏      | 397/1250 [09:13<26:18,  1.85s/it, loss=0.534]
Epoch 2/4:  32%|███▏      | 397/1250 [09:15<26:18,  1.85s/it, loss=0.609]
Epoch 2/4:  32%|███▏      | 398/1250 [09:15<25:37,  1.80s/it, loss=0.609]
Epoch 2/4:  32%|███▏      | 398/1250 [09:17<25:37,  1.80s/it, loss=0.729]
Epoch 2/4:  32%|███▏      | 399/1250 [09:17<25:19,  1.79s/it, loss=0.729]
Epoch 2/4:  32%|███▏      | 399/1250 [09:18<25:19,  1.79s/it, loss=0.738]
Epoch 2/4:  32%|███▏      | 400/1250 [09:18<25:33,  1.80s/it, loss=0.738]
Epoch 2/4:  32%|███▏      | 400/1250 [09:20<25:33,  1.80s/it, loss=1.31] 
Epoch 2/4:  32%|███▏      | 401/1250 [09:20<24:59,  1.77s/it, loss=1.31]
Epoch 2/4:  32%|███▏      | 401/1250 [09:22<24:59,  1.77s/it, loss=0.611]
Epoch 2/4:  32%|███▏      | 402/1250 [09:22<24:38,  1.74s/it, loss=0.611]
Epoch 2/4:  32%|███▏      | 402/1250 [09:23<24:38,  1.74s/it, loss=0.615]
Epoch 2/4:  32%|███▏      | 403/1250 [09:23<23:52,  1.69s/it, loss=0.615]
Epoch 2/4:  32%|███▏      | 403/1250 [09:25<23:52,  1.69s/it, loss=0.7]  
Epoch 2/4:  32%|███▏      | 404/1250 [09:25<25:04,  1.78s/it, loss=0.7]
Epoch 2/4:  32%|███▏      | 404/1250 [09:27<25:04,  1.78s/it, loss=0.988]
Epoch 2/4:  32%|███▏      | 405/1250 [09:27<26:32,  1.88s/it, loss=0.988]
Epoch 2/4:  32%|███▏      | 405/1250 [09:30<26:32,  1.88s/it, loss=0.495]
Epoch 2/4:  32%|███▏      | 406/1250 [09:30<27:52,  1.98s/it, loss=0.495]
Epoch 2/4:  32%|███▏      | 406/1250 [09:32<27:52,  1.98s/it, loss=0.859]
Epoch 2/4:  33%|███▎      | 407/1250 [09:32<28:08,  2.00s/it, loss=0.859]
Epoch 2/4:  33%|███▎      | 407/1250 [09:34<28:08,  2.00s/it, loss=1.01] 
Epoch 2/4:  33%|███▎      | 408/1250 [09:34<29:25,  2.10s/it, loss=1.01]
Epoch 2/4:  33%|███▎      | 408/1250 [09:36<29:25,  2.10s/it, loss=1.07]
Epoch 2/4:  33%|███▎      | 409/1250 [09:36<30:45,  2.19s/it, loss=1.07]
Epoch 2/4:  33%|███▎      | 409/1250 [09:39<30:45,  2.19s/it, loss=0.552]
Epoch 2/4:  33%|███▎      | 410/1250 [09:39<30:52,  2.21s/it, loss=0.552]
Epoch 2/4:  33%|███▎      | 410/1250 [09:41<30:52,  2.21s/it, loss=0.47] 
Epoch 2/4:  33%|███▎      | 411/1250 [09:41<30:09,  2.16s/it, loss=0.47]
Epoch 2/4:  33%|███▎      | 411/1250 [09:42<30:09,  2.16s/it, loss=0.603]
Epoch 2/4:  33%|███▎      | 412/1250 [09:42<27:59,  2.00s/it, loss=0.603]
Epoch 2/4:  33%|███▎      | 412/1250 [09:44<27:59,  2.00s/it, loss=0.698]
Epoch 2/4:  33%|███▎      | 413/1250 [09:44<28:18,  2.03s/it, loss=0.698]
Epoch 2/4:  33%|███▎      | 413/1250 [09:46<28:18,  2.03s/it, loss=0.597]
Epoch 2/4:  33%|███▎      | 414/1250 [09:46<26:02,  1.87s/it, loss=0.597]
Epoch 2/4:  33%|███▎      | 414/1250 [09:48<26:02,  1.87s/it, loss=0.937]
Epoch 2/4:  33%|███▎      | 415/1250 [09:48<25:04,  1.80s/it, loss=0.937]
Epoch 2/4:  33%|███▎      | 415/1250 [09:49<25:04,  1.80s/it, loss=0.791]
Epoch 2/4:  33%|███▎      | 416/1250 [09:49<24:17,  1.75s/it, loss=0.791]
Epoch 2/4:  33%|███▎      | 416/1250 [09:51<24:17,  1.75s/it, loss=0.72] 
Epoch 2/4:  33%|███▎      | 417/1250 [09:51<23:28,  1.69s/it, loss=0.72]
Epoch 2/4:  33%|███▎      | 417/1250 [09:53<23:28,  1.69s/it, loss=0.753]
Epoch 2/4:  33%|███▎      | 418/1250 [09:53<24:59,  1.80s/it, loss=0.753]
Epoch 2/4:  33%|███▎      | 418/1250 [09:55<24:59,  1.80s/it, loss=0.932]
Epoch 2/4:  34%|███▎      | 419/1250 [09:55<24:34,  1.77s/it, loss=0.932]
Epoch 2/4:  34%|███▎      | 419/1250 [09:56<24:34,  1.77s/it, loss=0.553]
Epoch 2/4:  34%|███▎      | 420/1250 [09:56<23:45,  1.72s/it, loss=0.553]
Epoch 2/4:  34%|███▎      | 420/1250 [09:58<23:45,  1.72s/it, loss=0.735]
Epoch 2/4:  34%|███▎      | 421/1250 [09:58<24:52,  1.80s/it, loss=0.735]
Epoch 2/4:  34%|███▎      | 421/1250 [10:00<24:52,  1.80s/it, loss=0.788]
Epoch 2/4:  34%|███▍      | 422/1250 [10:00<23:27,  1.70s/it, loss=0.788]
Epoch 2/4:  34%|███▍      | 422/1250 [10:01<23:27,  1.70s/it, loss=0.574]
Epoch 2/4:  34%|███▍      | 423/1250 [10:01<22:54,  1.66s/it, loss=0.574]
Epoch 2/4:  34%|███▍      | 423/1250 [10:03<22:54,  1.66s/it, loss=0.399]
Epoch 2/4:  34%|███▍      | 424/1250 [10:03<23:51,  1.73s/it, loss=0.399]
Epoch 2/4:  34%|███▍      | 424/1250 [10:05<23:51,  1.73s/it, loss=0.688]
Epoch 2/4:  34%|███▍      | 425/1250 [10:05<23:15,  1.69s/it, loss=0.688]
Epoch 2/4:  34%|███▍      | 425/1250 [10:06<23:15,  1.69s/it, loss=0.619]
Epoch 2/4:  34%|███▍      | 426/1250 [10:06<23:39,  1.72s/it, loss=0.619]
Epoch 2/4:  34%|███▍      | 426/1250 [10:08<23:39,  1.72s/it, loss=0.901]
Epoch 2/4:  34%|███▍      | 427/1250 [10:08<24:40,  1.80s/it, loss=0.901]
Epoch 2/4:  34%|███▍      | 427/1250 [10:10<24:40,  1.80s/it, loss=0.754]
Epoch 2/4:  34%|███▍      | 428/1250 [10:10<23:25,  1.71s/it, loss=0.754]
Epoch 2/4:  34%|███▍      | 428/1250 [10:12<23:25,  1.71s/it, loss=0.573]
Epoch 2/4:  34%|███▍      | 429/1250 [10:12<23:29,  1.72s/it, loss=0.573]
Epoch 2/4:  34%|███▍      | 429/1250 [10:14<23:29,  1.72s/it, loss=0.818]
Epoch 2/4:  34%|███▍      | 430/1250 [10:14<24:11,  1.77s/it, loss=0.818]
Epoch 2/4:  34%|███▍      | 430/1250 [10:15<24:11,  1.77s/it, loss=0.549]
Epoch 2/4:  34%|███▍      | 431/1250 [10:15<22:34,  1.65s/it, loss=0.549]
Epoch 2/4:  34%|███▍      | 431/1250 [10:17<22:34,  1.65s/it, loss=0.93] 
Epoch 2/4:  35%|███▍      | 432/1250 [10:17<22:07,  1.62s/it, loss=0.93]
Epoch 2/4:  35%|███▍      | 432/1250 [10:18<22:07,  1.62s/it, loss=0.591]
Epoch 2/4:  35%|███▍      | 433/1250 [10:18<23:28,  1.72s/it, loss=0.591]
Epoch 2/4:  35%|███▍      | 433/1250 [10:20<23:28,  1.72s/it, loss=0.659]
Epoch 2/4:  35%|███▍      | 434/1250 [10:20<22:35,  1.66s/it, loss=0.659]
Epoch 2/4:  35%|███▍      | 434/1250 [10:22<22:35,  1.66s/it, loss=0.652]
Epoch 2/4:  35%|███▍      | 435/1250 [10:22<22:08,  1.63s/it, loss=0.652]
Epoch 2/4:  35%|███▍      | 435/1250 [10:23<22:08,  1.63s/it, loss=0.539]
Epoch 2/4:  35%|███▍      | 436/1250 [10:23<22:59,  1.70s/it, loss=0.539]
Epoch 2/4:  35%|███▍      | 436/1250 [10:25<22:59,  1.70s/it, loss=0.913]
Epoch 2/4:  35%|███▍      | 437/1250 [10:25<22:25,  1.65s/it, loss=0.913]
Epoch 2/4:  35%|███▍      | 437/1250 [10:26<22:25,  1.65s/it, loss=0.494]
Epoch 2/4:  35%|███▌      | 438/1250 [10:26<21:41,  1.60s/it, loss=0.494]
Epoch 2/4:  35%|███▌      | 438/1250 [10:28<21:41,  1.60s/it, loss=1.34] 
Epoch 2/4:  35%|███▌      | 439/1250 [10:28<22:57,  1.70s/it, loss=1.34]
Epoch 2/4:  35%|███▌      | 439/1250 [10:30<22:57,  1.70s/it, loss=0.529]
Epoch 2/4:  35%|███▌      | 440/1250 [10:30<22:48,  1.69s/it, loss=0.529]
Epoch 2/4:  35%|███▌      | 440/1250 [10:32<22:48,  1.69s/it, loss=0.997]
Epoch 2/4:  35%|███▌      | 441/1250 [10:32<24:08,  1.79s/it, loss=0.997]
Epoch 2/4:  35%|███▌      | 441/1250 [10:34<24:08,  1.79s/it, loss=0.742]
Epoch 2/4:  35%|███▌      | 442/1250 [10:34<25:33,  1.90s/it, loss=0.742]
Epoch 2/4:  35%|███▌      | 442/1250 [10:36<25:33,  1.90s/it, loss=1.02] 
Epoch 2/4:  35%|███▌      | 443/1250 [10:36<25:25,  1.89s/it, loss=1.02]
Epoch 2/4:  35%|███▌      | 443/1250 [10:38<25:25,  1.89s/it, loss=0.458]
Epoch 2/4:  36%|███▌      | 444/1250 [10:38<26:12,  1.95s/it, loss=0.458]
Epoch 2/4:  36%|███▌      | 444/1250 [10:40<26:12,  1.95s/it, loss=1.08] 
Epoch 2/4:  36%|███▌      | 445/1250 [10:40<26:07,  1.95s/it, loss=1.08]
Epoch 2/4:  36%|███▌      | 445/1250 [10:42<26:07,  1.95s/it, loss=0.759]
Epoch 2/4:  36%|███▌      | 446/1250 [10:42<25:39,  1.91s/it, loss=0.759]
Epoch 2/4:  36%|███▌      | 446/1250 [10:44<25:39,  1.91s/it, loss=0.871]
Epoch 2/4:  36%|███▌      | 447/1250 [10:44<24:55,  1.86s/it, loss=0.871]
Epoch 2/4:  36%|███▌      | 447/1250 [10:45<24:55,  1.86s/it, loss=0.504]
Epoch 2/4:  36%|███▌      | 448/1250 [10:45<23:13,  1.74s/it, loss=0.504]
Epoch 2/4:  36%|███▌      | 448/1250 [10:47<23:13,  1.74s/it, loss=0.571]
Epoch 2/4:  36%|███▌      | 449/1250 [10:47<23:09,  1.73s/it, loss=0.571]
Epoch 2/4:  36%|███▌      | 449/1250 [10:48<23:09,  1.73s/it, loss=0.642]
Epoch 2/4:  36%|███▌      | 450/1250 [10:48<22:05,  1.66s/it, loss=0.642]
Epoch 2/4:  36%|███▌      | 450/1250 [10:50<22:05,  1.66s/it, loss=0.652]
Epoch 2/4:  36%|███▌      | 451/1250 [10:50<21:57,  1.65s/it, loss=0.652]
Epoch 2/4:  36%|███▌      | 451/1250 [10:52<21:57,  1.65s/it, loss=0.452]
Epoch 2/4:  36%|███▌      | 452/1250 [10:52<21:46,  1.64s/it, loss=0.452]
Epoch 2/4:  36%|███▌      | 452/1250 [10:53<21:46,  1.64s/it, loss=0.682]
Epoch 2/4:  36%|███▌      | 453/1250 [10:53<22:04,  1.66s/it, loss=0.682]
Epoch 2/4:  36%|███▌      | 453/1250 [10:55<22:04,  1.66s/it, loss=0.746]
Epoch 2/4:  36%|███▋      | 454/1250 [10:55<21:38,  1.63s/it, loss=0.746]
Epoch 2/4:  36%|███▋      | 454/1250 [10:56<21:38,  1.63s/it, loss=0.713]
Epoch 2/4:  36%|███▋      | 455/1250 [10:56<21:15,  1.60s/it, loss=0.713]
Epoch 2/4:  36%|███▋      | 455/1250 [10:58<21:15,  1.60s/it, loss=0.599]
Epoch 2/4:  36%|███▋      | 456/1250 [10:58<22:09,  1.67s/it, loss=0.599]
Epoch 2/4:  36%|███▋      | 456/1250 [11:00<22:09,  1.67s/it, loss=0.596]
Epoch 2/4:  37%|███▋      | 457/1250 [11:00<22:09,  1.68s/it, loss=0.596]
Epoch 2/4:  37%|███▋      | 457/1250 [11:01<22:09,  1.68s/it, loss=0.668]
Epoch 2/4:  37%|███▋      | 458/1250 [11:01<21:20,  1.62s/it, loss=0.668]
Epoch 2/4:  37%|███▋      | 458/1250 [11:03<21:20,  1.62s/it, loss=0.571]
Epoch 2/4:  37%|███▋      | 459/1250 [11:03<22:11,  1.68s/it, loss=0.571]
Epoch 2/4:  37%|███▋      | 459/1250 [11:05<22:11,  1.68s/it, loss=1.04] 
Epoch 2/4:  37%|███▋      | 460/1250 [11:05<23:01,  1.75s/it, loss=1.04]
Epoch 2/4:  37%|███▋      | 460/1250 [11:07<23:01,  1.75s/it, loss=0.57]
Epoch 2/4:  37%|███▋      | 461/1250 [11:07<23:00,  1.75s/it, loss=0.57]
Epoch 2/4:  37%|███▋      | 461/1250 [11:09<23:00,  1.75s/it, loss=0.776]
Epoch 2/4:  37%|███▋      | 462/1250 [11:09<24:08,  1.84s/it, loss=0.776]
Epoch 2/4:  37%|███▋      | 462/1250 [11:10<24:08,  1.84s/it, loss=0.878]
Epoch 2/4:  37%|███▋      | 463/1250 [11:10<23:02,  1.76s/it, loss=0.878]
Epoch 2/4:  37%|███▋      | 463/1250 [11:12<23:02,  1.76s/it, loss=0.581]
Epoch 2/4:  37%|███▋      | 464/1250 [11:12<23:39,  1.81s/it, loss=0.581]
Epoch 2/4:  37%|███▋      | 464/1250 [11:14<23:39,  1.81s/it, loss=0.613]
Epoch 2/4:  37%|███▋      | 465/1250 [11:14<22:40,  1.73s/it, loss=0.613]
Epoch 2/4:  37%|███▋      | 465/1250 [11:16<22:40,  1.73s/it, loss=0.458]
Epoch 2/4:  37%|███▋      | 466/1250 [11:16<22:23,  1.71s/it, loss=0.458]
Epoch 2/4:  37%|███▋      | 466/1250 [11:18<22:23,  1.71s/it, loss=0.718]
Epoch 2/4:  37%|███▋      | 467/1250 [11:18<24:01,  1.84s/it, loss=0.718]
Epoch 2/4:  37%|███▋      | 467/1250 [11:20<24:01,  1.84s/it, loss=0.945]
Epoch 2/4:  37%|███▋      | 468/1250 [11:20<23:46,  1.82s/it, loss=0.945]
Epoch 2/4:  37%|███▋      | 468/1250 [11:21<23:46,  1.82s/it, loss=0.872]
Epoch 2/4:  38%|███▊      | 469/1250 [11:21<23:38,  1.82s/it, loss=0.872]
Epoch 2/4:  38%|███▊      | 469/1250 [11:23<23:38,  1.82s/it, loss=0.849]
Epoch 2/4:  38%|███▊      | 470/1250 [11:23<22:40,  1.74s/it, loss=0.849]
Epoch 2/4:  38%|███▊      | 470/1250 [11:24<22:40,  1.74s/it, loss=0.728]
Epoch 2/4:  38%|███▊      | 471/1250 [11:24<21:53,  1.69s/it, loss=0.728]
Epoch 2/4:  38%|███▊      | 471/1250 [11:26<21:53,  1.69s/it, loss=0.403]
Epoch 2/4:  38%|███▊      | 472/1250 [11:26<21:10,  1.63s/it, loss=0.403]
Epoch 2/4:  38%|███▊      | 472/1250 [11:28<21:10,  1.63s/it, loss=0.644]
Epoch 2/4:  38%|███▊      | 473/1250 [11:28<21:08,  1.63s/it, loss=0.644]
Epoch 2/4:  38%|███▊      | 473/1250 [11:30<21:08,  1.63s/it, loss=0.589]
Epoch 2/4:  38%|███▊      | 474/1250 [11:30<22:52,  1.77s/it, loss=0.589]
Epoch 2/4:  38%|███▊      | 474/1250 [11:31<22:52,  1.77s/it, loss=0.676]
Epoch 2/4:  38%|███▊      | 475/1250 [11:31<21:42,  1.68s/it, loss=0.676]
Epoch 2/4:  38%|███▊      | 475/1250 [11:33<21:42,  1.68s/it, loss=0.588]
Epoch 2/4:  38%|███▊      | 476/1250 [11:33<21:28,  1.66s/it, loss=0.588]
Epoch 2/4:  38%|███▊      | 476/1250 [11:35<21:28,  1.66s/it, loss=0.725]
Epoch 2/4:  38%|███▊      | 477/1250 [11:35<22:52,  1.78s/it, loss=0.725]
Epoch 2/4:  38%|███▊      | 477/1250 [11:36<22:52,  1.78s/it, loss=0.703]
Epoch 2/4:  38%|███▊      | 478/1250 [11:36<22:06,  1.72s/it, loss=0.703]
Epoch 2/4:  38%|███▊      | 478/1250 [11:38<22:06,  1.72s/it, loss=0.906]
Epoch 2/4:  38%|███▊      | 479/1250 [11:38<23:09,  1.80s/it, loss=0.906]
Epoch 2/4:  38%|███▊      | 479/1250 [11:41<23:09,  1.80s/it, loss=0.763]
Epoch 2/4:  38%|███▊      | 480/1250 [11:41<24:16,  1.89s/it, loss=0.763]
Epoch 2/4:  38%|███▊      | 480/1250 [11:42<24:16,  1.89s/it, loss=1.54] 
Epoch 2/4:  38%|███▊      | 481/1250 [11:42<23:39,  1.85s/it, loss=1.54]
Epoch 2/4:  38%|███▊      | 481/1250 [11:44<23:39,  1.85s/it, loss=0.721]
Epoch 2/4:  39%|███▊      | 482/1250 [11:44<24:22,  1.90s/it, loss=0.721]
Epoch 2/4:  39%|███▊      | 482/1250 [11:46<24:22,  1.90s/it, loss=0.791]
Epoch 2/4:  39%|███▊      | 483/1250 [11:46<23:00,  1.80s/it, loss=0.791]
Epoch 2/4:  39%|███▊      | 483/1250 [11:47<23:00,  1.80s/it, loss=0.743]
Epoch 2/4:  39%|███▊      | 484/1250 [11:47<21:48,  1.71s/it, loss=0.743]
Epoch 2/4:  39%|███▊      | 484/1250 [11:49<21:48,  1.71s/it, loss=0.551]
Epoch 2/4:  39%|███▉      | 485/1250 [11:49<21:36,  1.69s/it, loss=0.551]
Epoch 2/4:  39%|███▉      | 485/1250 [11:51<21:36,  1.69s/it, loss=0.735]
Epoch 2/4:  39%|███▉      | 486/1250 [11:51<20:51,  1.64s/it, loss=0.735]
Epoch 2/4:  39%|███▉      | 486/1250 [11:52<20:51,  1.64s/it, loss=0.674]
Epoch 2/4:  39%|███▉      | 487/1250 [11:52<20:06,  1.58s/it, loss=0.674]
Epoch 2/4:  39%|███▉      | 487/1250 [11:54<20:06,  1.58s/it, loss=0.652]
Epoch 2/4:  39%|███▉      | 488/1250 [11:54<20:32,  1.62s/it, loss=0.652]
Epoch 2/4:  39%|███▉      | 488/1250 [11:55<20:32,  1.62s/it, loss=0.573]
Epoch 2/4:  39%|███▉      | 489/1250 [11:55<21:07,  1.67s/it, loss=0.573]
Epoch 2/4:  39%|███▉      | 489/1250 [11:57<21:07,  1.67s/it, loss=0.769]
Epoch 2/4:  39%|███▉      | 490/1250 [11:57<20:29,  1.62s/it, loss=0.769]
Epoch 2/4:  39%|███▉      | 490/1250 [11:59<20:29,  1.62s/it, loss=0.485]
Epoch 2/4:  39%|███▉      | 491/1250 [11:59<20:36,  1.63s/it, loss=0.485]
Epoch 2/4:  39%|███▉      | 491/1250 [12:00<20:36,  1.63s/it, loss=0.735]
Epoch 2/4:  39%|███▉      | 492/1250 [12:00<20:57,  1.66s/it, loss=0.735]
Epoch 2/4:  39%|███▉      | 492/1250 [12:02<20:57,  1.66s/it, loss=0.557]
Epoch 2/4:  39%|███▉      | 493/1250 [12:02<20:26,  1.62s/it, loss=0.557]
Epoch 2/4:  39%|███▉      | 493/1250 [12:04<20:26,  1.62s/it, loss=0.816]
Epoch 2/4:  40%|███▉      | 494/1250 [12:04<21:22,  1.70s/it, loss=0.816]
Epoch 2/4:  40%|███▉      | 494/1250 [12:06<21:22,  1.70s/it, loss=0.596]
Epoch 2/4:  40%|███▉      | 495/1250 [12:06<22:02,  1.75s/it, loss=0.596]
Epoch 2/4:  40%|███▉      | 495/1250 [12:07<22:02,  1.75s/it, loss=0.996]
Epoch 2/4:  40%|███▉      | 496/1250 [12:07<21:06,  1.68s/it, loss=0.996]
Epoch 2/4:  40%|███▉      | 496/1250 [12:09<21:06,  1.68s/it, loss=0.549]
Epoch 2/4:  40%|███▉      | 497/1250 [12:09<22:14,  1.77s/it, loss=0.549]
Epoch 2/4:  40%|███▉      | 497/1250 [12:11<22:14,  1.77s/it, loss=0.628]
Epoch 2/4:  40%|███▉      | 498/1250 [12:11<21:57,  1.75s/it, loss=0.628]
Epoch 2/4:  40%|███▉      | 498/1250 [12:12<21:57,  1.75s/it, loss=0.744]
Epoch 2/4:  40%|███▉      | 499/1250 [12:12<21:18,  1.70s/it, loss=0.744]
Epoch 2/4:  40%|███▉      | 499/1250 [12:14<21:18,  1.70s/it, loss=0.617]
Epoch 2/4:  40%|████      | 500/1250 [12:14<22:11,  1.78s/it, loss=0.617]
Epoch 2/4:  40%|████      | 500/1250 [12:16<22:11,  1.78s/it, loss=0.761]
Epoch 2/4:  40%|████      | 501/1250 [12:16<21:10,  1.70s/it, loss=0.761]
Epoch 2/4:  40%|████      | 501/1250 [12:18<21:10,  1.70s/it, loss=0.763]
Epoch 2/4:  40%|████      | 502/1250 [12:18<21:40,  1.74s/it, loss=0.763]
Epoch 2/4:  40%|████      | 502/1250 [12:20<21:40,  1.74s/it, loss=0.691]
Epoch 2/4:  40%|████      | 503/1250 [12:20<23:30,  1.89s/it, loss=0.691]
Epoch 2/4:  40%|████      | 503/1250 [12:22<23:30,  1.89s/it, loss=0.802]
Epoch 2/4:  40%|████      | 504/1250 [12:22<22:39,  1.82s/it, loss=0.802]
Epoch 2/4:  40%|████      | 504/1250 [12:24<22:39,  1.82s/it, loss=1.21] 
Epoch 2/4:  40%|████      | 505/1250 [12:24<22:53,  1.84s/it, loss=1.21]
Epoch 2/4:  40%|████      | 505/1250 [12:25<22:53,  1.84s/it, loss=0.692]
Epoch 2/4:  40%|████      | 506/1250 [12:25<21:41,  1.75s/it, loss=0.692]
Epoch 2/4:  40%|████      | 506/1250 [12:27<21:41,  1.75s/it, loss=0.768]
Epoch 2/4:  41%|████      | 507/1250 [12:27<21:19,  1.72s/it, loss=0.768]
Epoch 2/4:  41%|████      | 507/1250 [12:29<21:19,  1.72s/it, loss=0.684]
Epoch 2/4:  41%|████      | 508/1250 [12:29<21:33,  1.74s/it, loss=0.684]
Epoch 2/4:  41%|████      | 508/1250 [12:30<21:33,  1.74s/it, loss=0.611]
Epoch 2/4:  41%|████      | 509/1250 [12:30<22:05,  1.79s/it, loss=0.611]
Epoch 2/4:  41%|████      | 509/1250 [12:32<22:05,  1.79s/it, loss=0.55] 
Epoch 2/4:  41%|████      | 510/1250 [12:32<21:07,  1.71s/it, loss=0.55]
Epoch 2/4:  41%|████      | 510/1250 [12:34<21:07,  1.71s/it, loss=0.647]
Epoch 2/4:  41%|████      | 511/1250 [12:34<20:57,  1.70s/it, loss=0.647]
Epoch 2/4:  41%|████      | 511/1250 [12:36<20:57,  1.70s/it, loss=0.587]
Epoch 2/4:  41%|████      | 512/1250 [12:36<22:16,  1.81s/it, loss=0.587]
Epoch 2/4:  41%|████      | 512/1250 [12:37<22:16,  1.81s/it, loss=0.635]
Epoch 2/4:  41%|████      | 513/1250 [12:37<21:19,  1.74s/it, loss=0.635]
Epoch 2/4:  41%|████      | 513/1250 [12:39<21:19,  1.74s/it, loss=0.628]
Epoch 2/4:  41%|████      | 514/1250 [12:39<20:34,  1.68s/it, loss=0.628]
Epoch 2/4:  41%|████      | 514/1250 [12:41<20:34,  1.68s/it, loss=0.797]
Epoch 2/4:  41%|████      | 515/1250 [12:41<22:22,  1.83s/it, loss=0.797]
Epoch 2/4:  41%|████      | 515/1250 [12:43<22:22,  1.83s/it, loss=0.603]
Epoch 2/4:  41%|████▏     | 516/1250 [12:43<21:39,  1.77s/it, loss=0.603]
Epoch 2/4:  41%|████▏     | 516/1250 [12:45<21:39,  1.77s/it, loss=1.2]  
Epoch 2/4:  41%|████▏     | 517/1250 [12:45<22:36,  1.85s/it, loss=1.2]
Epoch 2/4:  41%|████▏     | 517/1250 [12:47<22:36,  1.85s/it, loss=0.527]
Epoch 2/4:  41%|████▏     | 518/1250 [12:47<23:04,  1.89s/it, loss=0.527]
Epoch 2/4:  41%|████▏     | 518/1250 [12:49<23:04,  1.89s/it, loss=0.727]
Epoch 2/4:  42%|████▏     | 519/1250 [12:49<23:20,  1.92s/it, loss=0.727]
Epoch 2/4:  42%|████▏     | 519/1250 [12:50<23:20,  1.92s/it, loss=0.597]
Epoch 2/4:  42%|████▏     | 520/1250 [12:50<21:39,  1.78s/it, loss=0.597]
Epoch 2/4:  42%|████▏     | 520/1250 [12:52<21:39,  1.78s/it, loss=0.77] 
Epoch 2/4:  42%|████▏     | 521/1250 [12:52<22:46,  1.87s/it, loss=0.77]
Epoch 2/4:  42%|████▏     | 521/1250 [12:54<22:46,  1.87s/it, loss=0.685]
Epoch 2/4:  42%|████▏     | 522/1250 [12:54<22:29,  1.85s/it, loss=0.685]
Epoch 2/4:  42%|████▏     | 522/1250 [12:55<22:29,  1.85s/it, loss=0.668]
Epoch 2/4:  42%|████▏     | 523/1250 [12:55<21:19,  1.76s/it, loss=0.668]
Epoch 2/4:  42%|████▏     | 523/1250 [12:57<21:19,  1.76s/it, loss=0.737]
Epoch 2/4:  42%|████▏     | 524/1250 [12:57<20:59,  1.73s/it, loss=0.737]
Epoch 2/4:  42%|████▏     | 524/1250 [12:59<20:59,  1.73s/it, loss=0.695]
Epoch 2/4:  42%|████▏     | 525/1250 [12:59<21:53,  1.81s/it, loss=0.695]
Epoch 2/4:  42%|████▏     | 525/1250 [13:01<21:53,  1.81s/it, loss=0.549]
Epoch 2/4:  42%|████▏     | 526/1250 [13:01<22:10,  1.84s/it, loss=0.549]
Epoch 2/4:  42%|████▏     | 526/1250 [13:03<22:10,  1.84s/it, loss=0.515]
Epoch 2/4:  42%|████▏     | 527/1250 [13:03<22:05,  1.83s/it, loss=0.515]
Epoch 2/4:  42%|████▏     | 527/1250 [13:04<22:05,  1.83s/it, loss=0.722]
Epoch 2/4:  42%|████▏     | 528/1250 [13:04<21:00,  1.75s/it, loss=0.722]
Epoch 2/4:  42%|████▏     | 528/1250 [13:06<21:00,  1.75s/it, loss=0.698]
Epoch 2/4:  42%|████▏     | 529/1250 [13:06<22:01,  1.83s/it, loss=0.698]
Epoch 2/4:  42%|████▏     | 529/1250 [13:08<22:01,  1.83s/it, loss=0.926]
Epoch 2/4:  42%|████▏     | 530/1250 [13:08<21:54,  1.83s/it, loss=0.926]
Epoch 2/4:  42%|████▏     | 530/1250 [13:10<21:54,  1.83s/it, loss=0.617]
Epoch 2/4:  42%|████▏     | 531/1250 [13:10<21:40,  1.81s/it, loss=0.617]
Epoch 2/4:  42%|████▏     | 531/1250 [13:12<21:40,  1.81s/it, loss=0.897]
Epoch 2/4:  43%|████▎     | 532/1250 [13:12<23:59,  2.00s/it, loss=0.897]
Epoch 2/4:  43%|████▎     | 532/1250 [13:14<23:59,  2.00s/it, loss=0.682]
Epoch 2/4:  43%|████▎     | 533/1250 [13:14<22:13,  1.86s/it, loss=0.682]
Epoch 2/4:  43%|████▎     | 533/1250 [13:16<22:13,  1.86s/it, loss=0.631]
Epoch 2/4:  43%|████▎     | 534/1250 [13:16<22:39,  1.90s/it, loss=0.631]
Epoch 2/4:  43%|████▎     | 534/1250 [13:18<22:39,  1.90s/it, loss=0.74] 
Epoch 2/4:  43%|████▎     | 535/1250 [13:18<23:28,  1.97s/it, loss=0.74]
Epoch 2/4:  43%|████▎     | 535/1250 [13:20<23:28,  1.97s/it, loss=0.958]
Epoch 2/4:  43%|████▎     | 536/1250 [13:20<23:46,  2.00s/it, loss=0.958]
Epoch 2/4:  43%|████▎     | 536/1250 [13:22<23:46,  2.00s/it, loss=1.14] 
Epoch 2/4:  43%|████▎     | 537/1250 [13:22<22:31,  1.90s/it, loss=1.14]
Epoch 2/4:  43%|████▎     | 537/1250 [13:24<22:31,  1.90s/it, loss=0.595]
Epoch 2/4:  43%|████▎     | 538/1250 [13:24<22:43,  1.92s/it, loss=0.595]
Epoch 2/4:  43%|████▎     | 538/1250 [13:25<22:43,  1.92s/it, loss=0.477]
Epoch 2/4:  43%|████▎     | 539/1250 [13:25<21:40,  1.83s/it, loss=0.477]
Epoch 2/4:  43%|████▎     | 539/1250 [13:27<21:40,  1.83s/it, loss=0.921]
Epoch 2/4:  43%|████▎     | 540/1250 [13:27<21:37,  1.83s/it, loss=0.921]
Epoch 2/4:  43%|████▎     | 540/1250 [13:29<21:37,  1.83s/it, loss=0.8]  
Epoch 2/4:  43%|████▎     | 541/1250 [13:29<22:44,  1.92s/it, loss=0.8]
Epoch 2/4:  43%|████▎     | 541/1250 [13:31<22:44,  1.92s/it, loss=0.581]
Epoch 2/4:  43%|████▎     | 542/1250 [13:31<22:12,  1.88s/it, loss=0.581]
Epoch 2/4:  43%|████▎     | 542/1250 [13:34<22:12,  1.88s/it, loss=0.513]
Epoch 2/4:  43%|████▎     | 543/1250 [13:34<23:46,  2.02s/it, loss=0.513]
Epoch 2/4:  43%|████▎     | 543/1250 [13:35<23:46,  2.02s/it, loss=0.69] 
Epoch 2/4:  44%|████▎     | 544/1250 [13:35<22:49,  1.94s/it, loss=0.69]
Epoch 2/4:  44%|████▎     | 544/1250 [13:37<22:49,  1.94s/it, loss=0.471]
Epoch 2/4:  44%|████▎     | 545/1250 [13:37<22:14,  1.89s/it, loss=0.471]
Epoch 2/4:  44%|████▎     | 545/1250 [13:39<22:14,  1.89s/it, loss=0.786]
Epoch 2/4:  44%|████▎     | 546/1250 [13:39<22:39,  1.93s/it, loss=0.786]
Epoch 2/4:  44%|████▎     | 546/1250 [13:41<22:39,  1.93s/it, loss=0.652]
Epoch 2/4:  44%|████▍     | 547/1250 [13:41<21:17,  1.82s/it, loss=0.652]
Epoch 2/4:  44%|████▍     | 547/1250 [13:43<21:17,  1.82s/it, loss=0.803]
Epoch 2/4:  44%|████▍     | 548/1250 [13:43<21:48,  1.86s/it, loss=0.803]
Epoch 2/4:  44%|████▍     | 548/1250 [13:44<21:48,  1.86s/it, loss=0.827]
Epoch 2/4:  44%|████▍     | 549/1250 [13:44<20:49,  1.78s/it, loss=0.827]
Epoch 2/4:  44%|████▍     | 549/1250 [13:46<20:49,  1.78s/it, loss=0.554]
Epoch 2/4:  44%|████▍     | 550/1250 [13:46<20:24,  1.75s/it, loss=0.554]
Epoch 2/4:  44%|████▍     | 550/1250 [13:48<20:24,  1.75s/it, loss=0.833]
Epoch 2/4:  44%|████▍     | 551/1250 [13:48<20:40,  1.77s/it, loss=0.833]
Epoch 2/4:  44%|████▍     | 551/1250 [13:50<20:40,  1.77s/it, loss=0.669]
Epoch 2/4:  44%|████▍     | 552/1250 [13:50<21:47,  1.87s/it, loss=0.669]
Epoch 2/4:  44%|████▍     | 552/1250 [13:52<21:47,  1.87s/it, loss=0.661]
Epoch 2/4:  44%|████▍     | 553/1250 [13:52<23:12,  2.00s/it, loss=0.661]
Epoch 2/4:  44%|████▍     | 553/1250 [13:54<23:12,  2.00s/it, loss=0.864]
Epoch 2/4:  44%|████▍     | 554/1250 [13:54<23:33,  2.03s/it, loss=0.864]
Epoch 2/4:  44%|████▍     | 554/1250 [13:56<23:33,  2.03s/it, loss=0.579]
Epoch 2/4:  44%|████▍     | 555/1250 [13:56<22:34,  1.95s/it, loss=0.579]
Epoch 2/4:  44%|████▍     | 555/1250 [13:58<22:34,  1.95s/it, loss=0.639]
Epoch 2/4:  44%|████▍     | 556/1250 [13:58<22:47,  1.97s/it, loss=0.639]
Epoch 2/4:  44%|████▍     | 556/1250 [14:00<22:47,  1.97s/it, loss=1.11] 
Epoch 2/4:  45%|████▍     | 557/1250 [14:00<21:56,  1.90s/it, loss=1.11]
Epoch 2/4:  45%|████▍     | 557/1250 [14:02<21:56,  1.90s/it, loss=0.712]
Epoch 2/4:  45%|████▍     | 558/1250 [14:02<22:26,  1.95s/it, loss=0.712]
Epoch 2/4:  45%|████▍     | 558/1250 [14:04<22:26,  1.95s/it, loss=0.873]
Epoch 2/4:  45%|████▍     | 559/1250 [14:04<24:13,  2.10s/it, loss=0.873]
Epoch 2/4:  45%|████▍     | 559/1250 [14:06<24:13,  2.10s/it, loss=0.725]
Epoch 2/4:  45%|████▍     | 560/1250 [14:06<21:44,  1.89s/it, loss=0.725]
Epoch 2/4:  45%|████▍     | 560/1250 [14:08<21:44,  1.89s/it, loss=0.555]
Epoch 2/4:  45%|████▍     | 561/1250 [14:08<22:41,  1.98s/it, loss=0.555]
Epoch 2/4:  45%|████▍     | 561/1250 [14:10<22:41,  1.98s/it, loss=0.637]
Epoch 2/4:  45%|████▍     | 562/1250 [14:10<22:45,  1.98s/it, loss=0.637]
Epoch 2/4:  45%|████▍     | 562/1250 [14:12<22:45,  1.98s/it, loss=0.768]
Epoch 2/4:  45%|████▌     | 563/1250 [14:12<22:07,  1.93s/it, loss=0.768]
Epoch 2/4:  45%|████▌     | 563/1250 [14:14<22:07,  1.93s/it, loss=0.563]
Epoch 2/4:  45%|████▌     | 564/1250 [14:14<21:50,  1.91s/it, loss=0.563]
Epoch 2/4:  45%|████▌     | 564/1250 [14:15<21:50,  1.91s/it, loss=0.605]
Epoch 2/4:  45%|████▌     | 565/1250 [14:15<21:35,  1.89s/it, loss=0.605]
Epoch 2/4:  45%|████▌     | 565/1250 [14:17<21:35,  1.89s/it, loss=0.503]
Epoch 2/4:  45%|████▌     | 566/1250 [14:17<20:47,  1.82s/it, loss=0.503]
Epoch 2/4:  45%|████▌     | 566/1250 [14:19<20:47,  1.82s/it, loss=0.668]
Epoch 2/4:  45%|████▌     | 567/1250 [14:19<22:07,  1.94s/it, loss=0.668]
Epoch 2/4:  45%|████▌     | 567/1250 [14:21<22:07,  1.94s/it, loss=0.917]
Epoch 2/4:  45%|████▌     | 568/1250 [14:21<22:20,  1.97s/it, loss=0.917]
Epoch 2/4:  45%|████▌     | 568/1250 [14:23<22:20,  1.97s/it, loss=0.464]
Epoch 2/4:  46%|████▌     | 569/1250 [14:23<22:29,  1.98s/it, loss=0.464]
Epoch 2/4:  46%|████▌     | 569/1250 [14:25<22:29,  1.98s/it, loss=0.709]
Epoch 2/4:  46%|████▌     | 570/1250 [14:25<21:47,  1.92s/it, loss=0.709]
Epoch 2/4:  46%|████▌     | 570/1250 [14:27<21:47,  1.92s/it, loss=0.621]
Epoch 2/4:  46%|████▌     | 571/1250 [14:27<21:47,  1.93s/it, loss=0.621]
Epoch 2/4:  46%|████▌     | 571/1250 [14:29<21:47,  1.93s/it, loss=0.618]
Epoch 2/4:  46%|████▌     | 572/1250 [14:29<21:09,  1.87s/it, loss=0.618]
Epoch 2/4:  46%|████▌     | 572/1250 [14:31<21:09,  1.87s/it, loss=0.554]
Epoch 2/4:  46%|████▌     | 573/1250 [14:31<22:03,  1.96s/it, loss=0.554]
Epoch 2/4:  46%|████▌     | 573/1250 [14:33<22:03,  1.96s/it, loss=0.385]
Epoch 2/4:  46%|████▌     | 574/1250 [14:33<21:20,  1.89s/it, loss=0.385]
Epoch 2/4:  46%|████▌     | 574/1250 [14:34<21:20,  1.89s/it, loss=0.989]
Epoch 2/4:  46%|████▌     | 575/1250 [14:34<20:52,  1.86s/it, loss=0.989]
Epoch 2/4:  46%|████▌     | 575/1250 [14:37<20:52,  1.86s/it, loss=0.711]
Epoch 2/4:  46%|████▌     | 576/1250 [14:37<21:59,  1.96s/it, loss=0.711]
Epoch 2/4:  46%|████▌     | 576/1250 [14:39<21:59,  1.96s/it, loss=0.83] 
Epoch 2/4:  46%|████▌     | 577/1250 [14:39<22:17,  1.99s/it, loss=0.83]
Epoch 2/4:  46%|████▌     | 577/1250 [14:41<22:17,  1.99s/it, loss=0.536]
Epoch 2/4:  46%|████▌     | 578/1250 [14:41<22:34,  2.02s/it, loss=0.536]
Epoch 2/4:  46%|████▌     | 578/1250 [14:42<22:34,  2.02s/it, loss=0.572]
Epoch 2/4:  46%|████▋     | 579/1250 [14:42<21:02,  1.88s/it, loss=0.572]
Epoch 2/4:  46%|████▋     | 579/1250 [14:44<21:02,  1.88s/it, loss=0.582]
Epoch 2/4:  46%|████▋     | 580/1250 [14:44<20:55,  1.87s/it, loss=0.582]
Epoch 2/4:  46%|████▋     | 580/1250 [14:46<20:55,  1.87s/it, loss=0.625]
Epoch 2/4:  46%|████▋     | 581/1250 [14:46<21:02,  1.89s/it, loss=0.625]
Epoch 2/4:  46%|████▋     | 581/1250 [14:48<21:02,  1.89s/it, loss=0.634]
Epoch 2/4:  47%|████▋     | 582/1250 [14:48<20:26,  1.84s/it, loss=0.634]
Epoch 2/4:  47%|████▋     | 582/1250 [14:50<20:26,  1.84s/it, loss=0.878]
Epoch 2/4:  47%|████▋     | 583/1250 [14:50<21:33,  1.94s/it, loss=0.878]
Epoch 2/4:  47%|████▋     | 583/1250 [14:52<21:33,  1.94s/it, loss=0.552]
Epoch 2/4:  47%|████▋     | 584/1250 [14:52<20:40,  1.86s/it, loss=0.552]
Epoch 2/4:  47%|████▋     | 584/1250 [14:53<20:40,  1.86s/it, loss=1.29] 
Epoch 2/4:  47%|████▋     | 585/1250 [14:53<20:29,  1.85s/it, loss=1.29]
Epoch 2/4:  47%|████▋     | 585/1250 [14:56<20:29,  1.85s/it, loss=0.737]
Epoch 2/4:  47%|████▋     | 586/1250 [14:56<21:14,  1.92s/it, loss=0.737]
Epoch 2/4:  47%|████▋     | 586/1250 [14:57<21:14,  1.92s/it, loss=0.776]
Epoch 2/4:  47%|████▋     | 587/1250 [14:57<20:34,  1.86s/it, loss=0.776]
Epoch 2/4:  47%|████▋     | 587/1250 [14:59<20:34,  1.86s/it, loss=0.737]
Epoch 2/4:  47%|████▋     | 588/1250 [14:59<20:30,  1.86s/it, loss=0.737]
Epoch 2/4:  47%|████▋     | 588/1250 [15:01<20:30,  1.86s/it, loss=0.655]
Epoch 2/4:  47%|████▋     | 589/1250 [15:01<21:17,  1.93s/it, loss=0.655]
Epoch 2/4:  47%|████▋     | 589/1250 [15:03<21:17,  1.93s/it, loss=0.441]
Epoch 2/4:  47%|████▋     | 590/1250 [15:03<21:37,  1.97s/it, loss=0.441]
Epoch 2/4:  47%|████▋     | 590/1250 [15:05<21:37,  1.97s/it, loss=0.751]
Epoch 2/4:  47%|████▋     | 591/1250 [15:05<21:31,  1.96s/it, loss=0.751]
Epoch 2/4:  47%|████▋     | 591/1250 [15:07<21:31,  1.96s/it, loss=0.67] 
Epoch 2/4:  47%|████▋     | 592/1250 [15:07<21:36,  1.97s/it, loss=0.67]
Epoch 2/4:  47%|████▋     | 592/1250 [15:09<21:36,  1.97s/it, loss=0.694]
Epoch 2/4:  47%|████▋     | 593/1250 [15:09<22:13,  2.03s/it, loss=0.694]
Epoch 2/4:  47%|████▋     | 593/1250 [15:11<22:13,  2.03s/it, loss=0.455]
Epoch 2/4:  48%|████▊     | 594/1250 [15:11<20:44,  1.90s/it, loss=0.455]
Epoch 2/4:  48%|████▊     | 594/1250 [15:13<20:44,  1.90s/it, loss=0.658]
Epoch 2/4:  48%|████▊     | 595/1250 [15:13<20:15,  1.86s/it, loss=0.658]
Epoch 2/4:  48%|████▊     | 595/1250 [15:15<20:15,  1.86s/it, loss=0.658]
Epoch 2/4:  48%|████▊     | 596/1250 [15:15<20:32,  1.88s/it, loss=0.658]
Epoch 2/4:  48%|████▊     | 596/1250 [15:17<20:32,  1.88s/it, loss=0.673]
Epoch 2/4:  48%|████▊     | 597/1250 [15:17<21:01,  1.93s/it, loss=0.673]
Epoch 2/4:  48%|████▊     | 597/1250 [15:19<21:01,  1.93s/it, loss=0.648]
Epoch 2/4:  48%|████▊     | 598/1250 [15:19<20:50,  1.92s/it, loss=0.648]
Epoch 2/4:  48%|████▊     | 598/1250 [15:21<20:50,  1.92s/it, loss=0.787]
Epoch 2/4:  48%|████▊     | 599/1250 [15:21<20:55,  1.93s/it, loss=0.787]
Epoch 2/4:  48%|████▊     | 599/1250 [15:22<20:55,  1.93s/it, loss=0.437]
Epoch 2/4:  48%|████▊     | 600/1250 [15:22<20:09,  1.86s/it, loss=0.437]
Epoch 2/4:  48%|████▊     | 600/1250 [15:24<20:09,  1.86s/it, loss=0.926]
Epoch 2/4:  48%|████▊     | 601/1250 [15:24<20:00,  1.85s/it, loss=0.926]
Epoch 2/4:  48%|████▊     | 601/1250 [15:26<20:00,  1.85s/it, loss=0.608]
Epoch 2/4:  48%|████▊     | 602/1250 [15:26<20:16,  1.88s/it, loss=0.608]
Epoch 2/4:  48%|████▊     | 602/1250 [15:28<20:16,  1.88s/it, loss=0.737]
Epoch 2/4:  48%|████▊     | 603/1250 [15:28<21:54,  2.03s/it, loss=0.737]
Epoch 2/4:  48%|████▊     | 603/1250 [15:30<21:54,  2.03s/it, loss=1.05] 
Epoch 2/4:  48%|████▊     | 604/1250 [15:30<19:29,  1.81s/it, loss=1.05]
Epoch 2/4:  48%|████▊     | 604/1250 [15:32<19:29,  1.81s/it, loss=0.556]
Epoch 2/4:  48%|████▊     | 605/1250 [15:32<20:20,  1.89s/it, loss=0.556]
Epoch 2/4:  48%|████▊     | 605/1250 [15:34<20:20,  1.89s/it, loss=0.714]
Epoch 2/4:  48%|████▊     | 606/1250 [15:34<19:45,  1.84s/it, loss=0.714]
Epoch 2/4:  48%|████▊     | 606/1250 [15:36<19:45,  1.84s/it, loss=0.517]
Epoch 2/4:  49%|████▊     | 607/1250 [15:36<20:14,  1.89s/it, loss=0.517]
Epoch 2/4:  49%|████▊     | 607/1250 [15:37<20:14,  1.89s/it, loss=0.638]
Epoch 2/4:  49%|████▊     | 608/1250 [15:37<20:15,  1.89s/it, loss=0.638]
Epoch 2/4:  49%|████▊     | 608/1250 [15:39<20:15,  1.89s/it, loss=0.62] 
Epoch 2/4:  49%|████▊     | 609/1250 [15:39<20:34,  1.93s/it, loss=0.62]
Epoch 2/4:  49%|████▊     | 609/1250 [15:42<20:34,  1.93s/it, loss=0.664]
Epoch 2/4:  49%|████▉     | 610/1250 [15:42<21:13,  1.99s/it, loss=0.664]
Epoch 2/4:  49%|████▉     | 610/1250 [15:44<21:13,  1.99s/it, loss=0.491]
Epoch 2/4:  49%|████▉     | 611/1250 [15:44<21:31,  2.02s/it, loss=0.491]
Epoch 2/4:  49%|████▉     | 611/1250 [15:46<21:31,  2.02s/it, loss=0.412]
Epoch 2/4:  49%|████▉     | 612/1250 [15:46<22:06,  2.08s/it, loss=0.412]
Epoch 2/4:  49%|████▉     | 612/1250 [15:47<22:06,  2.08s/it, loss=0.567]
Epoch 2/4:  49%|████▉     | 613/1250 [15:47<19:30,  1.84s/it, loss=0.567]
Epoch 2/4:  49%|████▉     | 613/1250 [15:49<19:30,  1.84s/it, loss=0.855]
Epoch 2/4:  49%|████▉     | 614/1250 [15:49<20:25,  1.93s/it, loss=0.855]
Epoch 2/4:  49%|████▉     | 614/1250 [15:51<20:25,  1.93s/it, loss=0.593]
Epoch 2/4:  49%|████▉     | 615/1250 [15:51<20:20,  1.92s/it, loss=0.593]
Epoch 2/4:  49%|████▉     | 615/1250 [15:53<20:20,  1.92s/it, loss=0.488]
Epoch 2/4:  49%|████▉     | 616/1250 [15:53<20:35,  1.95s/it, loss=0.488]
Epoch 2/4:  49%|████▉     | 616/1250 [15:55<20:35,  1.95s/it, loss=0.996]
Epoch 2/4:  49%|████▉     | 617/1250 [15:55<19:59,  1.89s/it, loss=0.996]
Epoch 2/4:  49%|████▉     | 617/1250 [15:57<19:59,  1.89s/it, loss=0.703]
Epoch 2/4:  49%|████▉     | 618/1250 [15:57<19:48,  1.88s/it, loss=0.703]
Epoch 2/4:  49%|████▉     | 618/1250 [15:59<19:48,  1.88s/it, loss=0.829]
Epoch 2/4:  50%|████▉     | 619/1250 [15:59<19:56,  1.90s/it, loss=0.829]
Epoch 2/4:  50%|████▉     | 619/1250 [16:01<19:56,  1.90s/it, loss=0.737]
Epoch 2/4:  50%|████▉     | 620/1250 [16:01<19:34,  1.86s/it, loss=0.737]
Epoch 2/4:  50%|████▉     | 620/1250 [16:03<19:34,  1.86s/it, loss=0.713]
Epoch 2/4:  50%|████▉     | 621/1250 [16:03<20:41,  1.97s/it, loss=0.713]
Epoch 2/4:  50%|████▉     | 621/1250 [16:05<20:41,  1.97s/it, loss=0.839]
Epoch 2/4:  50%|████▉     | 622/1250 [16:05<20:16,  1.94s/it, loss=0.839]
Epoch 2/4:  50%|████▉     | 622/1250 [16:07<20:16,  1.94s/it, loss=0.614]
Epoch 2/4:  50%|████▉     | 623/1250 [16:07<20:15,  1.94s/it, loss=0.614]
Epoch 2/4:  50%|████▉     | 623/1250 [16:09<20:15,  1.94s/it, loss=0.443]
Epoch 2/4:  50%|████▉     | 624/1250 [16:09<21:46,  2.09s/it, loss=0.443]
Epoch 2/4:  50%|████▉     | 624/1250 [16:11<21:46,  2.09s/it, loss=0.538]
Epoch 2/4:  50%|█████     | 625/1250 [16:11<20:50,  2.00s/it, loss=0.538]
Epoch 2/4:  50%|█████     | 625/1250 [16:13<20:50,  2.00s/it, loss=0.672]
Epoch 2/4:  50%|█████     | 626/1250 [16:13<20:21,  1.96s/it, loss=0.672]
Epoch 2/4:  50%|█████     | 626/1250 [16:15<20:21,  1.96s/it, loss=0.595]
Epoch 2/4:  50%|█████     | 627/1250 [16:15<20:42,  1.99s/it, loss=0.595]
Epoch 2/4:  50%|█████     | 627/1250 [16:17<20:42,  1.99s/it, loss=0.764]
Epoch 2/4:  50%|█████     | 628/1250 [16:17<19:52,  1.92s/it, loss=0.764]
Epoch 2/4:  50%|█████     | 628/1250 [16:19<19:52,  1.92s/it, loss=0.529]
Epoch 2/4:  50%|█████     | 629/1250 [16:19<21:19,  2.06s/it, loss=0.529]
Epoch 2/4:  50%|█████     | 629/1250 [16:21<21:19,  2.06s/it, loss=0.766]
Epoch 2/4:  50%|█████     | 630/1250 [16:21<20:17,  1.96s/it, loss=0.766]
Epoch 2/4:  50%|█████     | 630/1250 [16:23<20:17,  1.96s/it, loss=0.816]
Epoch 2/4:  50%|█████     | 631/1250 [16:23<20:26,  1.98s/it, loss=0.816]
Epoch 2/4:  50%|█████     | 631/1250 [16:25<20:26,  1.98s/it, loss=0.857]
Epoch 2/4:  51%|█████     | 632/1250 [16:25<20:24,  1.98s/it, loss=0.857]
Epoch 2/4:  51%|█████     | 632/1250 [16:27<20:24,  1.98s/it, loss=0.443]
Epoch 2/4:  51%|█████     | 633/1250 [16:27<20:37,  2.01s/it, loss=0.443]
Epoch 2/4:  51%|█████     | 633/1250 [16:29<20:37,  2.01s/it, loss=0.984]
Epoch 2/4:  51%|█████     | 634/1250 [16:29<21:18,  2.07s/it, loss=0.984]
Epoch 2/4:  51%|█████     | 634/1250 [16:31<21:18,  2.07s/it, loss=1.02] 
Epoch 2/4:  51%|█████     | 635/1250 [16:31<19:49,  1.93s/it, loss=1.02]
Epoch 2/4:  51%|█████     | 635/1250 [16:33<19:49,  1.93s/it, loss=0.78]
Epoch 2/4:  51%|█████     | 636/1250 [16:33<20:17,  1.98s/it, loss=0.78]
Epoch 2/4:  51%|█████     | 636/1250 [16:35<20:17,  1.98s/it, loss=0.969]
Epoch 2/4:  51%|█████     | 637/1250 [16:35<21:12,  2.08s/it, loss=0.969]
Epoch 2/4:  51%|█████     | 637/1250 [16:37<21:12,  2.08s/it, loss=0.769]
Epoch 2/4:  51%|█████     | 638/1250 [16:37<19:50,  1.95s/it, loss=0.769]
Epoch 2/4:  51%|█████     | 638/1250 [16:39<19:50,  1.95s/it, loss=0.559]
Epoch 2/4:  51%|█████     | 639/1250 [16:39<20:30,  2.01s/it, loss=0.559]
Epoch 2/4:  51%|█████     | 639/1250 [16:41<20:30,  2.01s/it, loss=0.599]
Epoch 2/4:  51%|█████     | 640/1250 [16:41<19:53,  1.96s/it, loss=0.599]
Epoch 2/4:  51%|█████     | 640/1250 [16:43<19:53,  1.96s/it, loss=1.11] 
Epoch 2/4:  51%|█████▏    | 641/1250 [16:43<20:20,  2.00s/it, loss=1.11]
Epoch 2/4:  51%|█████▏    | 641/1250 [16:45<20:20,  2.00s/it, loss=0.561]
Epoch 2/4:  51%|█████▏    | 642/1250 [16:45<20:33,  2.03s/it, loss=0.561]
Epoch 2/4:  51%|█████▏    | 642/1250 [16:47<20:33,  2.03s/it, loss=0.752]
Epoch 2/4:  51%|█████▏    | 643/1250 [16:47<20:22,  2.01s/it, loss=0.752]
Epoch 2/4:  51%|█████▏    | 643/1250 [16:49<20:22,  2.01s/it, loss=0.714]
Epoch 2/4:  52%|█████▏    | 644/1250 [16:49<19:32,  1.94s/it, loss=0.714]
Epoch 2/4:  52%|█████▏    | 644/1250 [16:50<19:32,  1.94s/it, loss=0.839]
Epoch 2/4:  52%|█████▏    | 645/1250 [16:50<19:18,  1.91s/it, loss=0.839]
Epoch 2/4:  52%|█████▏    | 645/1250 [16:53<19:18,  1.91s/it, loss=0.628]
Epoch 2/4:  52%|█████▏    | 646/1250 [16:53<20:26,  2.03s/it, loss=0.628]
Epoch 2/4:  52%|█████▏    | 646/1250 [16:55<20:26,  2.03s/it, loss=0.754]
Epoch 2/4:  52%|█████▏    | 647/1250 [16:55<21:42,  2.16s/it, loss=0.754]
Epoch 2/4:  52%|█████▏    | 647/1250 [16:57<21:42,  2.16s/it, loss=0.553]
Epoch 2/4:  52%|█████▏    | 648/1250 [16:57<19:28,  1.94s/it, loss=0.553]
Epoch 2/4:  52%|█████▏    | 648/1250 [16:59<19:28,  1.94s/it, loss=0.681]
Epoch 2/4:  52%|█████▏    | 649/1250 [16:59<19:47,  1.98s/it, loss=0.681]
Epoch 2/4:  52%|█████▏    | 649/1250 [17:01<19:47,  1.98s/it, loss=0.692]
Epoch 2/4:  52%|█████▏    | 650/1250 [17:01<19:48,  1.98s/it, loss=0.692]
Epoch 2/4:  52%|█████▏    | 650/1250 [17:03<19:48,  1.98s/it, loss=0.681]
Epoch 2/4:  52%|█████▏    | 651/1250 [17:03<19:33,  1.96s/it, loss=0.681]
Epoch 2/4:  52%|█████▏    | 651/1250 [17:05<19:33,  1.96s/it, loss=0.914]
Epoch 2/4:  52%|█████▏    | 652/1250 [17:05<20:19,  2.04s/it, loss=0.914]
Epoch 2/4:  52%|█████▏    | 652/1250 [17:07<20:19,  2.04s/it, loss=0.643]
Epoch 2/4:  52%|█████▏    | 653/1250 [17:07<19:57,  2.01s/it, loss=0.643]
Epoch 2/4:  52%|█████▏    | 653/1250 [17:08<19:57,  2.01s/it, loss=0.97] 
Epoch 2/4:  52%|█████▏    | 654/1250 [17:08<18:53,  1.90s/it, loss=0.97]
Epoch 2/4:  52%|█████▏    | 654/1250 [17:11<18:53,  1.90s/it, loss=0.572]
Epoch 2/4:  52%|█████▏    | 655/1250 [17:11<19:38,  1.98s/it, loss=0.572]
Epoch 2/4:  52%|█████▏    | 655/1250 [17:12<19:38,  1.98s/it, loss=0.562]
Epoch 2/4:  52%|█████▏    | 656/1250 [17:12<19:35,  1.98s/it, loss=0.562]
Epoch 2/4:  52%|█████▏    | 656/1250 [17:15<19:35,  1.98s/it, loss=0.946]
Epoch 2/4:  53%|█████▎    | 657/1250 [17:15<20:30,  2.08s/it, loss=0.946]
Epoch 2/4:  53%|█████▎    | 657/1250 [17:17<20:30,  2.08s/it, loss=0.775]
Epoch 2/4:  53%|█████▎    | 658/1250 [17:17<19:45,  2.00s/it, loss=0.775]
Epoch 2/4:  53%|█████▎    | 658/1250 [17:19<19:45,  2.00s/it, loss=0.588]
Epoch 2/4:  53%|█████▎    | 659/1250 [17:19<19:37,  1.99s/it, loss=0.588]
Epoch 2/4:  53%|█████▎    | 659/1250 [17:21<19:37,  1.99s/it, loss=0.897]
Epoch 2/4:  53%|█████▎    | 660/1250 [17:21<20:16,  2.06s/it, loss=0.897]
Epoch 2/4:  53%|█████▎    | 660/1250 [17:22<20:16,  2.06s/it, loss=0.659]
Epoch 2/4:  53%|█████▎    | 661/1250 [17:22<18:57,  1.93s/it, loss=0.659]
Epoch 2/4:  53%|█████▎    | 661/1250 [17:25<18:57,  1.93s/it, loss=0.738]
Epoch 2/4:  53%|█████▎    | 662/1250 [17:25<19:56,  2.03s/it, loss=0.738]
Epoch 2/4:  53%|█████▎    | 662/1250 [17:26<19:56,  2.03s/it, loss=0.913]
Epoch 2/4:  53%|█████▎    | 663/1250 [17:26<18:57,  1.94s/it, loss=0.913]
Epoch 2/4:  53%|█████▎    | 663/1250 [17:29<18:57,  1.94s/it, loss=0.775]
Epoch 2/4:  53%|█████▎    | 664/1250 [17:29<20:01,  2.05s/it, loss=0.775]
Epoch 2/4:  53%|█████▎    | 664/1250 [17:31<20:01,  2.05s/it, loss=0.578]
Epoch 2/4:  53%|█████▎    | 665/1250 [17:31<19:57,  2.05s/it, loss=0.578]
Epoch 2/4:  53%|█████▎    | 665/1250 [17:33<19:57,  2.05s/it, loss=0.663]
Epoch 2/4:  53%|█████▎    | 666/1250 [17:33<19:34,  2.01s/it, loss=0.663]
Epoch 2/4:  53%|█████▎    | 666/1250 [17:35<19:34,  2.01s/it, loss=0.778]
Epoch 2/4:  53%|█████▎    | 667/1250 [17:35<18:57,  1.95s/it, loss=0.778]
Epoch 2/4:  53%|█████▎    | 667/1250 [17:37<18:57,  1.95s/it, loss=0.803]
Epoch 2/4:  53%|█████▎    | 668/1250 [17:37<19:36,  2.02s/it, loss=0.803]
Epoch 2/4:  53%|█████▎    | 668/1250 [17:39<19:36,  2.02s/it, loss=0.888]
Epoch 2/4:  54%|█████▎    | 669/1250 [17:39<20:03,  2.07s/it, loss=0.888]
Epoch 2/4:  54%|█████▎    | 669/1250 [17:41<20:03,  2.07s/it, loss=0.885]
Epoch 2/4:  54%|█████▎    | 670/1250 [17:41<20:01,  2.07s/it, loss=0.885]
Epoch 2/4:  54%|█████▎    | 670/1250 [17:43<20:01,  2.07s/it, loss=0.58] 
Epoch 2/4:  54%|█████▎    | 671/1250 [17:43<18:51,  1.95s/it, loss=0.58]
Epoch 2/4:  54%|█████▎    | 671/1250 [17:45<18:51,  1.95s/it, loss=0.913]
Epoch 2/4:  54%|█████▍    | 672/1250 [17:45<19:02,  1.98s/it, loss=0.913]
Epoch 2/4:  54%|█████▍    | 672/1250 [17:46<19:02,  1.98s/it, loss=0.612]
Epoch 2/4:  54%|█████▍    | 673/1250 [17:46<18:23,  1.91s/it, loss=0.612]
Epoch 2/4:  54%|█████▍    | 673/1250 [17:49<18:23,  1.91s/it, loss=0.908]
Epoch 2/4:  54%|█████▍    | 674/1250 [17:49<19:11,  2.00s/it, loss=0.908]
Epoch 2/4:  54%|█████▍    | 674/1250 [17:50<19:11,  2.00s/it, loss=0.583]
Epoch 2/4:  54%|█████▍    | 675/1250 [17:50<18:43,  1.95s/it, loss=0.583]
Epoch 2/4:  54%|█████▍    | 675/1250 [17:53<18:43,  1.95s/it, loss=0.73] 
Epoch 2/4:  54%|█████▍    | 676/1250 [17:53<19:03,  1.99s/it, loss=0.73]
Epoch 2/4:  54%|█████▍    | 676/1250 [17:55<19:03,  1.99s/it, loss=0.637]
Epoch 2/4:  54%|█████▍    | 677/1250 [17:55<19:41,  2.06s/it, loss=0.637]
Epoch 2/4:  54%|█████▍    | 677/1250 [17:57<19:41,  2.06s/it, loss=0.628]
Epoch 2/4:  54%|█████▍    | 678/1250 [17:57<19:07,  2.01s/it, loss=0.628]
Epoch 2/4:  54%|█████▍    | 678/1250 [17:59<19:07,  2.01s/it, loss=0.492]
Epoch 2/4:  54%|█████▍    | 679/1250 [17:59<19:06,  2.01s/it, loss=0.492]
Epoch 2/4:  54%|█████▍    | 679/1250 [18:00<19:06,  2.01s/it, loss=0.524]
Epoch 2/4:  54%|█████▍    | 680/1250 [18:00<18:29,  1.95s/it, loss=0.524]
Epoch 2/4:  54%|█████▍    | 680/1250 [18:03<18:29,  1.95s/it, loss=0.566]
Epoch 2/4:  54%|█████▍    | 681/1250 [18:03<18:57,  2.00s/it, loss=0.566]
Epoch 2/4:  54%|█████▍    | 681/1250 [18:05<18:57,  2.00s/it, loss=0.59] 
Epoch 2/4:  55%|█████▍    | 682/1250 [18:05<19:43,  2.08s/it, loss=0.59]
Epoch 2/4:  55%|█████▍    | 682/1250 [18:07<19:43,  2.08s/it, loss=1.06]
Epoch 2/4:  55%|█████▍    | 683/1250 [18:07<18:58,  2.01s/it, loss=1.06]
Epoch 2/4:  55%|█████▍    | 683/1250 [18:09<18:58,  2.01s/it, loss=0.77]
Epoch 2/4:  55%|█████▍    | 684/1250 [18:09<18:21,  1.95s/it, loss=0.77]
Epoch 2/4:  55%|█████▍    | 684/1250 [18:10<18:21,  1.95s/it, loss=0.538]
Epoch 2/4:  55%|█████▍    | 685/1250 [18:10<18:23,  1.95s/it, loss=0.538]
Epoch 2/4:  55%|█████▍    | 685/1250 [18:13<18:23,  1.95s/it, loss=0.813]
Epoch 2/4:  55%|█████▍    | 686/1250 [18:13<19:20,  2.06s/it, loss=0.813]
Epoch 2/4:  55%|█████▍    | 686/1250 [18:15<19:20,  2.06s/it, loss=0.654]
Epoch 2/4:  55%|█████▍    | 687/1250 [18:15<18:49,  2.01s/it, loss=0.654]
Epoch 2/4:  55%|█████▍    | 687/1250 [18:16<18:49,  2.01s/it, loss=0.6]  
Epoch 2/4:  55%|█████▌    | 688/1250 [18:16<18:10,  1.94s/it, loss=0.6]
Epoch 2/4:  55%|█████▌    | 688/1250 [18:19<18:10,  1.94s/it, loss=1.09]
Epoch 2/4:  55%|█████▌    | 689/1250 [18:19<18:49,  2.01s/it, loss=1.09]
Epoch 2/4:  55%|█████▌    | 689/1250 [18:21<18:49,  2.01s/it, loss=1]   
Epoch 2/4:  55%|█████▌    | 690/1250 [18:21<18:46,  2.01s/it, loss=1]
Epoch 2/4:  55%|█████▌    | 690/1250 [18:23<18:46,  2.01s/it, loss=0.406]
Epoch 2/4:  55%|█████▌    | 691/1250 [18:23<18:46,  2.01s/it, loss=0.406]
Epoch 2/4:  55%|█████▌    | 691/1250 [18:25<18:46,  2.01s/it, loss=0.748]
Epoch 2/4:  55%|█████▌    | 692/1250 [18:25<19:05,  2.05s/it, loss=0.748]
Epoch 2/4:  55%|█████▌    | 692/1250 [18:27<19:05,  2.05s/it, loss=0.79] 
Epoch 2/4:  55%|█████▌    | 693/1250 [18:27<18:23,  1.98s/it, loss=0.79]
Epoch 2/4:  55%|█████▌    | 693/1250 [18:29<18:23,  1.98s/it, loss=0.927]
Epoch 2/4:  56%|█████▌    | 694/1250 [18:29<18:54,  2.04s/it, loss=0.927]
Epoch 2/4:  56%|█████▌    | 694/1250 [18:31<18:54,  2.04s/it, loss=0.754]
Epoch 2/4:  56%|█████▌    | 695/1250 [18:31<18:37,  2.01s/it, loss=0.754]
Epoch 2/4:  56%|█████▌    | 695/1250 [18:33<18:37,  2.01s/it, loss=0.823]
Epoch 2/4:  56%|█████▌    | 696/1250 [18:33<18:54,  2.05s/it, loss=0.823]
Epoch 2/4:  56%|█████▌    | 696/1250 [18:35<18:54,  2.05s/it, loss=0.587]
Epoch 2/4:  56%|█████▌    | 697/1250 [18:35<19:37,  2.13s/it, loss=0.587]
Epoch 2/4:  56%|█████▌    | 697/1250 [18:37<19:37,  2.13s/it, loss=0.676]
Epoch 2/4:  56%|█████▌    | 698/1250 [18:37<19:25,  2.11s/it, loss=0.676]
Epoch 2/4:  56%|█████▌    | 698/1250 [18:40<19:25,  2.11s/it, loss=0.601]
Epoch 2/4:  56%|█████▌    | 699/1250 [18:40<19:52,  2.16s/it, loss=0.601]
Epoch 2/4:  56%|█████▌    | 699/1250 [18:42<19:52,  2.16s/it, loss=0.861]
Epoch 2/4:  56%|█████▌    | 700/1250 [18:42<20:07,  2.20s/it, loss=0.861]
Epoch 2/4:  56%|█████▌    | 700/1250 [18:44<20:07,  2.20s/it, loss=0.915]
Epoch 2/4:  56%|█████▌    | 701/1250 [18:44<19:27,  2.13s/it, loss=0.915]
Epoch 2/4:  56%|█████▌    | 701/1250 [18:46<19:27,  2.13s/it, loss=0.743]
Epoch 2/4:  56%|█████▌    | 702/1250 [18:46<20:05,  2.20s/it, loss=0.743]
Epoch 2/4:  56%|█████▌    | 702/1250 [18:47<20:05,  2.20s/it, loss=0.921]
Epoch 2/4:  56%|█████▌    | 703/1250 [18:47<17:34,  1.93s/it, loss=0.921]
Epoch 2/4:  56%|█████▌    | 703/1250 [18:50<17:34,  1.93s/it, loss=0.644]
Epoch 2/4:  56%|█████▋    | 704/1250 [18:50<18:03,  1.98s/it, loss=0.644]
Epoch 2/4:  56%|█████▋    | 704/1250 [18:51<18:03,  1.98s/it, loss=1]    
Epoch 2/4:  56%|█████▋    | 705/1250 [18:51<17:26,  1.92s/it, loss=1]
Epoch 2/4:  56%|█████▋    | 705/1250 [18:53<17:26,  1.92s/it, loss=0.523]
Epoch 2/4:  56%|█████▋    | 706/1250 [18:53<17:17,  1.91s/it, loss=0.523]
Epoch 2/4:  56%|█████▋    | 706/1250 [18:55<17:17,  1.91s/it, loss=0.769]
Epoch 2/4:  57%|█████▋    | 707/1250 [18:55<18:05,  2.00s/it, loss=0.769]
Epoch 2/4:  57%|█████▋    | 707/1250 [18:57<18:05,  2.00s/it, loss=0.785]
Epoch 2/4:  57%|█████▋    | 708/1250 [18:57<16:54,  1.87s/it, loss=0.785]
Epoch 2/4:  57%|█████▋    | 708/1250 [18:59<16:54,  1.87s/it, loss=0.77] 
Epoch 2/4:  57%|█████▋    | 709/1250 [18:59<17:05,  1.90s/it, loss=0.77]
Epoch 2/4:  57%|█████▋    | 709/1250 [19:01<17:05,  1.90s/it, loss=0.681]
Epoch 2/4:  57%|█████▋    | 710/1250 [19:01<18:32,  2.06s/it, loss=0.681]
Epoch 2/4:  57%|█████▋    | 710/1250 [19:03<18:32,  2.06s/it, loss=0.613]
Epoch 2/4:  57%|█████▋    | 711/1250 [19:03<18:10,  2.02s/it, loss=0.613]
Epoch 2/4:  57%|█████▋    | 711/1250 [19:05<18:10,  2.02s/it, loss=0.805]
Epoch 2/4:  57%|█████▋    | 712/1250 [19:05<18:17,  2.04s/it, loss=0.805]
Epoch 2/4:  57%|█████▋    | 712/1250 [19:07<18:17,  2.04s/it, loss=0.584]
Epoch 2/4:  57%|█████▋    | 713/1250 [19:07<18:06,  2.02s/it, loss=0.584]
Epoch 2/4:  57%|█████▋    | 713/1250 [19:09<18:06,  2.02s/it, loss=0.503]
Epoch 2/4:  57%|█████▋    | 714/1250 [19:09<18:04,  2.02s/it, loss=0.503]
Epoch 2/4:  57%|█████▋    | 714/1250 [19:11<18:04,  2.02s/it, loss=0.834]
Epoch 2/4:  57%|█████▋    | 715/1250 [19:11<18:06,  2.03s/it, loss=0.834]
Epoch 2/4:  57%|█████▋    | 715/1250 [19:14<18:06,  2.03s/it, loss=1.08] 
Epoch 2/4:  57%|█████▋    | 716/1250 [19:14<18:16,  2.05s/it, loss=1.08]
Epoch 2/4:  57%|█████▋    | 716/1250 [19:16<18:16,  2.05s/it, loss=0.533]
Epoch 2/4:  57%|█████▋    | 717/1250 [19:16<18:37,  2.10s/it, loss=0.533]
Epoch 2/4:  57%|█████▋    | 717/1250 [19:18<18:37,  2.10s/it, loss=0.52] 
Epoch 2/4:  57%|█████▋    | 718/1250 [19:18<17:51,  2.01s/it, loss=0.52]
Epoch 2/4:  57%|█████▋    | 718/1250 [19:20<17:51,  2.01s/it, loss=0.353]
Epoch 2/4:  58%|█████▊    | 719/1250 [19:20<18:02,  2.04s/it, loss=0.353]
Epoch 2/4:  58%|█████▊    | 719/1250 [19:22<18:02,  2.04s/it, loss=0.864]
Epoch 2/4:  58%|█████▊    | 720/1250 [19:22<17:42,  2.01s/it, loss=0.864]
Epoch 2/4:  58%|█████▊    | 720/1250 [19:24<17:42,  2.01s/it, loss=0.6]  
Epoch 2/4:  58%|█████▊    | 721/1250 [19:24<17:47,  2.02s/it, loss=0.6]
Epoch 2/4:  58%|█████▊    | 721/1250 [19:26<17:47,  2.02s/it, loss=0.655]
Epoch 2/4:  58%|█████▊    | 722/1250 [19:26<18:41,  2.12s/it, loss=0.655]
Epoch 2/4:  58%|█████▊    | 722/1250 [19:27<18:41,  2.12s/it, loss=0.561]
Epoch 2/4:  58%|█████▊    | 723/1250 [19:27<16:39,  1.90s/it, loss=0.561]
Epoch 2/4:  58%|█████▊    | 723/1250 [19:29<16:39,  1.90s/it, loss=0.921]
Epoch 2/4:  58%|█████▊    | 724/1250 [19:29<16:50,  1.92s/it, loss=0.921]
Epoch 2/4:  58%|█████▊    | 724/1250 [19:31<16:50,  1.92s/it, loss=1.52] 
Epoch 2/4:  58%|█████▊    | 725/1250 [19:31<17:12,  1.97s/it, loss=1.52]
Epoch 2/4:  58%|█████▊    | 725/1250 [19:33<17:12,  1.97s/it, loss=0.552]
Epoch 2/4:  58%|█████▊    | 726/1250 [19:33<17:13,  1.97s/it, loss=0.552]
Epoch 2/4:  58%|█████▊    | 726/1250 [19:35<17:13,  1.97s/it, loss=0.887]
Epoch 2/4:  58%|█████▊    | 727/1250 [19:35<17:02,  1.95s/it, loss=0.887]
Epoch 2/4:  58%|█████▊    | 727/1250 [19:37<17:02,  1.95s/it, loss=0.57] 
Epoch 2/4:  58%|█████▊    | 728/1250 [19:37<16:59,  1.95s/it, loss=0.57]
Epoch 2/4:  58%|█████▊    | 728/1250 [19:39<16:59,  1.95s/it, loss=0.569]
Epoch 2/4:  58%|█████▊    | 729/1250 [19:39<16:42,  1.92s/it, loss=0.569]
Epoch 2/4:  58%|█████▊    | 729/1250 [19:41<16:42,  1.92s/it, loss=0.75] 
Epoch 2/4:  58%|█████▊    | 730/1250 [19:41<17:03,  1.97s/it, loss=0.75]
Epoch 2/4:  58%|█████▊    | 730/1250 [19:43<17:03,  1.97s/it, loss=0.922]
Epoch 2/4:  58%|█████▊    | 731/1250 [19:43<16:37,  1.92s/it, loss=0.922]
Epoch 2/4:  58%|█████▊    | 731/1250 [19:45<16:37,  1.92s/it, loss=1.13] 
Epoch 2/4:  59%|█████▊    | 732/1250 [19:45<16:27,  1.91s/it, loss=1.13]
Epoch 2/4:  59%|█████▊    | 732/1250 [19:47<16:27,  1.91s/it, loss=0.485]
Epoch 2/4:  59%|█████▊    | 733/1250 [19:47<16:15,  1.89s/it, loss=0.485]
Epoch 2/4:  59%|█████▊    | 733/1250 [19:49<16:15,  1.89s/it, loss=0.568]
Epoch 2/4:  59%|█████▊    | 734/1250 [19:49<16:45,  1.95s/it, loss=0.568]
Epoch 2/4:  59%|█████▊    | 734/1250 [19:51<16:45,  1.95s/it, loss=0.819]
Epoch 2/4:  59%|█████▉    | 735/1250 [19:51<17:17,  2.01s/it, loss=0.819]
Epoch 2/4:  59%|█████▉    | 735/1250 [19:53<17:17,  2.01s/it, loss=0.697]
Epoch 2/4:  59%|█████▉    | 736/1250 [19:53<16:11,  1.89s/it, loss=0.697]
Epoch 2/4:  59%|█████▉    | 736/1250 [19:54<16:11,  1.89s/it, loss=0.823]
Epoch 2/4:  59%|█████▉    | 737/1250 [19:54<15:21,  1.80s/it, loss=0.823]
Epoch 2/4:  59%|█████▉    | 737/1250 [19:56<15:21,  1.80s/it, loss=0.559]
Epoch 2/4:  59%|█████▉    | 738/1250 [19:56<16:10,  1.89s/it, loss=0.559]
Epoch 2/4:  59%|█████▉    | 738/1250 [19:58<16:10,  1.89s/it, loss=0.59] 
Epoch 2/4:  59%|█████▉    | 739/1250 [19:58<16:02,  1.88s/it, loss=0.59]
Epoch 2/4:  59%|█████▉    | 739/1250 [20:00<16:02,  1.88s/it, loss=0.515]
Epoch 2/4:  59%|█████▉    | 740/1250 [20:00<16:06,  1.89s/it, loss=0.515]
Epoch 2/4:  59%|█████▉    | 740/1250 [20:02<16:06,  1.89s/it, loss=0.632]
Epoch 2/4:  59%|█████▉    | 741/1250 [20:02<15:58,  1.88s/it, loss=0.632]
Epoch 2/4:  59%|█████▉    | 741/1250 [20:04<15:58,  1.88s/it, loss=0.461]
Epoch 2/4:  59%|█████▉    | 742/1250 [20:04<15:33,  1.84s/it, loss=0.461]
Epoch 2/4:  59%|█████▉    | 742/1250 [20:06<15:33,  1.84s/it, loss=0.671]
Epoch 2/4:  59%|█████▉    | 743/1250 [20:06<15:31,  1.84s/it, loss=0.671]
Epoch 2/4:  59%|█████▉    | 743/1250 [20:07<15:31,  1.84s/it, loss=0.448]
Epoch 2/4:  60%|█████▉    | 744/1250 [20:07<15:43,  1.86s/it, loss=0.448]
Epoch 2/4:  60%|█████▉    | 744/1250 [20:10<15:43,  1.86s/it, loss=0.709]
Epoch 2/4:  60%|█████▉    | 745/1250 [20:10<16:14,  1.93s/it, loss=0.709]
Epoch 2/4:  60%|█████▉    | 745/1250 [20:12<16:14,  1.93s/it, loss=0.853]
Epoch 2/4:  60%|█████▉    | 746/1250 [20:12<16:16,  1.94s/it, loss=0.853]
Epoch 2/4:  60%|█████▉    | 746/1250 [20:13<16:16,  1.94s/it, loss=0.65] 
Epoch 2/4:  60%|█████▉    | 747/1250 [20:13<16:17,  1.94s/it, loss=0.65]
Epoch 2/4:  60%|█████▉    | 747/1250 [20:15<16:17,  1.94s/it, loss=0.639]
Epoch 2/4:  60%|█████▉    | 748/1250 [20:15<15:49,  1.89s/it, loss=0.639]
Epoch 2/4:  60%|█████▉    | 748/1250 [20:17<15:49,  1.89s/it, loss=0.739]
Epoch 2/4:  60%|█████▉    | 749/1250 [20:17<16:01,  1.92s/it, loss=0.739]
Epoch 2/4:  60%|█████▉    | 749/1250 [20:19<16:01,  1.92s/it, loss=0.931]
Epoch 2/4:  60%|██████    | 750/1250 [20:19<16:18,  1.96s/it, loss=0.931]
Epoch 2/4:  60%|██████    | 750/1250 [20:22<16:18,  1.96s/it, loss=0.6]  
Epoch 2/4:  60%|██████    | 751/1250 [20:22<17:06,  2.06s/it, loss=0.6]
Epoch 2/4:  60%|██████    | 751/1250 [20:24<17:06,  2.06s/it, loss=0.897]
Epoch 2/4:  60%|██████    | 752/1250 [20:24<17:12,  2.07s/it, loss=0.897]
Epoch 2/4:  60%|██████    | 752/1250 [20:25<17:12,  2.07s/it, loss=0.56] 
Epoch 2/4:  60%|██████    | 753/1250 [20:25<16:13,  1.96s/it, loss=0.56]
Epoch 2/4:  60%|██████    | 753/1250 [20:27<16:13,  1.96s/it, loss=0.537]
Epoch 2/4:  60%|██████    | 754/1250 [20:27<15:58,  1.93s/it, loss=0.537]
Epoch 2/4:  60%|██████    | 754/1250 [20:29<15:58,  1.93s/it, loss=0.731]
Epoch 2/4:  60%|██████    | 755/1250 [20:29<16:19,  1.98s/it, loss=0.731]
Epoch 2/4:  60%|██████    | 755/1250 [20:31<16:19,  1.98s/it, loss=0.956]
Epoch 2/4:  60%|██████    | 756/1250 [20:31<16:21,  1.99s/it, loss=0.956]
Epoch 2/4:  60%|██████    | 756/1250 [20:33<16:21,  1.99s/it, loss=0.973]
Epoch 2/4:  61%|██████    | 757/1250 [20:33<16:02,  1.95s/it, loss=0.973]
Epoch 2/4:  61%|██████    | 757/1250 [20:35<16:02,  1.95s/it, loss=0.557]
Epoch 2/4:  61%|██████    | 758/1250 [20:35<16:18,  1.99s/it, loss=0.557]
Epoch 2/4:  61%|██████    | 758/1250 [20:38<16:18,  1.99s/it, loss=0.749]
Epoch 2/4:  61%|██████    | 759/1250 [20:38<17:07,  2.09s/it, loss=0.749]
Epoch 2/4:  61%|██████    | 759/1250 [20:39<17:07,  2.09s/it, loss=0.635]
Epoch 2/4:  61%|██████    | 760/1250 [20:39<16:02,  1.96s/it, loss=0.635]
Epoch 2/4:  61%|██████    | 760/1250 [20:42<16:02,  1.96s/it, loss=0.496]
Epoch 2/4:  61%|██████    | 761/1250 [20:42<16:48,  2.06s/it, loss=0.496]
Epoch 2/4:  61%|██████    | 761/1250 [20:43<16:48,  2.06s/it, loss=0.636]
Epoch 2/4:  61%|██████    | 762/1250 [20:43<15:59,  1.97s/it, loss=0.636]
Epoch 2/4:  61%|██████    | 762/1250 [20:45<15:59,  1.97s/it, loss=0.542]
Epoch 2/4:  61%|██████    | 763/1250 [20:45<16:29,  2.03s/it, loss=0.542]
Epoch 2/4:  61%|██████    | 763/1250 [20:47<16:29,  2.03s/it, loss=0.823]
Epoch 2/4:  61%|██████    | 764/1250 [20:47<15:54,  1.96s/it, loss=0.823]
Epoch 2/4:  61%|██████    | 764/1250 [20:49<15:54,  1.96s/it, loss=0.604]
Epoch 2/4:  61%|██████    | 765/1250 [20:49<16:21,  2.02s/it, loss=0.604]
Epoch 2/4:  61%|██████    | 765/1250 [20:52<16:21,  2.02s/it, loss=0.499]
Epoch 2/4:  61%|██████▏   | 766/1250 [20:52<17:14,  2.14s/it, loss=0.499]
Epoch 2/4:  61%|██████▏   | 766/1250 [20:53<17:14,  2.14s/it, loss=0.643]
Epoch 2/4:  61%|██████▏   | 767/1250 [20:53<15:41,  1.95s/it, loss=0.643]
Epoch 2/4:  61%|██████▏   | 767/1250 [20:56<15:41,  1.95s/it, loss=0.852]
Epoch 2/4:  61%|██████▏   | 768/1250 [20:56<17:07,  2.13s/it, loss=0.852]
Epoch 2/4:  61%|██████▏   | 768/1250 [20:58<17:07,  2.13s/it, loss=0.782]
Epoch 2/4:  62%|██████▏   | 769/1250 [20:58<16:38,  2.08s/it, loss=0.782]
Epoch 2/4:  62%|██████▏   | 769/1250 [21:00<16:38,  2.08s/it, loss=0.626]
Epoch 2/4:  62%|██████▏   | 770/1250 [21:00<15:49,  1.98s/it, loss=0.626]
Epoch 2/4:  62%|██████▏   | 770/1250 [21:02<15:49,  1.98s/it, loss=0.626]
Epoch 2/4:  62%|██████▏   | 771/1250 [21:02<16:15,  2.04s/it, loss=0.626]
Epoch 2/4:  62%|██████▏   | 771/1250 [21:04<16:15,  2.04s/it, loss=0.483]
Epoch 2/4:  62%|██████▏   | 772/1250 [21:04<15:55,  2.00s/it, loss=0.483]
Epoch 2/4:  62%|██████▏   | 772/1250 [21:06<15:55,  2.00s/it, loss=0.636]
Epoch 2/4:  62%|██████▏   | 773/1250 [21:06<15:59,  2.01s/it, loss=0.636]
Epoch 2/4:  62%|██████▏   | 773/1250 [21:08<15:59,  2.01s/it, loss=0.822]
Epoch 2/4:  62%|██████▏   | 774/1250 [21:08<15:49,  2.00s/it, loss=0.822]
Epoch 2/4:  62%|██████▏   | 774/1250 [21:10<15:49,  2.00s/it, loss=1.01] 
Epoch 2/4:  62%|██████▏   | 775/1250 [21:10<15:47,  2.00s/it, loss=1.01]
Epoch 2/4:  62%|██████▏   | 775/1250 [21:12<15:47,  2.00s/it, loss=0.591]
Epoch 2/4:  62%|██████▏   | 776/1250 [21:12<16:32,  2.09s/it, loss=0.591]
Epoch 2/4:  62%|██████▏   | 776/1250 [21:14<16:32,  2.09s/it, loss=0.753]
Epoch 2/4:  62%|██████▏   | 777/1250 [21:14<15:34,  1.97s/it, loss=0.753]
Epoch 2/4:  62%|██████▏   | 777/1250 [21:16<15:34,  1.97s/it, loss=0.7]  
Epoch 2/4:  62%|██████▏   | 778/1250 [21:16<15:37,  1.99s/it, loss=0.7]
Epoch 2/4:  62%|██████▏   | 778/1250 [21:18<15:37,  1.99s/it, loss=0.826]
Epoch 2/4:  62%|██████▏   | 779/1250 [21:18<15:14,  1.94s/it, loss=0.826]
Epoch 2/4:  62%|██████▏   | 779/1250 [21:20<15:14,  1.94s/it, loss=1.01] 
Epoch 2/4:  62%|██████▏   | 780/1250 [21:20<15:46,  2.01s/it, loss=1.01]
Epoch 2/4:  62%|██████▏   | 780/1250 [21:22<15:46,  2.01s/it, loss=0.921]
Epoch 2/4:  62%|██████▏   | 781/1250 [21:22<15:28,  1.98s/it, loss=0.921]
Epoch 2/4:  62%|██████▏   | 781/1250 [21:23<15:28,  1.98s/it, loss=0.682]
Epoch 2/4:  63%|██████▎   | 782/1250 [21:23<14:55,  1.91s/it, loss=0.682]
Epoch 2/4:  63%|██████▎   | 782/1250 [21:25<14:55,  1.91s/it, loss=1.13] 
Epoch 2/4:  63%|██████▎   | 783/1250 [21:25<14:54,  1.92s/it, loss=1.13]
Epoch 2/4:  63%|██████▎   | 783/1250 [21:27<14:54,  1.92s/it, loss=0.574]
Epoch 2/4:  63%|██████▎   | 784/1250 [21:27<14:59,  1.93s/it, loss=0.574]
Epoch 2/4:  63%|██████▎   | 784/1250 [21:29<14:59,  1.93s/it, loss=0.615]
Epoch 2/4:  63%|██████▎   | 785/1250 [21:29<15:16,  1.97s/it, loss=0.615]
Epoch 2/4:  63%|██████▎   | 785/1250 [21:31<15:16,  1.97s/it, loss=0.695]
Epoch 2/4:  63%|██████▎   | 786/1250 [21:31<15:24,  1.99s/it, loss=0.695]
Epoch 2/4:  63%|██████▎   | 786/1250 [21:33<15:24,  1.99s/it, loss=0.609]
Epoch 2/4:  63%|██████▎   | 787/1250 [21:33<14:57,  1.94s/it, loss=0.609]
Epoch 2/4:  63%|██████▎   | 787/1250 [21:35<14:57,  1.94s/it, loss=1.12] 
Epoch 2/4:  63%|██████▎   | 788/1250 [21:35<14:18,  1.86s/it, loss=1.12]
Epoch 2/4:  63%|██████▎   | 788/1250 [21:37<14:18,  1.86s/it, loss=0.581]
Epoch 2/4:  63%|██████▎   | 789/1250 [21:37<15:46,  2.05s/it, loss=0.581]
Epoch 2/4:  63%|██████▎   | 789/1250 [21:39<15:46,  2.05s/it, loss=1.14] 
Epoch 2/4:  63%|██████▎   | 790/1250 [21:39<14:41,  1.92s/it, loss=1.14]
Epoch 2/4:  63%|██████▎   | 790/1250 [21:41<14:41,  1.92s/it, loss=0.305]
Epoch 2/4:  63%|██████▎   | 791/1250 [21:41<15:11,  1.99s/it, loss=0.305]
Epoch 2/4:  63%|██████▎   | 791/1250 [21:43<15:11,  1.99s/it, loss=0.932]
Epoch 2/4:  63%|██████▎   | 792/1250 [21:43<15:30,  2.03s/it, loss=0.932]
Epoch 2/4:  63%|██████▎   | 792/1250 [21:45<15:30,  2.03s/it, loss=0.782]
Epoch 2/4:  63%|██████▎   | 793/1250 [21:45<15:20,  2.01s/it, loss=0.782]
Epoch 2/4:  63%|██████▎   | 793/1250 [21:47<15:20,  2.01s/it, loss=0.781]
Epoch 2/4:  64%|██████▎   | 794/1250 [21:47<14:55,  1.96s/it, loss=0.781]
Epoch 2/4:  64%|██████▎   | 794/1250 [21:49<14:55,  1.96s/it, loss=0.785]
Epoch 2/4:  64%|██████▎   | 795/1250 [21:49<14:41,  1.94s/it, loss=0.785]
Epoch 2/4:  64%|██████▎   | 795/1250 [21:51<14:41,  1.94s/it, loss=0.768]
Epoch 2/4:  64%|██████▎   | 796/1250 [21:51<15:17,  2.02s/it, loss=0.768]
Epoch 2/4:  64%|██████▎   | 796/1250 [21:53<15:17,  2.02s/it, loss=0.543]
Epoch 2/4:  64%|██████▍   | 797/1250 [21:53<15:01,  1.99s/it, loss=0.543]
Epoch 2/4:  64%|██████▍   | 797/1250 [21:55<15:01,  1.99s/it, loss=0.929]
Epoch 2/4:  64%|██████▍   | 798/1250 [21:55<15:03,  2.00s/it, loss=0.929]
Epoch 2/4:  64%|██████▍   | 798/1250 [21:57<15:03,  2.00s/it, loss=0.7]  
Epoch 2/4:  64%|██████▍   | 799/1250 [21:57<15:45,  2.10s/it, loss=0.7]
Epoch 2/4:  64%|██████▍   | 799/1250 [21:59<15:45,  2.10s/it, loss=0.722]
Epoch 2/4:  64%|██████▍   | 800/1250 [21:59<14:44,  1.97s/it, loss=0.722]
Epoch 2/4:  64%|██████▍   | 800/1250 [22:01<14:44,  1.97s/it, loss=0.651]
Epoch 2/4:  64%|██████▍   | 801/1250 [22:01<14:51,  1.98s/it, loss=0.651]
Epoch 2/4:  64%|██████▍   | 801/1250 [22:03<14:51,  1.98s/it, loss=0.565]
Epoch 2/4:  64%|██████▍   | 802/1250 [22:03<15:16,  2.05s/it, loss=0.565]
Epoch 2/4:  64%|██████▍   | 802/1250 [22:05<15:16,  2.05s/it, loss=0.716]
Epoch 2/4:  64%|██████▍   | 803/1250 [22:05<15:05,  2.03s/it, loss=0.716]
Epoch 2/4:  64%|██████▍   | 803/1250 [22:07<15:05,  2.03s/it, loss=0.713]
Epoch 2/4:  64%|██████▍   | 804/1250 [22:07<15:15,  2.05s/it, loss=0.713]
Epoch 2/4:  64%|██████▍   | 804/1250 [22:09<15:15,  2.05s/it, loss=0.613]
Epoch 2/4:  64%|██████▍   | 805/1250 [22:09<14:36,  1.97s/it, loss=0.613]
Epoch 2/4:  64%|██████▍   | 805/1250 [22:11<14:36,  1.97s/it, loss=0.703]
Epoch 2/4:  64%|██████▍   | 806/1250 [22:11<14:51,  2.01s/it, loss=0.703]
Epoch 2/4:  64%|██████▍   | 806/1250 [22:13<14:51,  2.01s/it, loss=0.4]  
Epoch 2/4:  65%|██████▍   | 807/1250 [22:13<14:58,  2.03s/it, loss=0.4]
Epoch 2/4:  65%|██████▍   | 807/1250 [22:15<14:58,  2.03s/it, loss=0.614]
Epoch 2/4:  65%|██████▍   | 808/1250 [22:15<14:17,  1.94s/it, loss=0.614]
Epoch 2/4:  65%|██████▍   | 808/1250 [22:17<14:17,  1.94s/it, loss=0.627]
Epoch 2/4:  65%|██████▍   | 809/1250 [22:17<14:15,  1.94s/it, loss=0.627]
Epoch 2/4:  65%|██████▍   | 809/1250 [22:19<14:15,  1.94s/it, loss=0.829]
Epoch 2/4:  65%|██████▍   | 810/1250 [22:19<14:22,  1.96s/it, loss=0.829]
Epoch 2/4:  65%|██████▍   | 810/1250 [22:21<14:22,  1.96s/it, loss=0.719]
Epoch 2/4:  65%|██████▍   | 811/1250 [22:21<14:08,  1.93s/it, loss=0.719]
Epoch 2/4:  65%|██████▍   | 811/1250 [22:23<14:08,  1.93s/it, loss=0.688]
Epoch 2/4:  65%|██████▍   | 812/1250 [22:23<14:23,  1.97s/it, loss=0.688]
Epoch 2/4:  65%|██████▍   | 812/1250 [22:25<14:23,  1.97s/it, loss=0.847]
Epoch 2/4:  65%|██████▌   | 813/1250 [22:25<14:16,  1.96s/it, loss=0.847]
Epoch 2/4:  65%|██████▌   | 813/1250 [22:27<14:16,  1.96s/it, loss=0.371]
Epoch 2/4:  65%|██████▌   | 814/1250 [22:27<14:19,  1.97s/it, loss=0.371]
Epoch 2/4:  65%|██████▌   | 814/1250 [22:29<14:19,  1.97s/it, loss=0.632]
Epoch 2/4:  65%|██████▌   | 815/1250 [22:29<14:19,  1.98s/it, loss=0.632]
Epoch 2/4:  65%|██████▌   | 815/1250 [22:31<14:19,  1.98s/it, loss=0.997]
Epoch 2/4:  65%|██████▌   | 816/1250 [22:31<14:25,  1.99s/it, loss=0.997]
Epoch 2/4:  65%|██████▌   | 816/1250 [22:33<14:25,  1.99s/it, loss=0.592]
Epoch 2/4:  65%|██████▌   | 817/1250 [22:33<14:37,  2.03s/it, loss=0.592]
Epoch 2/4:  65%|██████▌   | 817/1250 [22:35<14:37,  2.03s/it, loss=0.571]
Epoch 2/4:  65%|██████▌   | 818/1250 [22:35<14:36,  2.03s/it, loss=0.571]
Epoch 2/4:  65%|██████▌   | 818/1250 [22:37<14:36,  2.03s/it, loss=0.777]
Epoch 2/4:  66%|██████▌   | 819/1250 [22:37<14:55,  2.08s/it, loss=0.777]
Epoch 2/4:  66%|██████▌   | 819/1250 [22:39<14:55,  2.08s/it, loss=0.781]
Epoch 2/4:  66%|██████▌   | 820/1250 [22:39<14:23,  2.01s/it, loss=0.781]
Epoch 2/4:  66%|██████▌   | 820/1250 [22:41<14:23,  2.01s/it, loss=0.468]
Epoch 2/4:  66%|██████▌   | 821/1250 [22:41<14:58,  2.09s/it, loss=0.468]
Epoch 2/4:  66%|██████▌   | 821/1250 [22:44<14:58,  2.09s/it, loss=0.864]
Epoch 2/4:  66%|██████▌   | 822/1250 [22:44<14:57,  2.10s/it, loss=0.864]
Epoch 2/4:  66%|██████▌   | 822/1250 [22:46<14:57,  2.10s/it, loss=1.01] 
Epoch 2/4:  66%|██████▌   | 823/1250 [22:46<14:55,  2.10s/it, loss=1.01]
Epoch 2/4:  66%|██████▌   | 823/1250 [22:48<14:55,  2.10s/it, loss=0.635]
Epoch 2/4:  66%|██████▌   | 824/1250 [22:48<14:45,  2.08s/it, loss=0.635]
Epoch 2/4:  66%|██████▌   | 824/1250 [22:49<14:45,  2.08s/it, loss=0.607]
Epoch 2/4:  66%|██████▌   | 825/1250 [22:49<14:09,  2.00s/it, loss=0.607]
Epoch 2/4:  66%|██████▌   | 825/1250 [22:51<14:09,  2.00s/it, loss=0.616]
Epoch 2/4:  66%|██████▌   | 826/1250 [22:51<14:10,  2.01s/it, loss=0.616]
Epoch 2/4:  66%|██████▌   | 826/1250 [22:53<14:10,  2.01s/it, loss=0.819]
Epoch 2/4:  66%|██████▌   | 827/1250 [22:53<13:43,  1.95s/it, loss=0.819]
Epoch 2/4:  66%|██████▌   | 827/1250 [22:55<13:43,  1.95s/it, loss=0.623]
Epoch 2/4:  66%|██████▌   | 828/1250 [22:55<13:52,  1.97s/it, loss=0.623]
Epoch 2/4:  66%|██████▌   | 828/1250 [22:57<13:52,  1.97s/it, loss=0.624]
Epoch 2/4:  66%|██████▋   | 829/1250 [22:57<14:04,  2.01s/it, loss=0.624]
Epoch 2/4:  66%|██████▋   | 829/1250 [22:59<14:04,  2.01s/it, loss=0.538]
Epoch 2/4:  66%|██████▋   | 830/1250 [22:59<13:41,  1.96s/it, loss=0.538]
Epoch 2/4:  66%|██████▋   | 830/1250 [23:01<13:41,  1.96s/it, loss=0.701]
Epoch 2/4:  66%|██████▋   | 831/1250 [23:01<13:46,  1.97s/it, loss=0.701]
Epoch 2/4:  66%|██████▋   | 831/1250 [23:03<13:46,  1.97s/it, loss=0.642]
Epoch 2/4:  67%|██████▋   | 832/1250 [23:03<13:40,  1.96s/it, loss=0.642]
Epoch 2/4:  67%|██████▋   | 832/1250 [23:05<13:40,  1.96s/it, loss=0.972]
Epoch 2/4:  67%|██████▋   | 833/1250 [23:05<13:49,  1.99s/it, loss=0.972]
Epoch 2/4:  67%|██████▋   | 833/1250 [23:07<13:49,  1.99s/it, loss=1.22] 
Epoch 2/4:  67%|██████▋   | 834/1250 [23:07<13:53,  2.00s/it, loss=1.22]
Epoch 2/4:  67%|██████▋   | 834/1250 [23:09<13:53,  2.00s/it, loss=0.733]
Epoch 2/4:  67%|██████▋   | 835/1250 [23:09<13:44,  1.99s/it, loss=0.733]
Epoch 2/4:  67%|██████▋   | 835/1250 [23:11<13:44,  1.99s/it, loss=0.624]
Epoch 2/4:  67%|██████▋   | 836/1250 [23:12<14:18,  2.07s/it, loss=0.624]
Epoch 2/4:  67%|██████▋   | 836/1250 [23:14<14:18,  2.07s/it, loss=0.725]
Epoch 2/4:  67%|██████▋   | 837/1250 [23:14<14:12,  2.06s/it, loss=0.725]
Epoch 2/4:  67%|██████▋   | 837/1250 [23:15<14:12,  2.06s/it, loss=0.738]
Epoch 2/4:  67%|██████▋   | 838/1250 [23:15<13:55,  2.03s/it, loss=0.738]
Epoch 2/4:  67%|██████▋   | 838/1250 [23:17<13:55,  2.03s/it, loss=0.781]
Epoch 2/4:  67%|██████▋   | 839/1250 [23:17<12:58,  1.89s/it, loss=0.781]
Epoch 2/4:  67%|██████▋   | 839/1250 [23:19<12:58,  1.89s/it, loss=0.565]
Epoch 2/4:  67%|██████▋   | 840/1250 [23:19<13:24,  1.96s/it, loss=0.565]
Epoch 2/4:  67%|██████▋   | 840/1250 [23:21<13:24,  1.96s/it, loss=0.615]
Epoch 2/4:  67%|██████▋   | 841/1250 [23:21<13:19,  1.95s/it, loss=0.615]
Epoch 2/4:  67%|██████▋   | 841/1250 [23:23<13:19,  1.95s/it, loss=0.452]
Epoch 2/4:  67%|██████▋   | 842/1250 [23:23<14:01,  2.06s/it, loss=0.452]
Epoch 2/4:  67%|██████▋   | 842/1250 [23:25<14:01,  2.06s/it, loss=0.459]
Epoch 2/4:  67%|██████▋   | 843/1250 [23:25<13:14,  1.95s/it, loss=0.459]
Epoch 2/4:  67%|██████▋   | 843/1250 [23:27<13:14,  1.95s/it, loss=0.718]
Epoch 2/4:  68%|██████▊   | 844/1250 [23:27<13:31,  2.00s/it, loss=0.718]
Epoch 2/4:  68%|██████▊   | 844/1250 [23:29<13:31,  2.00s/it, loss=0.885]
Epoch 2/4:  68%|██████▊   | 845/1250 [23:29<13:04,  1.94s/it, loss=0.885]
Epoch 2/4:  68%|██████▊   | 845/1250 [23:31<13:04,  1.94s/it, loss=0.532]
Epoch 2/4:  68%|██████▊   | 846/1250 [23:31<13:05,  1.94s/it, loss=0.532]
Epoch 2/4:  68%|██████▊   | 846/1250 [23:33<13:05,  1.94s/it, loss=0.594]
Epoch 2/4:  68%|██████▊   | 847/1250 [23:33<13:58,  2.08s/it, loss=0.594]
Epoch 2/4:  68%|██████▊   | 847/1250 [23:35<13:58,  2.08s/it, loss=0.546]
Epoch 2/4:  68%|██████▊   | 848/1250 [23:35<13:00,  1.94s/it, loss=0.546]
Epoch 2/4:  68%|██████▊   | 848/1250 [23:37<13:00,  1.94s/it, loss=0.872]
Epoch 2/4:  68%|██████▊   | 849/1250 [23:37<14:00,  2.10s/it, loss=0.872]
Epoch 2/4:  68%|██████▊   | 849/1250 [23:39<14:00,  2.10s/it, loss=0.502]
Epoch 2/4:  68%|██████▊   | 850/1250 [23:39<13:42,  2.06s/it, loss=0.502]
Epoch 2/4:  68%|██████▊   | 850/1250 [23:42<13:42,  2.06s/it, loss=0.654]
Epoch 2/4:  68%|██████▊   | 851/1250 [23:42<13:45,  2.07s/it, loss=0.654]
Epoch 2/4:  68%|██████▊   | 851/1250 [23:44<13:45,  2.07s/it, loss=0.649]
Epoch 2/4:  68%|██████▊   | 852/1250 [23:44<13:32,  2.04s/it, loss=0.649]
Epoch 2/4:  68%|██████▊   | 852/1250 [23:46<13:32,  2.04s/it, loss=0.928]
Epoch 2/4:  68%|██████▊   | 853/1250 [23:46<13:51,  2.10s/it, loss=0.928]
Epoch 2/4:  68%|██████▊   | 853/1250 [23:48<13:51,  2.10s/it, loss=0.551]
Epoch 2/4:  68%|██████▊   | 854/1250 [23:48<13:40,  2.07s/it, loss=0.551]
Epoch 2/4:  68%|██████▊   | 854/1250 [23:50<13:40,  2.07s/it, loss=0.898]
Epoch 2/4:  68%|██████▊   | 855/1250 [23:50<13:22,  2.03s/it, loss=0.898]
Epoch 2/4:  68%|██████▊   | 855/1250 [23:51<13:22,  2.03s/it, loss=0.711]
Epoch 2/4:  68%|██████▊   | 856/1250 [23:51<12:42,  1.93s/it, loss=0.711]
Epoch 2/4:  68%|██████▊   | 856/1250 [23:54<12:42,  1.93s/it, loss=0.851]
Epoch 2/4:  69%|██████▊   | 857/1250 [23:54<13:08,  2.01s/it, loss=0.851]
Epoch 2/4:  69%|██████▊   | 857/1250 [23:55<13:08,  2.01s/it, loss=0.495]
Epoch 2/4:  69%|██████▊   | 858/1250 [23:55<12:19,  1.89s/it, loss=0.495]
Epoch 2/4:  69%|██████▊   | 858/1250 [23:57<12:19,  1.89s/it, loss=0.943]
Epoch 2/4:  69%|██████▊   | 859/1250 [23:57<12:56,  1.99s/it, loss=0.943]
Epoch 2/4:  69%|██████▊   | 859/1250 [24:00<12:56,  1.99s/it, loss=0.381]
Epoch 2/4:  69%|██████▉   | 860/1250 [24:00<13:17,  2.04s/it, loss=0.381]
Epoch 2/4:  69%|██████▉   | 860/1250 [24:01<13:17,  2.04s/it, loss=0.849]
Epoch 2/4:  69%|██████▉   | 861/1250 [24:01<12:18,  1.90s/it, loss=0.849]
Epoch 2/4:  69%|██████▉   | 861/1250 [24:03<12:18,  1.90s/it, loss=0.761]
Epoch 2/4:  69%|██████▉   | 862/1250 [24:03<12:32,  1.94s/it, loss=0.761]
Epoch 2/4:  69%|██████▉   | 862/1250 [24:05<12:32,  1.94s/it, loss=0.717]
Epoch 2/4:  69%|██████▉   | 863/1250 [24:05<12:14,  1.90s/it, loss=0.717]
Epoch 2/4:  69%|██████▉   | 863/1250 [24:07<12:14,  1.90s/it, loss=0.557]
Epoch 2/4:  69%|██████▉   | 864/1250 [24:07<12:50,  2.00s/it, loss=0.557]
Epoch 2/4:  69%|██████▉   | 864/1250 [24:09<12:50,  2.00s/it, loss=0.55] 
Epoch 2/4:  69%|██████▉   | 865/1250 [24:09<12:57,  2.02s/it, loss=0.55]
Epoch 2/4:  69%|██████▉   | 865/1250 [24:11<12:57,  2.02s/it, loss=0.78]
Epoch 2/4:  69%|██████▉   | 866/1250 [24:11<12:18,  1.92s/it, loss=0.78]
Epoch 2/4:  69%|██████▉   | 866/1250 [24:13<12:18,  1.92s/it, loss=0.563]
Epoch 2/4:  69%|██████▉   | 867/1250 [24:13<12:39,  1.98s/it, loss=0.563]
Epoch 2/4:  69%|██████▉   | 867/1250 [24:15<12:39,  1.98s/it, loss=0.786]
Epoch 2/4:  69%|██████▉   | 868/1250 [24:15<12:27,  1.96s/it, loss=0.786]
Epoch 2/4:  69%|██████▉   | 868/1250 [24:17<12:27,  1.96s/it, loss=0.761]
Epoch 2/4:  70%|██████▉   | 869/1250 [24:17<11:45,  1.85s/it, loss=0.761]
Epoch 2/4:  70%|██████▉   | 869/1250 [24:19<11:45,  1.85s/it, loss=1.02] 
Epoch 2/4:  70%|██████▉   | 870/1250 [24:19<13:01,  2.06s/it, loss=1.02]
Epoch 2/4:  70%|██████▉   | 870/1250 [24:21<13:01,  2.06s/it, loss=0.619]
Epoch 2/4:  70%|██████▉   | 871/1250 [24:21<12:13,  1.93s/it, loss=0.619]
Epoch 2/4:  70%|██████▉   | 871/1250 [24:23<12:13,  1.93s/it, loss=0.552]
Epoch 2/4:  70%|██████▉   | 872/1250 [24:23<12:23,  1.97s/it, loss=0.552]
Epoch 2/4:  70%|██████▉   | 872/1250 [24:25<12:23,  1.97s/it, loss=1.05] 
Epoch 2/4:  70%|██████▉   | 873/1250 [24:25<12:14,  1.95s/it, loss=1.05]
Epoch 2/4:  70%|██████▉   | 873/1250 [24:27<12:14,  1.95s/it, loss=0.78]
Epoch 2/4:  70%|██████▉   | 874/1250 [24:27<12:23,  1.98s/it, loss=0.78]
Epoch 2/4:  70%|██████▉   | 874/1250 [24:29<12:23,  1.98s/it, loss=1.17]
Epoch 2/4:  70%|███████   | 875/1250 [24:29<12:36,  2.02s/it, loss=1.17]
Epoch 2/4:  70%|███████   | 875/1250 [24:31<12:36,  2.02s/it, loss=0.639]
Epoch 2/4:  70%|███████   | 876/1250 [24:31<11:59,  1.92s/it, loss=0.639]
Epoch 2/4:  70%|███████   | 876/1250 [24:33<11:59,  1.92s/it, loss=0.523]
Epoch 2/4:  70%|███████   | 877/1250 [24:33<12:06,  1.95s/it, loss=0.523]
Epoch 2/4:  70%|███████   | 877/1250 [24:35<12:06,  1.95s/it, loss=0.646]
Epoch 2/4:  70%|███████   | 878/1250 [24:35<12:07,  1.96s/it, loss=0.646]
Epoch 2/4:  70%|███████   | 878/1250 [24:36<12:07,  1.96s/it, loss=0.594]
Epoch 2/4:  70%|███████   | 879/1250 [24:36<11:47,  1.91s/it, loss=0.594]
Epoch 2/4:  70%|███████   | 879/1250 [24:39<11:47,  1.91s/it, loss=0.369]
Epoch 2/4:  70%|███████   | 880/1250 [24:39<12:27,  2.02s/it, loss=0.369]
Epoch 2/4:  70%|███████   | 880/1250 [24:41<12:27,  2.02s/it, loss=0.714]
Epoch 2/4:  70%|███████   | 881/1250 [24:41<12:18,  2.00s/it, loss=0.714]
Epoch 2/4:  70%|███████   | 881/1250 [24:42<12:18,  2.00s/it, loss=0.675]
Epoch 2/4:  71%|███████   | 882/1250 [24:42<11:41,  1.91s/it, loss=0.675]
Epoch 2/4:  71%|███████   | 882/1250 [24:44<11:41,  1.91s/it, loss=0.598]
Epoch 2/4:  71%|███████   | 883/1250 [24:44<12:00,  1.96s/it, loss=0.598]
Epoch 2/4:  71%|███████   | 883/1250 [24:46<12:00,  1.96s/it, loss=0.739]
Epoch 2/4:  71%|███████   | 884/1250 [24:46<11:14,  1.84s/it, loss=0.739]
Epoch 2/4:  71%|███████   | 884/1250 [24:48<11:14,  1.84s/it, loss=0.927]
Epoch 2/4:  71%|███████   | 885/1250 [24:48<11:34,  1.90s/it, loss=0.927]
Epoch 2/4:  71%|███████   | 885/1250 [24:50<11:34,  1.90s/it, loss=0.532]
Epoch 2/4:  71%|███████   | 886/1250 [24:50<11:38,  1.92s/it, loss=0.532]
Epoch 2/4:  71%|███████   | 886/1250 [24:52<11:38,  1.92s/it, loss=0.603]
Epoch 2/4:  71%|███████   | 887/1250 [24:52<12:03,  1.99s/it, loss=0.603]
Epoch 2/4:  71%|███████   | 887/1250 [24:54<12:03,  1.99s/it, loss=0.779]
Epoch 2/4:  71%|███████   | 888/1250 [24:54<11:48,  1.96s/it, loss=0.779]
Epoch 2/4:  71%|███████   | 888/1250 [24:56<11:48,  1.96s/it, loss=0.822]
Epoch 2/4:  71%|███████   | 889/1250 [24:56<11:41,  1.94s/it, loss=0.822]
Epoch 2/4:  71%|███████   | 889/1250 [24:58<11:41,  1.94s/it, loss=0.489]
Epoch 2/4:  71%|███████   | 890/1250 [24:58<11:09,  1.86s/it, loss=0.489]
Epoch 2/4:  71%|███████   | 890/1250 [25:00<11:09,  1.86s/it, loss=0.962]
Epoch 2/4:  71%|███████▏  | 891/1250 [25:00<11:43,  1.96s/it, loss=0.962]
Epoch 2/4:  71%|███████▏  | 891/1250 [25:01<11:43,  1.96s/it, loss=0.848]
Epoch 2/4:  71%|███████▏  | 892/1250 [25:01<10:47,  1.81s/it, loss=0.848]
Epoch 2/4:  71%|███████▏  | 892/1250 [25:03<10:47,  1.81s/it, loss=0.738]
Epoch 2/4:  71%|███████▏  | 893/1250 [25:03<11:27,  1.92s/it, loss=0.738]
Epoch 2/4:  71%|███████▏  | 893/1250 [25:05<11:27,  1.92s/it, loss=0.715]
Epoch 2/4:  72%|███████▏  | 894/1250 [25:05<11:19,  1.91s/it, loss=0.715]
Epoch 2/4:  72%|███████▏  | 894/1250 [25:07<11:19,  1.91s/it, loss=0.461]
Epoch 2/4:  72%|███████▏  | 895/1250 [25:07<11:21,  1.92s/it, loss=0.461]
Epoch 2/4:  72%|███████▏  | 895/1250 [25:09<11:21,  1.92s/it, loss=0.795]
Epoch 2/4:  72%|███████▏  | 896/1250 [25:09<11:46,  2.00s/it, loss=0.795]
Epoch 2/4:  72%|███████▏  | 896/1250 [25:11<11:46,  2.00s/it, loss=0.872]
Epoch 2/4:  72%|███████▏  | 897/1250 [25:11<11:18,  1.92s/it, loss=0.872]
Epoch 2/4:  72%|███████▏  | 897/1250 [25:13<11:18,  1.92s/it, loss=0.676]
Epoch 2/4:  72%|███████▏  | 898/1250 [25:13<11:40,  1.99s/it, loss=0.676]
Epoch 2/4:  72%|███████▏  | 898/1250 [25:15<11:40,  1.99s/it, loss=1.09] 
Epoch 2/4:  72%|███████▏  | 899/1250 [25:15<11:30,  1.97s/it, loss=1.09]
Epoch 2/4:  72%|███████▏  | 899/1250 [25:17<11:30,  1.97s/it, loss=0.859]
Epoch 2/4:  72%|███████▏  | 900/1250 [25:17<11:29,  1.97s/it, loss=0.859]
Epoch 2/4:  72%|███████▏  | 900/1250 [25:19<11:29,  1.97s/it, loss=0.645]
Epoch 2/4:  72%|███████▏  | 901/1250 [25:19<11:15,  1.93s/it, loss=0.645]
Epoch 2/4:  72%|███████▏  | 901/1250 [25:21<11:15,  1.93s/it, loss=0.426]
Epoch 2/4:  72%|███████▏  | 902/1250 [25:21<11:04,  1.91s/it, loss=0.426]
Epoch 2/4:  72%|███████▏  | 902/1250 [25:23<11:04,  1.91s/it, loss=0.498]
Epoch 2/4:  72%|███████▏  | 903/1250 [25:23<10:42,  1.85s/it, loss=0.498]
Epoch 2/4:  72%|███████▏  | 903/1250 [25:25<10:42,  1.85s/it, loss=0.639]
Epoch 2/4:  72%|███████▏  | 904/1250 [25:25<11:14,  1.95s/it, loss=0.639]
Epoch 2/4:  72%|███████▏  | 904/1250 [25:27<11:14,  1.95s/it, loss=0.682]
Epoch 2/4:  72%|███████▏  | 905/1250 [25:27<11:39,  2.03s/it, loss=0.682]
Epoch 2/4:  72%|███████▏  | 905/1250 [25:29<11:39,  2.03s/it, loss=0.76] 
Epoch 2/4:  72%|███████▏  | 906/1250 [25:29<11:06,  1.94s/it, loss=0.76]
Epoch 2/4:  72%|███████▏  | 906/1250 [25:30<11:06,  1.94s/it, loss=0.698]
Epoch 2/4:  73%|███████▎  | 907/1250 [25:30<10:43,  1.88s/it, loss=0.698]
Epoch 2/4:  73%|███████▎  | 907/1250 [25:32<10:43,  1.88s/it, loss=0.763]
Epoch 2/4:  73%|███████▎  | 908/1250 [25:32<10:56,  1.92s/it, loss=0.763]
Epoch 2/4:  73%|███████▎  | 908/1250 [25:34<10:56,  1.92s/it, loss=0.841]
Epoch 2/4:  73%|███████▎  | 909/1250 [25:34<11:07,  1.96s/it, loss=0.841]
Epoch 2/4:  73%|███████▎  | 909/1250 [25:36<11:07,  1.96s/it, loss=0.648]
Epoch 2/4:  73%|███████▎  | 910/1250 [25:36<10:39,  1.88s/it, loss=0.648]
Epoch 2/4:  73%|███████▎  | 910/1250 [25:38<10:39,  1.88s/it, loss=0.79] 
Epoch 2/4:  73%|███████▎  | 911/1250 [25:38<11:11,  1.98s/it, loss=0.79]
Epoch 2/4:  73%|███████▎  | 911/1250 [25:40<11:11,  1.98s/it, loss=0.638]
Epoch 2/4:  73%|███████▎  | 912/1250 [25:40<10:59,  1.95s/it, loss=0.638]
Epoch 2/4:  73%|███████▎  | 912/1250 [25:42<10:59,  1.95s/it, loss=0.564]
Epoch 2/4:  73%|███████▎  | 913/1250 [25:42<11:09,  1.99s/it, loss=0.564]
Epoch 2/4:  73%|███████▎  | 913/1250 [25:44<11:09,  1.99s/it, loss=0.691]
Epoch 2/4:  73%|███████▎  | 914/1250 [25:44<10:52,  1.94s/it, loss=0.691]
Epoch 2/4:  73%|███████▎  | 914/1250 [25:46<10:52,  1.94s/it, loss=0.556]
Epoch 2/4:  73%|███████▎  | 915/1250 [25:46<10:38,  1.91s/it, loss=0.556]
Epoch 2/4:  73%|███████▎  | 915/1250 [25:48<10:38,  1.91s/it, loss=0.494]
Epoch 2/4:  73%|███████▎  | 916/1250 [25:48<10:50,  1.95s/it, loss=0.494]
Epoch 2/4:  73%|███████▎  | 916/1250 [25:50<10:50,  1.95s/it, loss=0.893]
Epoch 2/4:  73%|███████▎  | 917/1250 [25:50<10:53,  1.96s/it, loss=0.893]
Epoch 2/4:  73%|███████▎  | 917/1250 [25:52<10:53,  1.96s/it, loss=0.681]
Epoch 2/4:  73%|███████▎  | 918/1250 [25:52<10:50,  1.96s/it, loss=0.681]
Epoch 2/4:  73%|███████▎  | 918/1250 [25:55<10:50,  1.96s/it, loss=0.516]
Epoch 2/4:  74%|███████▎  | 919/1250 [25:55<12:11,  2.21s/it, loss=0.516]
Epoch 2/4:  74%|███████▎  | 919/1250 [25:57<12:11,  2.21s/it, loss=0.604]
Epoch 2/4:  74%|███████▎  | 920/1250 [25:57<12:08,  2.21s/it, loss=0.604]
Epoch 2/4:  74%|███████▎  | 920/1250 [25:58<12:08,  2.21s/it, loss=0.619]
Epoch 2/4:  74%|███████▎  | 921/1250 [25:58<10:01,  1.83s/it, loss=0.619]
Epoch 2/4:  74%|███████▎  | 921/1250 [25:59<10:01,  1.83s/it, loss=0.675]
Epoch 2/4:  74%|███████▍  | 922/1250 [25:59<08:51,  1.62s/it, loss=0.675]
Epoch 2/4:  74%|███████▍  | 922/1250 [26:01<08:51,  1.62s/it, loss=0.97] 
Epoch 2/4:  74%|███████▍  | 923/1250 [26:01<08:28,  1.56s/it, loss=0.97]
Epoch 2/4:  74%|███████▍  | 923/1250 [26:02<08:28,  1.56s/it, loss=0.82]
Epoch 2/4:  74%|███████▍  | 924/1250 [26:02<07:56,  1.46s/it, loss=0.82]
Epoch 2/4:  74%|███████▍  | 924/1250 [26:03<07:56,  1.46s/it, loss=0.641]
Epoch 2/4:  74%|███████▍  | 925/1250 [26:03<07:30,  1.39s/it, loss=0.641]
Epoch 2/4:  74%|███████▍  | 925/1250 [26:04<07:30,  1.39s/it, loss=0.548]
Epoch 2/4:  74%|███████▍  | 926/1250 [26:04<07:12,  1.34s/it, loss=0.548]
Epoch 2/4:  74%|███████▍  | 926/1250 [26:05<07:12,  1.34s/it, loss=0.455]
Epoch 2/4:  74%|███████▍  | 927/1250 [26:05<07:02,  1.31s/it, loss=0.455]
Epoch 2/4:  74%|███████▍  | 927/1250 [26:07<07:02,  1.31s/it, loss=0.956]
Epoch 2/4:  74%|███████▍  | 928/1250 [26:07<07:00,  1.31s/it, loss=0.956]
Epoch 2/4:  74%|███████▍  | 928/1250 [26:08<07:00,  1.31s/it, loss=0.538]
Epoch 2/4:  74%|███████▍  | 929/1250 [26:08<06:51,  1.28s/it, loss=0.538]
Epoch 2/4:  74%|███████▍  | 929/1250 [26:09<06:51,  1.28s/it, loss=0.805]
Epoch 2/4:  74%|███████▍  | 930/1250 [26:09<06:44,  1.26s/it, loss=0.805]
Epoch 2/4:  74%|███████▍  | 930/1250 [26:10<06:44,  1.26s/it, loss=0.934]
Epoch 2/4:  74%|███████▍  | 931/1250 [26:10<06:45,  1.27s/it, loss=0.934]
Epoch 2/4:  74%|███████▍  | 931/1250 [26:12<06:45,  1.27s/it, loss=0.779]
Epoch 2/4:  75%|███████▍  | 932/1250 [26:12<06:40,  1.26s/it, loss=0.779]
Epoch 2/4:  75%|███████▍  | 932/1250 [26:13<06:40,  1.26s/it, loss=0.553]
Epoch 2/4:  75%|███████▍  | 933/1250 [26:13<06:38,  1.26s/it, loss=0.553]
Epoch 2/4:  75%|███████▍  | 933/1250 [26:14<06:38,  1.26s/it, loss=0.696]
Epoch 2/4:  75%|███████▍  | 934/1250 [26:14<06:30,  1.24s/it, loss=0.696]
Epoch 2/4:  75%|███████▍  | 934/1250 [26:15<06:30,  1.24s/it, loss=0.591]
Epoch 2/4:  75%|███████▍  | 935/1250 [26:15<06:30,  1.24s/it, loss=0.591]
Epoch 2/4:  75%|███████▍  | 935/1250 [26:17<06:30,  1.24s/it, loss=0.551]
Epoch 2/4:  75%|███████▍  | 936/1250 [26:17<06:24,  1.23s/it, loss=0.551]
Epoch 2/4:  75%|███████▍  | 936/1250 [26:18<06:24,  1.23s/it, loss=0.787]
Epoch 2/4:  75%|███████▍  | 937/1250 [26:18<06:23,  1.23s/it, loss=0.787]
Epoch 2/4:  75%|███████▍  | 937/1250 [26:19<06:23,  1.23s/it, loss=0.935]
Epoch 2/4:  75%|███████▌  | 938/1250 [26:19<06:22,  1.23s/it, loss=0.935]
Epoch 2/4:  75%|███████▌  | 938/1250 [26:20<06:22,  1.23s/it, loss=0.633]
Epoch 2/4:  75%|███████▌  | 939/1250 [26:20<06:17,  1.21s/it, loss=0.633]
Epoch 2/4:  75%|███████▌  | 939/1250 [26:21<06:17,  1.21s/it, loss=0.691]
Epoch 2/4:  75%|███████▌  | 940/1250 [26:21<06:14,  1.21s/it, loss=0.691]
Epoch 2/4:  75%|███████▌  | 940/1250 [26:23<06:14,  1.21s/it, loss=0.819]
Epoch 2/4:  75%|███████▌  | 941/1250 [26:23<06:23,  1.24s/it, loss=0.819]
Epoch 2/4:  75%|███████▌  | 941/1250 [26:24<06:23,  1.24s/it, loss=0.43] 
Epoch 2/4:  75%|███████▌  | 942/1250 [26:24<06:21,  1.24s/it, loss=0.43]
Epoch 2/4:  75%|███████▌  | 942/1250 [26:25<06:21,  1.24s/it, loss=0.547]
Epoch 2/4:  75%|███████▌  | 943/1250 [26:25<06:20,  1.24s/it, loss=0.547]
Epoch 2/4:  75%|███████▌  | 943/1250 [26:26<06:20,  1.24s/it, loss=0.542]
Epoch 2/4:  76%|███████▌  | 944/1250 [26:26<06:12,  1.22s/it, loss=0.542]
Epoch 2/4:  76%|███████▌  | 944/1250 [26:28<06:12,  1.22s/it, loss=0.608]
Epoch 2/4:  76%|███████▌  | 945/1250 [26:28<06:07,  1.21s/it, loss=0.608]
Epoch 2/4:  76%|███████▌  | 945/1250 [26:29<06:07,  1.21s/it, loss=0.692]
Epoch 2/4:  76%|███████▌  | 946/1250 [26:29<06:05,  1.20s/it, loss=0.692]
Epoch 2/4:  76%|███████▌  | 946/1250 [26:30<06:05,  1.20s/it, loss=0.842]
Epoch 2/4:  76%|███████▌  | 947/1250 [26:30<05:57,  1.18s/it, loss=0.842]
Epoch 2/4:  76%|███████▌  | 947/1250 [26:31<05:57,  1.18s/it, loss=0.48] 
Epoch 2/4:  76%|███████▌  | 948/1250 [26:31<05:52,  1.17s/it, loss=0.48]
Epoch 2/4:  76%|███████▌  | 948/1250 [26:32<05:52,  1.17s/it, loss=0.503]
Epoch 2/4:  76%|███████▌  | 949/1250 [26:32<05:53,  1.17s/it, loss=0.503]
Epoch 2/4:  76%|███████▌  | 949/1250 [26:33<05:53,  1.17s/it, loss=0.717]
Epoch 2/4:  76%|███████▌  | 950/1250 [26:33<05:55,  1.19s/it, loss=0.717]
Epoch 2/4:  76%|███████▌  | 950/1250 [26:35<05:55,  1.19s/it, loss=0.757]
Epoch 2/4:  76%|███████▌  | 951/1250 [26:35<05:52,  1.18s/it, loss=0.757]
Epoch 2/4:  76%|███████▌  | 951/1250 [26:36<05:52,  1.18s/it, loss=0.516]
Epoch 2/4:  76%|███████▌  | 952/1250 [26:36<05:50,  1.18s/it, loss=0.516]
Epoch 2/4:  76%|███████▌  | 952/1250 [26:37<05:50,  1.18s/it, loss=0.71] 
Epoch 2/4:  76%|███████▌  | 953/1250 [26:37<05:47,  1.17s/it, loss=0.71]
Epoch 2/4:  76%|███████▌  | 953/1250 [26:38<05:47,  1.17s/it, loss=0.5] 
Epoch 2/4:  76%|███████▋  | 954/1250 [26:38<05:50,  1.18s/it, loss=0.5]
Epoch 2/4:  76%|███████▋  | 954/1250 [26:39<05:50,  1.18s/it, loss=0.892]
Epoch 2/4:  76%|███████▋  | 955/1250 [26:39<05:49,  1.18s/it, loss=0.892]
Epoch 2/4:  76%|███████▋  | 955/1250 [26:40<05:49,  1.18s/it, loss=0.657]
Epoch 2/4:  76%|███████▋  | 956/1250 [26:40<05:43,  1.17s/it, loss=0.657]
Epoch 2/4:  76%|███████▋  | 956/1250 [26:42<05:43,  1.17s/it, loss=1.07] 
Epoch 2/4:  77%|███████▋  | 957/1250 [26:42<05:47,  1.19s/it, loss=1.07]
Epoch 2/4:  77%|███████▋  | 957/1250 [26:43<05:47,  1.19s/it, loss=0.879]
Epoch 2/4:  77%|███████▋  | 958/1250 [26:43<05:52,  1.21s/it, loss=0.879]
Epoch 2/4:  77%|███████▋  | 958/1250 [26:44<05:52,  1.21s/it, loss=1.05] 
Epoch 2/4:  77%|███████▋  | 959/1250 [26:44<05:54,  1.22s/it, loss=1.05]
Epoch 2/4:  77%|███████▋  | 959/1250 [26:45<05:54,  1.22s/it, loss=0.614]
Epoch 2/4:  77%|███████▋  | 960/1250 [26:45<05:48,  1.20s/it, loss=0.614]
Epoch 2/4:  77%|███████▋  | 960/1250 [26:46<05:48,  1.20s/it, loss=0.612]
Epoch 2/4:  77%|███████▋  | 961/1250 [26:46<05:39,  1.17s/it, loss=0.612]
Epoch 2/4:  77%|███████▋  | 961/1250 [26:48<05:39,  1.17s/it, loss=0.665]
Epoch 2/4:  77%|███████▋  | 962/1250 [26:48<05:39,  1.18s/it, loss=0.665]
Epoch 2/4:  77%|███████▋  | 962/1250 [26:49<05:39,  1.18s/it, loss=0.657]
Epoch 2/4:  77%|███████▋  | 963/1250 [26:49<05:39,  1.18s/it, loss=0.657]
Epoch 2/4:  77%|███████▋  | 963/1250 [26:50<05:39,  1.18s/it, loss=0.784]
Epoch 2/4:  77%|███████▋  | 964/1250 [26:50<05:34,  1.17s/it, loss=0.784]
Epoch 2/4:  77%|███████▋  | 964/1250 [26:51<05:34,  1.17s/it, loss=0.618]
Epoch 2/4:  77%|███████▋  | 965/1250 [26:51<05:32,  1.17s/it, loss=0.618]
Epoch 2/4:  77%|███████▋  | 965/1250 [26:52<05:32,  1.17s/it, loss=0.711]
Epoch 2/4:  77%|███████▋  | 966/1250 [26:52<05:40,  1.20s/it, loss=0.711]
Epoch 2/4:  77%|███████▋  | 966/1250 [26:54<05:40,  1.20s/it, loss=0.737]
Epoch 2/4:  77%|███████▋  | 967/1250 [26:54<05:43,  1.21s/it, loss=0.737]
Epoch 2/4:  77%|███████▋  | 967/1250 [26:55<05:43,  1.21s/it, loss=0.519]
Epoch 2/4:  77%|███████▋  | 968/1250 [26:55<05:37,  1.20s/it, loss=0.519]
Epoch 2/4:  77%|███████▋  | 968/1250 [26:56<05:37,  1.20s/it, loss=0.933]
Epoch 2/4:  78%|███████▊  | 969/1250 [26:56<05:28,  1.17s/it, loss=0.933]
Epoch 2/4:  78%|███████▊  | 969/1250 [26:57<05:28,  1.17s/it, loss=0.626]
Epoch 2/4:  78%|███████▊  | 970/1250 [26:57<05:35,  1.20s/it, loss=0.626]
Epoch 2/4:  78%|███████▊  | 970/1250 [26:58<05:35,  1.20s/it, loss=0.511]
Epoch 2/4:  78%|███████▊  | 971/1250 [26:58<05:38,  1.21s/it, loss=0.511]
Epoch 2/4:  78%|███████▊  | 971/1250 [27:00<05:38,  1.21s/it, loss=0.939]
Epoch 2/4:  78%|███████▊  | 972/1250 [27:00<05:33,  1.20s/it, loss=0.939]
Epoch 2/4:  78%|███████▊  | 972/1250 [27:01<05:33,  1.20s/it, loss=0.772]
Epoch 2/4:  78%|███████▊  | 973/1250 [27:01<05:27,  1.18s/it, loss=0.772]
Epoch 2/4:  78%|███████▊  | 973/1250 [27:02<05:27,  1.18s/it, loss=0.757]
Epoch 2/4:  78%|███████▊  | 974/1250 [27:02<05:29,  1.19s/it, loss=0.757]
Epoch 2/4:  78%|███████▊  | 974/1250 [27:03<05:29,  1.19s/it, loss=0.645]
Epoch 2/4:  78%|███████▊  | 975/1250 [27:03<05:33,  1.21s/it, loss=0.645]
Epoch 2/4:  78%|███████▊  | 975/1250 [27:04<05:33,  1.21s/it, loss=0.511]
Epoch 2/4:  78%|███████▊  | 976/1250 [27:04<05:29,  1.20s/it, loss=0.511]
Epoch 2/4:  78%|███████▊  | 976/1250 [27:05<05:29,  1.20s/it, loss=0.695]
Epoch 2/4:  78%|███████▊  | 977/1250 [27:05<05:22,  1.18s/it, loss=0.695]
Epoch 2/4:  78%|███████▊  | 977/1250 [27:07<05:22,  1.18s/it, loss=0.538]
Epoch 2/4:  78%|███████▊  | 978/1250 [27:07<05:21,  1.18s/it, loss=0.538]
Epoch 2/4:  78%|███████▊  | 978/1250 [27:08<05:21,  1.18s/it, loss=0.85] 
Epoch 2/4:  78%|███████▊  | 979/1250 [27:08<05:24,  1.20s/it, loss=0.85]
Epoch 2/4:  78%|███████▊  | 979/1250 [27:09<05:24,  1.20s/it, loss=0.669]
Epoch 2/4:  78%|███████▊  | 980/1250 [27:09<05:20,  1.19s/it, loss=0.669]
Epoch 2/4:  78%|███████▊  | 980/1250 [27:10<05:20,  1.19s/it, loss=0.499]
Epoch 2/4:  78%|███████▊  | 981/1250 [27:10<05:14,  1.17s/it, loss=0.499]
Epoch 2/4:  78%|███████▊  | 981/1250 [27:11<05:14,  1.17s/it, loss=0.595]
Epoch 2/4:  79%|███████▊  | 982/1250 [27:11<05:16,  1.18s/it, loss=0.595]
Epoch 2/4:  79%|███████▊  | 982/1250 [27:13<05:16,  1.18s/it, loss=0.71] 
Epoch 2/4:  79%|███████▊  | 983/1250 [27:13<05:31,  1.24s/it, loss=0.71]
Epoch 2/4:  79%|███████▊  | 983/1250 [27:14<05:31,  1.24s/it, loss=0.863]
Epoch 2/4:  79%|███████▊  | 984/1250 [27:14<05:34,  1.26s/it, loss=0.863]
Epoch 2/4:  79%|███████▊  | 984/1250 [27:15<05:34,  1.26s/it, loss=0.745]
Epoch 2/4:  79%|███████▉  | 985/1250 [27:15<05:26,  1.23s/it, loss=0.745]
Epoch 2/4:  79%|███████▉  | 985/1250 [27:16<05:26,  1.23s/it, loss=0.612]
Epoch 2/4:  79%|███████▉  | 986/1250 [27:16<05:18,  1.21s/it, loss=0.612]
Epoch 2/4:  79%|███████▉  | 986/1250 [27:18<05:18,  1.21s/it, loss=0.901]
Epoch 2/4:  79%|███████▉  | 987/1250 [27:18<05:13,  1.19s/it, loss=0.901]
Epoch 2/4:  79%|███████▉  | 987/1250 [27:19<05:13,  1.19s/it, loss=0.683]
Epoch 2/4:  79%|███████▉  | 988/1250 [27:19<05:12,  1.19s/it, loss=0.683]
Epoch 2/4:  79%|███████▉  | 988/1250 [27:20<05:12,  1.19s/it, loss=0.616]
Epoch 2/4:  79%|███████▉  | 989/1250 [27:20<05:06,  1.17s/it, loss=0.616]
Epoch 2/4:  79%|███████▉  | 989/1250 [27:21<05:06,  1.17s/it, loss=0.78] 
Epoch 2/4:  79%|███████▉  | 990/1250 [27:21<05:01,  1.16s/it, loss=0.78]
Epoch 2/4:  79%|███████▉  | 990/1250 [27:22<05:01,  1.16s/it, loss=0.839]
Epoch 2/4:  79%|███████▉  | 991/1250 [27:22<05:07,  1.19s/it, loss=0.839]
Epoch 2/4:  79%|███████▉  | 991/1250 [27:24<05:07,  1.19s/it, loss=0.505]
Epoch 2/4:  79%|███████▉  | 992/1250 [27:24<05:11,  1.21s/it, loss=0.505]
Epoch 2/4:  79%|███████▉  | 992/1250 [27:25<05:11,  1.21s/it, loss=0.774]
Epoch 2/4:  79%|███████▉  | 993/1250 [27:25<05:07,  1.20s/it, loss=0.774]
Epoch 2/4:  79%|███████▉  | 993/1250 [27:26<05:07,  1.20s/it, loss=0.619]
Epoch 2/4:  80%|███████▉  | 994/1250 [27:26<05:01,  1.18s/it, loss=0.619]
Epoch 2/4:  80%|███████▉  | 994/1250 [27:27<05:01,  1.18s/it, loss=0.968]
Epoch 2/4:  80%|███████▉  | 995/1250 [27:27<05:00,  1.18s/it, loss=0.968]
Epoch 2/4:  80%|███████▉  | 995/1250 [27:28<05:00,  1.18s/it, loss=0.932]
Epoch 2/4:  80%|███████▉  | 996/1250 [27:28<05:02,  1.19s/it, loss=0.932]
Epoch 2/4:  80%|███████▉  | 996/1250 [27:29<05:02,  1.19s/it, loss=0.66] 
Epoch 2/4:  80%|███████▉  | 997/1250 [27:29<04:59,  1.19s/it, loss=0.66]
Epoch 2/4:  80%|███████▉  | 997/1250 [27:31<04:59,  1.19s/it, loss=0.597]
Epoch 2/4:  80%|███████▉  | 998/1250 [27:31<04:53,  1.16s/it, loss=0.597]
Epoch 2/4:  80%|███████▉  | 998/1250 [27:32<04:53,  1.16s/it, loss=0.387]
Epoch 2/4:  80%|███████▉  | 999/1250 [27:32<04:50,  1.16s/it, loss=0.387]
Epoch 2/4:  80%|███████▉  | 999/1250 [27:33<04:50,  1.16s/it, loss=0.727]
Epoch 2/4:  80%|████████  | 1000/1250 [27:33<04:59,  1.20s/it, loss=0.727]
Epoch 2/4:  80%|████████  | 1000/1250 [27:34<04:59,  1.20s/it, loss=0.716]
Epoch 2/4:  80%|████████  | 1001/1250 [27:34<05:01,  1.21s/it, loss=0.716]
Epoch 2/4:  80%|████████  | 1001/1250 [27:35<05:01,  1.21s/it, loss=0.858]
Epoch 2/4:  80%|████████  | 1002/1250 [27:35<04:53,  1.18s/it, loss=0.858]
Epoch 2/4:  80%|████████  | 1002/1250 [27:36<04:53,  1.18s/it, loss=0.696]
Epoch 2/4:  80%|████████  | 1003/1250 [27:36<04:46,  1.16s/it, loss=0.696]
Epoch 2/4:  80%|████████  | 1003/1250 [27:38<04:46,  1.16s/it, loss=0.975]
Epoch 2/4:  80%|████████  | 1004/1250 [27:38<04:52,  1.19s/it, loss=0.975]
Epoch 2/4:  80%|████████  | 1004/1250 [27:39<04:52,  1.19s/it, loss=0.643]
Epoch 2/4:  80%|████████  | 1005/1250 [27:39<05:00,  1.23s/it, loss=0.643]
Epoch 2/4:  80%|████████  | 1005/1250 [27:40<05:00,  1.23s/it, loss=0.667]
Epoch 2/4:  80%|████████  | 1006/1250 [27:40<04:55,  1.21s/it, loss=0.667]
Epoch 2/4:  80%|████████  | 1006/1250 [27:41<04:55,  1.21s/it, loss=0.938]
Epoch 2/4:  81%|████████  | 1007/1250 [27:41<04:46,  1.18s/it, loss=0.938]
Epoch 2/4:  81%|████████  | 1007/1250 [27:43<04:46,  1.18s/it, loss=0.73] 
Epoch 2/4:  81%|████████  | 1008/1250 [27:43<04:51,  1.20s/it, loss=0.73]
Epoch 2/4:  81%|████████  | 1008/1250 [27:44<04:51,  1.20s/it, loss=0.379]
Epoch 2/4:  81%|████████  | 1009/1250 [27:44<04:57,  1.23s/it, loss=0.379]
Epoch 2/4:  81%|████████  | 1009/1250 [27:45<04:57,  1.23s/it, loss=0.547]
Epoch 2/4:  81%|████████  | 1010/1250 [27:45<04:53,  1.22s/it, loss=0.547]
Epoch 2/4:  81%|████████  | 1010/1250 [27:46<04:53,  1.22s/it, loss=0.712]
Epoch 2/4:  81%|████████  | 1011/1250 [27:46<04:41,  1.18s/it, loss=0.712]
Epoch 2/4:  81%|████████  | 1011/1250 [27:47<04:41,  1.18s/it, loss=0.563]
Epoch 2/4:  81%|████████  | 1012/1250 [27:47<04:35,  1.16s/it, loss=0.563]
Epoch 2/4:  81%|████████  | 1012/1250 [27:48<04:35,  1.16s/it, loss=0.77] 
Epoch 2/4:  81%|████████  | 1013/1250 [27:48<04:35,  1.16s/it, loss=0.77]
Epoch 2/4:  81%|████████  | 1013/1250 [27:50<04:35,  1.16s/it, loss=0.633]
Epoch 2/4:  81%|████████  | 1014/1250 [27:50<04:34,  1.16s/it, loss=0.633]
Epoch 2/4:  81%|████████  | 1014/1250 [27:51<04:34,  1.16s/it, loss=0.824]
Epoch 2/4:  81%|████████  | 1015/1250 [27:51<04:32,  1.16s/it, loss=0.824]
Epoch 2/4:  81%|████████  | 1015/1250 [27:52<04:32,  1.16s/it, loss=0.547]
Epoch 2/4:  81%|████████▏ | 1016/1250 [27:52<04:32,  1.16s/it, loss=0.547]
Epoch 2/4:  81%|████████▏ | 1016/1250 [27:53<04:32,  1.16s/it, loss=0.949]
Epoch 2/4:  81%|████████▏ | 1017/1250 [27:53<04:35,  1.18s/it, loss=0.949]
Epoch 2/4:  81%|████████▏ | 1017/1250 [27:54<04:35,  1.18s/it, loss=0.964]
Epoch 2/4:  81%|████████▏ | 1018/1250 [27:54<04:35,  1.19s/it, loss=0.964]
Epoch 2/4:  81%|████████▏ | 1018/1250 [27:55<04:35,  1.19s/it, loss=0.862]
Epoch 2/4:  82%|████████▏ | 1019/1250 [27:55<04:31,  1.17s/it, loss=0.862]
Epoch 2/4:  82%|████████▏ | 1019/1250 [27:57<04:31,  1.17s/it, loss=0.719]
Epoch 2/4:  82%|████████▏ | 1020/1250 [27:57<04:25,  1.15s/it, loss=0.719]
Epoch 2/4:  82%|████████▏ | 1020/1250 [27:58<04:25,  1.15s/it, loss=0.955]
Epoch 2/4:  82%|████████▏ | 1021/1250 [27:58<04:27,  1.17s/it, loss=0.955]
Epoch 2/4:  82%|████████▏ | 1021/1250 [27:59<04:27,  1.17s/it, loss=0.603]
Epoch 2/4:  82%|████████▏ | 1022/1250 [27:59<04:30,  1.19s/it, loss=0.603]
Epoch 2/4:  82%|████████▏ | 1022/1250 [28:00<04:30,  1.19s/it, loss=0.528]
Epoch 2/4:  82%|████████▏ | 1023/1250 [28:00<04:30,  1.19s/it, loss=0.528]
Epoch 2/4:  82%|████████▏ | 1023/1250 [28:01<04:30,  1.19s/it, loss=0.578]
Epoch 2/4:  82%|████████▏ | 1024/1250 [28:01<04:27,  1.19s/it, loss=0.578]
Epoch 2/4:  82%|████████▏ | 1024/1250 [28:03<04:27,  1.19s/it, loss=0.634]
Epoch 2/4:  82%|████████▏ | 1025/1250 [28:03<04:30,  1.20s/it, loss=0.634]
Epoch 2/4:  82%|████████▏ | 1025/1250 [28:04<04:30,  1.20s/it, loss=0.542]
Epoch 2/4:  82%|████████▏ | 1026/1250 [28:04<04:35,  1.23s/it, loss=0.542]
Epoch 2/4:  82%|████████▏ | 1026/1250 [28:05<04:35,  1.23s/it, loss=0.799]
Epoch 2/4:  82%|████████▏ | 1027/1250 [28:05<04:35,  1.23s/it, loss=0.799]
Epoch 2/4:  82%|████████▏ | 1027/1250 [28:06<04:35,  1.23s/it, loss=0.687]
Epoch 2/4:  82%|████████▏ | 1028/1250 [28:06<04:33,  1.23s/it, loss=0.687]
Epoch 2/4:  82%|████████▏ | 1028/1250 [28:08<04:33,  1.23s/it, loss=0.53] 
Epoch 2/4:  82%|████████▏ | 1029/1250 [28:08<04:31,  1.23s/it, loss=0.53]
Epoch 2/4:  82%|████████▏ | 1029/1250 [28:09<04:31,  1.23s/it, loss=0.74]
Epoch 2/4:  82%|████████▏ | 1030/1250 [28:09<04:36,  1.26s/it, loss=0.74]
Epoch 2/4:  82%|████████▏ | 1030/1250 [28:10<04:36,  1.26s/it, loss=0.662]
Epoch 2/4:  82%|████████▏ | 1031/1250 [28:10<04:39,  1.28s/it, loss=0.662]
Epoch 2/4:  82%|████████▏ | 1031/1250 [28:11<04:39,  1.28s/it, loss=0.577]
Epoch 2/4:  83%|████████▎ | 1032/1250 [28:11<04:35,  1.26s/it, loss=0.577]
Epoch 2/4:  83%|████████▎ | 1032/1250 [28:13<04:35,  1.26s/it, loss=0.756]
Epoch 2/4:  83%|████████▎ | 1033/1250 [28:13<04:34,  1.26s/it, loss=0.756]
Epoch 2/4:  83%|████████▎ | 1033/1250 [28:14<04:34,  1.26s/it, loss=0.751]
Epoch 2/4:  83%|████████▎ | 1034/1250 [28:14<04:28,  1.24s/it, loss=0.751]
Epoch 2/4:  83%|████████▎ | 1034/1250 [28:15<04:28,  1.24s/it, loss=0.669]
Epoch 2/4:  83%|████████▎ | 1035/1250 [28:15<04:28,  1.25s/it, loss=0.669]
Epoch 2/4:  83%|████████▎ | 1035/1250 [28:16<04:28,  1.25s/it, loss=0.54] 
Epoch 2/4:  83%|████████▎ | 1036/1250 [28:16<04:22,  1.23s/it, loss=0.54]
Epoch 2/4:  83%|████████▎ | 1036/1250 [28:18<04:22,  1.23s/it, loss=0.964]
Epoch 2/4:  83%|████████▎ | 1037/1250 [28:18<04:18,  1.21s/it, loss=0.964]
Epoch 2/4:  83%|████████▎ | 1037/1250 [28:19<04:18,  1.21s/it, loss=0.479]
Epoch 2/4:  83%|████████▎ | 1038/1250 [28:19<04:18,  1.22s/it, loss=0.479]
Epoch 2/4:  83%|████████▎ | 1038/1250 [28:20<04:18,  1.22s/it, loss=0.537]
Epoch 2/4:  83%|████████▎ | 1039/1250 [28:20<04:17,  1.22s/it, loss=0.537]
Epoch 2/4:  83%|████████▎ | 1039/1250 [28:21<04:17,  1.22s/it, loss=0.756]
Epoch 2/4:  83%|████████▎ | 1040/1250 [28:21<04:10,  1.19s/it, loss=0.756]
Epoch 2/4:  83%|████████▎ | 1040/1250 [28:22<04:10,  1.19s/it, loss=0.758]
Epoch 2/4:  83%|████████▎ | 1041/1250 [28:22<04:02,  1.16s/it, loss=0.758]
Epoch 2/4:  83%|████████▎ | 1041/1250 [28:23<04:02,  1.16s/it, loss=1.03] 
Epoch 2/4:  83%|████████▎ | 1042/1250 [28:23<04:01,  1.16s/it, loss=1.03]
Epoch 2/4:  83%|████████▎ | 1042/1250 [28:25<04:01,  1.16s/it, loss=0.963]
Epoch 2/4:  83%|████████▎ | 1043/1250 [28:25<04:02,  1.17s/it, loss=0.963]
Epoch 2/4:  83%|████████▎ | 1043/1250 [28:26<04:02,  1.17s/it, loss=0.705]
Epoch 2/4:  84%|████████▎ | 1044/1250 [28:26<03:59,  1.16s/it, loss=0.705]
Epoch 2/4:  84%|████████▎ | 1044/1250 [28:27<03:59,  1.16s/it, loss=0.76] 
Epoch 2/4:  84%|████████▎ | 1045/1250 [28:27<03:56,  1.15s/it, loss=0.76]
Epoch 2/4:  84%|████████▎ | 1045/1250 [28:28<03:56,  1.15s/it, loss=0.651]
Epoch 2/4:  84%|████████▎ | 1046/1250 [28:28<04:02,  1.19s/it, loss=0.651]
Epoch 2/4:  84%|████████▎ | 1046/1250 [28:29<04:02,  1.19s/it, loss=0.968]
Epoch 2/4:  84%|████████▍ | 1047/1250 [28:29<04:08,  1.22s/it, loss=0.968]
Epoch 2/4:  84%|████████▍ | 1047/1250 [28:31<04:08,  1.22s/it, loss=0.751]
Epoch 2/4:  84%|████████▍ | 1048/1250 [28:31<04:02,  1.20s/it, loss=0.751]
Epoch 2/4:  84%|████████▍ | 1048/1250 [28:32<04:02,  1.20s/it, loss=0.559]
Epoch 2/4:  84%|████████▍ | 1049/1250 [28:32<03:53,  1.16s/it, loss=0.559]
Epoch 2/4:  84%|████████▍ | 1049/1250 [28:33<03:53,  1.16s/it, loss=0.775]
Epoch 2/4:  84%|████████▍ | 1050/1250 [28:33<03:52,  1.16s/it, loss=0.775]
Epoch 2/4:  84%|████████▍ | 1050/1250 [28:34<03:52,  1.16s/it, loss=0.772]
Epoch 2/4:  84%|████████▍ | 1051/1250 [28:34<03:57,  1.19s/it, loss=0.772]
Epoch 2/4:  84%|████████▍ | 1051/1250 [28:35<03:57,  1.19s/it, loss=0.827]
Epoch 2/4:  84%|████████▍ | 1052/1250 [28:35<03:58,  1.21s/it, loss=0.827]
Epoch 2/4:  84%|████████▍ | 1052/1250 [28:37<03:58,  1.21s/it, loss=0.605]
Epoch 2/4:  84%|████████▍ | 1053/1250 [28:37<03:59,  1.21s/it, loss=0.605]
Epoch 2/4:  84%|████████▍ | 1053/1250 [28:38<03:59,  1.21s/it, loss=0.655]
Epoch 2/4:  84%|████████▍ | 1054/1250 [28:38<04:04,  1.25s/it, loss=0.655]
Epoch 2/4:  84%|████████▍ | 1054/1250 [28:39<04:04,  1.25s/it, loss=0.733]
Epoch 2/4:  84%|████████▍ | 1055/1250 [28:39<04:02,  1.24s/it, loss=0.733]
Epoch 2/4:  84%|████████▍ | 1055/1250 [28:40<04:02,  1.24s/it, loss=0.88] 
Epoch 2/4:  84%|████████▍ | 1056/1250 [28:40<03:59,  1.23s/it, loss=0.88]
Epoch 2/4:  84%|████████▍ | 1056/1250 [28:42<03:59,  1.23s/it, loss=0.809]
Epoch 2/4:  85%|████████▍ | 1057/1250 [28:42<03:58,  1.24s/it, loss=0.809]
Epoch 2/4:  85%|████████▍ | 1057/1250 [28:43<03:58,  1.24s/it, loss=0.929]
Epoch 2/4:  85%|████████▍ | 1058/1250 [28:43<03:52,  1.21s/it, loss=0.929]
Epoch 2/4:  85%|████████▍ | 1058/1250 [28:44<03:52,  1.21s/it, loss=0.745]
Epoch 2/4:  85%|████████▍ | 1059/1250 [28:44<03:52,  1.21s/it, loss=0.745]
Epoch 2/4:  85%|████████▍ | 1059/1250 [28:45<03:52,  1.21s/it, loss=0.614]
Epoch 2/4:  85%|████████▍ | 1060/1250 [28:45<03:49,  1.21s/it, loss=0.614]
Epoch 2/4:  85%|████████▍ | 1060/1250 [28:46<03:49,  1.21s/it, loss=0.84] 
Epoch 2/4:  85%|████████▍ | 1061/1250 [28:46<03:42,  1.18s/it, loss=0.84]
Epoch 2/4:  85%|████████▍ | 1061/1250 [28:47<03:42,  1.18s/it, loss=0.638]
Epoch 2/4:  85%|████████▍ | 1062/1250 [28:47<03:39,  1.17s/it, loss=0.638]
Epoch 2/4:  85%|████████▍ | 1062/1250 [28:49<03:39,  1.17s/it, loss=0.675]
Epoch 2/4:  85%|████████▌ | 1063/1250 [28:49<03:42,  1.19s/it, loss=0.675]
Epoch 2/4:  85%|████████▌ | 1063/1250 [28:50<03:42,  1.19s/it, loss=0.895]
Epoch 2/4:  85%|████████▌ | 1064/1250 [28:50<03:42,  1.20s/it, loss=0.895]
Epoch 2/4:  85%|████████▌ | 1064/1250 [28:51<03:42,  1.20s/it, loss=0.519]
Epoch 2/4:  85%|████████▌ | 1065/1250 [28:51<03:38,  1.18s/it, loss=0.519]
Epoch 2/4:  85%|████████▌ | 1065/1250 [28:52<03:38,  1.18s/it, loss=0.676]
Epoch 2/4:  85%|████████▌ | 1066/1250 [28:52<03:33,  1.16s/it, loss=0.676]
Epoch 2/4:  85%|████████▌ | 1066/1250 [28:53<03:33,  1.16s/it, loss=0.697]
Epoch 2/4:  85%|████████▌ | 1067/1250 [28:53<03:33,  1.17s/it, loss=0.697]
Epoch 2/4:  85%|████████▌ | 1067/1250 [28:54<03:33,  1.17s/it, loss=0.977]
Epoch 2/4:  85%|████████▌ | 1068/1250 [28:54<03:35,  1.18s/it, loss=0.977]
Epoch 2/4:  85%|████████▌ | 1068/1250 [28:56<03:35,  1.18s/it, loss=0.52] 
Epoch 2/4:  86%|████████▌ | 1069/1250 [28:56<03:34,  1.19s/it, loss=0.52]
Epoch 2/4:  86%|████████▌ | 1069/1250 [28:57<03:34,  1.19s/it, loss=0.6] 
Epoch 2/4:  86%|████████▌ | 1070/1250 [28:57<03:30,  1.17s/it, loss=0.6]
Epoch 2/4:  86%|████████▌ | 1070/1250 [28:58<03:30,  1.17s/it, loss=0.759]
Epoch 2/4:  86%|████████▌ | 1071/1250 [28:58<03:28,  1.17s/it, loss=0.759]
Epoch 2/4:  86%|████████▌ | 1071/1250 [28:59<03:28,  1.17s/it, loss=0.588]
Epoch 2/4:  86%|████████▌ | 1072/1250 [28:59<03:28,  1.17s/it, loss=0.588]
Epoch 2/4:  86%|████████▌ | 1072/1250 [29:00<03:28,  1.17s/it, loss=0.708]
Epoch 2/4:  86%|████████▌ | 1073/1250 [29:00<03:26,  1.16s/it, loss=0.708]
Epoch 2/4:  86%|████████▌ | 1073/1250 [29:01<03:26,  1.16s/it, loss=0.545]
Epoch 2/4:  86%|████████▌ | 1074/1250 [29:01<03:23,  1.15s/it, loss=0.545]
Epoch 2/4:  86%|████████▌ | 1074/1250 [29:03<03:23,  1.15s/it, loss=0.793]
Epoch 2/4:  86%|████████▌ | 1075/1250 [29:03<03:23,  1.17s/it, loss=0.793]
Epoch 2/4:  86%|████████▌ | 1075/1250 [29:04<03:23,  1.17s/it, loss=0.716]
Epoch 2/4:  86%|████████▌ | 1076/1250 [29:04<03:28,  1.20s/it, loss=0.716]
Epoch 2/4:  86%|████████▌ | 1076/1250 [29:05<03:28,  1.20s/it, loss=0.564]
Epoch 2/4:  86%|████████▌ | 1077/1250 [29:05<03:28,  1.20s/it, loss=0.564]
Epoch 2/4:  86%|████████▌ | 1077/1250 [29:06<03:28,  1.20s/it, loss=0.637]
Epoch 2/4:  86%|████████▌ | 1078/1250 [29:06<03:23,  1.18s/it, loss=0.637]
Epoch 2/4:  86%|████████▌ | 1078/1250 [29:07<03:23,  1.18s/it, loss=0.611]
Epoch 2/4:  86%|████████▋ | 1079/1250 [29:07<03:21,  1.18s/it, loss=0.611]
Epoch 2/4:  86%|████████▋ | 1079/1250 [29:09<03:21,  1.18s/it, loss=0.694]
Epoch 2/4:  86%|████████▋ | 1080/1250 [29:09<03:23,  1.20s/it, loss=0.694]
Epoch 2/4:  86%|████████▋ | 1080/1250 [29:10<03:23,  1.20s/it, loss=0.85] 
Epoch 2/4:  86%|████████▋ | 1081/1250 [29:10<03:22,  1.20s/it, loss=0.85]
Epoch 2/4:  86%|████████▋ | 1081/1250 [29:11<03:22,  1.20s/it, loss=0.76]
Epoch 2/4:  87%|████████▋ | 1082/1250 [29:11<03:17,  1.18s/it, loss=0.76]
Epoch 2/4:  87%|████████▋ | 1082/1250 [29:12<03:17,  1.18s/it, loss=0.791]
Epoch 2/4:  87%|████████▋ | 1083/1250 [29:12<03:14,  1.17s/it, loss=0.791]
Epoch 2/4:  87%|████████▋ | 1083/1250 [29:13<03:14,  1.17s/it, loss=0.61] 
Epoch 2/4:  87%|████████▋ | 1084/1250 [29:13<03:17,  1.19s/it, loss=0.61]
Epoch 2/4:  87%|████████▋ | 1084/1250 [29:15<03:17,  1.19s/it, loss=0.613]
Epoch 2/4:  87%|████████▋ | 1085/1250 [29:15<03:17,  1.19s/it, loss=0.613]
Epoch 2/4:  87%|████████▋ | 1085/1250 [29:16<03:17,  1.19s/it, loss=0.873]
Epoch 2/4:  87%|████████▋ | 1086/1250 [29:16<03:12,  1.18s/it, loss=0.873]
Epoch 2/4:  87%|████████▋ | 1086/1250 [29:17<03:12,  1.18s/it, loss=0.521]
Epoch 2/4:  87%|████████▋ | 1087/1250 [29:17<03:08,  1.16s/it, loss=0.521]
Epoch 2/4:  87%|████████▋ | 1087/1250 [29:18<03:08,  1.16s/it, loss=0.639]
Epoch 2/4:  87%|████████▋ | 1088/1250 [29:18<03:09,  1.17s/it, loss=0.639]
Epoch 2/4:  87%|████████▋ | 1088/1250 [29:19<03:09,  1.17s/it, loss=0.863]
Epoch 2/4:  87%|████████▋ | 1089/1250 [29:19<03:12,  1.20s/it, loss=0.863]
Epoch 2/4:  87%|████████▋ | 1089/1250 [29:20<03:12,  1.20s/it, loss=0.712]
Epoch 2/4:  87%|████████▋ | 1090/1250 [29:20<03:10,  1.19s/it, loss=0.712]
Epoch 2/4:  87%|████████▋ | 1090/1250 [29:22<03:10,  1.19s/it, loss=0.678]
Epoch 2/4:  87%|████████▋ | 1091/1250 [29:22<03:04,  1.16s/it, loss=0.678]
Epoch 2/4:  87%|████████▋ | 1091/1250 [29:23<03:04,  1.16s/it, loss=1.06] 
Epoch 2/4:  87%|████████▋ | 1092/1250 [29:23<03:01,  1.15s/it, loss=1.06]
Epoch 2/4:  87%|████████▋ | 1092/1250 [29:24<03:01,  1.15s/it, loss=0.6] 
Epoch 2/4:  87%|████████▋ | 1093/1250 [29:24<03:07,  1.20s/it, loss=0.6]
Epoch 2/4:  87%|████████▋ | 1093/1250 [29:25<03:07,  1.20s/it, loss=0.572]
Epoch 2/4:  88%|████████▊ | 1094/1250 [29:25<03:07,  1.20s/it, loss=0.572]
Epoch 2/4:  88%|████████▊ | 1094/1250 [29:26<03:07,  1.20s/it, loss=0.552]
Epoch 2/4:  88%|████████▊ | 1095/1250 [29:26<03:02,  1.18s/it, loss=0.552]
Epoch 2/4:  88%|████████▊ | 1095/1250 [29:27<03:02,  1.18s/it, loss=1.63] 
Epoch 2/4:  88%|████████▊ | 1096/1250 [29:27<02:57,  1.16s/it, loss=1.63]
Epoch 2/4:  88%|████████▊ | 1096/1250 [29:29<02:57,  1.16s/it, loss=0.88]
Epoch 2/4:  88%|████████▊ | 1097/1250 [29:29<02:57,  1.16s/it, loss=0.88]
Epoch 2/4:  88%|████████▊ | 1097/1250 [29:30<02:57,  1.16s/it, loss=0.703]
Epoch 2/4:  88%|████████▊ | 1098/1250 [29:30<02:58,  1.17s/it, loss=0.703]
Epoch 2/4:  88%|████████▊ | 1098/1250 [29:31<02:58,  1.17s/it, loss=0.722]
Epoch 2/4:  88%|████████▊ | 1099/1250 [29:31<02:56,  1.17s/it, loss=0.722]
Epoch 2/4:  88%|████████▊ | 1099/1250 [29:32<02:56,  1.17s/it, loss=1.13] 
Epoch 2/4:  88%|████████▊ | 1100/1250 [29:32<02:52,  1.15s/it, loss=1.13]
Epoch 2/4:  88%|████████▊ | 1100/1250 [29:33<02:52,  1.15s/it, loss=0.364]
Epoch 2/4:  88%|████████▊ | 1101/1250 [29:33<02:53,  1.16s/it, loss=0.364]
Epoch 2/4:  88%|████████▊ | 1101/1250 [29:35<02:53,  1.16s/it, loss=0.538]
Epoch 2/4:  88%|████████▊ | 1102/1250 [29:35<02:59,  1.21s/it, loss=0.538]
Epoch 2/4:  88%|████████▊ | 1102/1250 [29:36<02:59,  1.21s/it, loss=0.851]
Epoch 2/4:  88%|████████▊ | 1103/1250 [29:36<02:57,  1.21s/it, loss=0.851]
Epoch 2/4:  88%|████████▊ | 1103/1250 [29:37<02:57,  1.21s/it, loss=0.781]
Epoch 2/4:  88%|████████▊ | 1104/1250 [29:37<02:51,  1.17s/it, loss=0.781]
Epoch 2/4:  88%|████████▊ | 1104/1250 [29:38<02:51,  1.17s/it, loss=0.643]
Epoch 2/4:  88%|████████▊ | 1105/1250 [29:38<02:49,  1.17s/it, loss=0.643]
Epoch 2/4:  88%|████████▊ | 1105/1250 [29:39<02:49,  1.17s/it, loss=0.832]
Epoch 2/4:  88%|████████▊ | 1106/1250 [29:39<02:49,  1.18s/it, loss=0.832]
Epoch 2/4:  88%|████████▊ | 1106/1250 [29:40<02:49,  1.18s/it, loss=0.67] 
Epoch 2/4:  89%|████████▊ | 1107/1250 [29:40<02:48,  1.18s/it, loss=0.67]
Epoch 2/4:  89%|████████▊ | 1107/1250 [29:41<02:48,  1.18s/it, loss=0.694]
Epoch 2/4:  89%|████████▊ | 1108/1250 [29:41<02:45,  1.16s/it, loss=0.694]
Epoch 2/4:  89%|████████▊ | 1108/1250 [29:43<02:45,  1.16s/it, loss=1.03] 
Epoch 2/4:  89%|████████▊ | 1109/1250 [29:43<02:44,  1.17s/it, loss=1.03]
Epoch 2/4:  89%|████████▊ | 1109/1250 [29:44<02:44,  1.17s/it, loss=0.472]
Epoch 2/4:  89%|████████▉ | 1110/1250 [29:44<02:47,  1.20s/it, loss=0.472]
Epoch 2/4:  89%|████████▉ | 1110/1250 [29:45<02:47,  1.20s/it, loss=0.604]
Epoch 2/4:  89%|████████▉ | 1111/1250 [29:45<02:48,  1.22s/it, loss=0.604]
Epoch 2/4:  89%|████████▉ | 1111/1250 [29:46<02:48,  1.22s/it, loss=0.727]
Epoch 2/4:  89%|████████▉ | 1112/1250 [29:46<02:45,  1.20s/it, loss=0.727]
Epoch 2/4:  89%|████████▉ | 1112/1250 [29:48<02:45,  1.20s/it, loss=0.735]
Epoch 2/4:  89%|████████▉ | 1113/1250 [29:48<02:41,  1.18s/it, loss=0.735]
Epoch 2/4:  89%|████████▉ | 1113/1250 [29:49<02:41,  1.18s/it, loss=1.07] 
Epoch 2/4:  89%|████████▉ | 1114/1250 [29:49<02:39,  1.17s/it, loss=1.07]
Epoch 2/4:  89%|████████▉ | 1114/1250 [29:50<02:39,  1.17s/it, loss=0.61]
Epoch 2/4:  89%|████████▉ | 1115/1250 [29:50<02:37,  1.17s/it, loss=0.61]
Epoch 2/4:  89%|████████▉ | 1115/1250 [29:51<02:37,  1.17s/it, loss=0.502]
Epoch 2/4:  89%|████████▉ | 1116/1250 [29:51<02:35,  1.16s/it, loss=0.502]
Epoch 2/4:  89%|████████▉ | 1116/1250 [29:52<02:35,  1.16s/it, loss=0.59] 
Epoch 2/4:  89%|████████▉ | 1117/1250 [29:52<02:32,  1.15s/it, loss=0.59]
Epoch 2/4:  89%|████████▉ | 1117/1250 [29:53<02:32,  1.15s/it, loss=0.855]
Epoch 2/4:  89%|████████▉ | 1118/1250 [29:53<02:32,  1.16s/it, loss=0.855]
Epoch 2/4:  89%|████████▉ | 1118/1250 [29:55<02:32,  1.16s/it, loss=0.684]
Epoch 2/4:  90%|████████▉ | 1119/1250 [29:55<02:36,  1.19s/it, loss=0.684]
Epoch 2/4:  90%|████████▉ | 1119/1250 [29:56<02:36,  1.19s/it, loss=0.781]
Epoch 2/4:  90%|████████▉ | 1120/1250 [29:56<02:35,  1.19s/it, loss=0.781]
Epoch 2/4:  90%|████████▉ | 1120/1250 [29:57<02:35,  1.19s/it, loss=0.801]
Epoch 2/4:  90%|████████▉ | 1121/1250 [29:57<02:30,  1.16s/it, loss=0.801]
Epoch 2/4:  90%|████████▉ | 1121/1250 [29:58<02:30,  1.16s/it, loss=0.855]
Epoch 2/4:  90%|████████▉ | 1122/1250 [29:58<02:26,  1.14s/it, loss=0.855]
Epoch 2/4:  90%|████████▉ | 1122/1250 [29:59<02:26,  1.14s/it, loss=0.612]
Epoch 2/4:  90%|████████▉ | 1123/1250 [29:59<02:27,  1.16s/it, loss=0.612]
Epoch 2/4:  90%|████████▉ | 1123/1250 [30:00<02:27,  1.16s/it, loss=0.578]
Epoch 2/4:  90%|████████▉ | 1124/1250 [30:00<02:28,  1.18s/it, loss=0.578]
Epoch 2/4:  90%|████████▉ | 1124/1250 [30:01<02:28,  1.18s/it, loss=1.12] 
Epoch 2/4:  90%|█████████ | 1125/1250 [30:01<02:25,  1.17s/it, loss=1.12]
Epoch 2/4:  90%|█████████ | 1125/1250 [30:03<02:25,  1.17s/it, loss=0.764]
Epoch 2/4:  90%|█████████ | 1126/1250 [30:03<02:22,  1.15s/it, loss=0.764]
Epoch 2/4:  90%|█████████ | 1126/1250 [30:04<02:22,  1.15s/it, loss=0.681]
Epoch 2/4:  90%|█████████ | 1127/1250 [30:04<02:23,  1.17s/it, loss=0.681]
Epoch 2/4:  90%|█████████ | 1127/1250 [30:05<02:23,  1.17s/it, loss=0.754]
Epoch 2/4:  90%|█████████ | 1128/1250 [30:05<02:24,  1.18s/it, loss=0.754]
Epoch 2/4:  90%|█████████ | 1128/1250 [30:06<02:24,  1.18s/it, loss=0.793]
Epoch 2/4:  90%|█████████ | 1129/1250 [30:06<02:22,  1.18s/it, loss=0.793]
Epoch 2/4:  90%|█████████ | 1129/1250 [30:07<02:22,  1.18s/it, loss=0.532]
Epoch 2/4:  90%|█████████ | 1130/1250 [30:07<02:18,  1.16s/it, loss=0.532]
Epoch 2/4:  90%|█████████ | 1130/1250 [30:08<02:18,  1.16s/it, loss=0.545]
Epoch 2/4:  90%|█████████ | 1131/1250 [30:08<02:18,  1.16s/it, loss=0.545]
Epoch 2/4:  90%|█████████ | 1131/1250 [30:10<02:18,  1.16s/it, loss=0.531]
Epoch 2/4:  91%|█████████ | 1132/1250 [30:10<02:20,  1.19s/it, loss=0.531]
Epoch 2/4:  91%|█████████ | 1132/1250 [30:11<02:20,  1.19s/it, loss=0.702]
Epoch 2/4:  91%|█████████ | 1133/1250 [30:11<02:18,  1.18s/it, loss=0.702]
Epoch 2/4:  91%|█████████ | 1133/1250 [30:12<02:18,  1.18s/it, loss=0.719]
Epoch 2/4:  91%|█████████ | 1134/1250 [30:12<02:15,  1.17s/it, loss=0.719]
Epoch 2/4:  91%|█████████ | 1134/1250 [30:13<02:15,  1.17s/it, loss=0.589]
Epoch 2/4:  91%|█████████ | 1135/1250 [30:13<02:15,  1.18s/it, loss=0.589]
Epoch 2/4:  91%|█████████ | 1135/1250 [30:14<02:15,  1.18s/it, loss=0.73] 
Epoch 2/4:  91%|█████████ | 1136/1250 [30:14<02:14,  1.18s/it, loss=0.73]
Epoch 2/4:  91%|█████████ | 1136/1250 [30:16<02:14,  1.18s/it, loss=0.897]
Epoch 2/4:  91%|█████████ | 1137/1250 [30:16<02:14,  1.19s/it, loss=0.897]
Epoch 2/4:  91%|█████████ | 1137/1250 [30:17<02:14,  1.19s/it, loss=1.99] 
Epoch 2/4:  91%|█████████ | 1138/1250 [30:17<02:11,  1.18s/it, loss=1.99]
Epoch 2/4:  91%|█████████ | 1138/1250 [30:18<02:11,  1.18s/it, loss=0.523]
Epoch 2/4:  91%|█████████ | 1139/1250 [30:18<02:07,  1.15s/it, loss=0.523]
Epoch 2/4:  91%|█████████ | 1139/1250 [30:19<02:07,  1.15s/it, loss=0.578]
Epoch 2/4:  91%|█████████ | 1140/1250 [30:19<02:06,  1.15s/it, loss=0.578]
Epoch 2/4:  91%|█████████ | 1140/1250 [30:20<02:06,  1.15s/it, loss=0.681]
Epoch 2/4:  91%|█████████▏| 1141/1250 [30:20<02:06,  1.16s/it, loss=0.681]
Epoch 2/4:  91%|█████████▏| 1141/1250 [30:21<02:06,  1.16s/it, loss=0.831]
Epoch 2/4:  91%|█████████▏| 1142/1250 [30:21<02:00,  1.12s/it, loss=0.831]
Epoch 2/4:  91%|█████████▏| 1142/1250 [30:22<02:00,  1.12s/it, loss=0.731]
Epoch 2/4:  91%|█████████▏| 1143/1250 [30:22<01:57,  1.09s/it, loss=0.731]
Epoch 2/4:  91%|█████████▏| 1143/1250 [30:23<01:57,  1.09s/it, loss=0.516]
Epoch 2/4:  92%|█████████▏| 1144/1250 [30:23<02:00,  1.13s/it, loss=0.516]
Epoch 2/4:  92%|█████████▏| 1144/1250 [30:25<02:00,  1.13s/it, loss=0.725]
Epoch 2/4:  92%|█████████▏| 1145/1250 [30:25<02:04,  1.18s/it, loss=0.725]
Epoch 2/4:  92%|█████████▏| 1145/1250 [30:26<02:04,  1.18s/it, loss=0.822]
Epoch 2/4:  92%|█████████▏| 1146/1250 [30:26<02:03,  1.19s/it, loss=0.822]
Epoch 2/4:  92%|█████████▏| 1146/1250 [30:27<02:03,  1.19s/it, loss=0.548]
Epoch 2/4:  92%|█████████▏| 1147/1250 [30:27<01:59,  1.16s/it, loss=0.548]
Epoch 2/4:  92%|█████████▏| 1147/1250 [30:28<01:59,  1.16s/it, loss=0.545]
Epoch 2/4:  92%|█████████▏| 1148/1250 [30:28<01:53,  1.11s/it, loss=0.545]
Epoch 2/4:  92%|█████████▏| 1148/1250 [30:29<01:53,  1.11s/it, loss=0.589]
Epoch 2/4:  92%|█████████▏| 1149/1250 [30:29<01:52,  1.11s/it, loss=0.589]
Epoch 2/4:  92%|█████████▏| 1149/1250 [30:30<01:52,  1.11s/it, loss=0.737]
Epoch 2/4:  92%|█████████▏| 1150/1250 [30:30<01:55,  1.16s/it, loss=0.737]
Epoch 2/4:  92%|█████████▏| 1150/1250 [30:32<01:55,  1.16s/it, loss=0.472]
Epoch 2/4:  92%|█████████▏| 1151/1250 [30:32<01:55,  1.17s/it, loss=0.472]
Epoch 2/4:  92%|█████████▏| 1151/1250 [30:33<01:55,  1.17s/it, loss=0.777]
Epoch 2/4:  92%|█████████▏| 1152/1250 [30:33<01:52,  1.15s/it, loss=0.777]
Epoch 2/4:  92%|█████████▏| 1152/1250 [30:34<01:52,  1.15s/it, loss=0.799]
Epoch 2/4:  92%|█████████▏| 1153/1250 [30:34<01:51,  1.15s/it, loss=0.799]
Epoch 2/4:  92%|█████████▏| 1153/1250 [30:35<01:51,  1.15s/it, loss=0.601]
Epoch 2/4:  92%|█████████▏| 1154/1250 [30:35<01:52,  1.17s/it, loss=0.601]
Epoch 2/4:  92%|█████████▏| 1154/1250 [30:36<01:52,  1.17s/it, loss=0.633]
Epoch 2/4:  92%|█████████▏| 1155/1250 [30:36<01:50,  1.17s/it, loss=0.633]
Epoch 2/4:  92%|█████████▏| 1155/1250 [30:37<01:50,  1.17s/it, loss=0.761]
Epoch 2/4:  92%|█████████▏| 1156/1250 [30:37<01:47,  1.15s/it, loss=0.761]
Epoch 2/4:  92%|█████████▏| 1156/1250 [30:39<01:47,  1.15s/it, loss=0.817]
Epoch 2/4:  93%|█████████▎| 1157/1250 [30:39<01:47,  1.15s/it, loss=0.817]
Epoch 2/4:  93%|█████████▎| 1157/1250 [30:40<01:47,  1.15s/it, loss=0.668]
Epoch 2/4:  93%|█████████▎| 1158/1250 [30:40<01:47,  1.17s/it, loss=0.668]
Epoch 2/4:  93%|█████████▎| 1158/1250 [30:41<01:47,  1.17s/it, loss=0.707]
Epoch 2/4:  93%|█████████▎| 1159/1250 [30:41<01:46,  1.18s/it, loss=0.707]
Epoch 2/4:  93%|█████████▎| 1159/1250 [30:42<01:46,  1.18s/it, loss=0.6]  
Epoch 2/4:  93%|█████████▎| 1160/1250 [30:42<01:45,  1.18s/it, loss=0.6]
Epoch 2/4:  93%|█████████▎| 1160/1250 [30:43<01:45,  1.18s/it, loss=0.81]
Epoch 2/4:  93%|█████████▎| 1161/1250 [30:43<01:46,  1.20s/it, loss=0.81]
Epoch 2/4:  93%|█████████▎| 1161/1250 [30:45<01:46,  1.20s/it, loss=0.54]
Epoch 2/4:  93%|█████████▎| 1162/1250 [30:45<01:44,  1.19s/it, loss=0.54]
Epoch 2/4:  93%|█████████▎| 1162/1250 [30:46<01:44,  1.19s/it, loss=0.764]
Epoch 2/4:  93%|█████████▎| 1163/1250 [30:46<01:43,  1.19s/it, loss=0.764]
Epoch 2/4:  93%|█████████▎| 1163/1250 [30:47<01:43,  1.19s/it, loss=0.749]
Epoch 2/4:  93%|█████████▎| 1164/1250 [30:47<01:41,  1.18s/it, loss=0.749]
Epoch 2/4:  93%|█████████▎| 1164/1250 [30:48<01:41,  1.18s/it, loss=0.55] 
Epoch 2/4:  93%|█████████▎| 1165/1250 [30:48<01:38,  1.15s/it, loss=0.55]
Epoch 2/4:  93%|█████████▎| 1165/1250 [30:49<01:38,  1.15s/it, loss=0.753]
Epoch 2/4:  93%|█████████▎| 1166/1250 [30:49<01:36,  1.15s/it, loss=0.753]
Epoch 2/4:  93%|█████████▎| 1166/1250 [30:50<01:36,  1.15s/it, loss=0.805]
Epoch 2/4:  93%|█████████▎| 1167/1250 [30:50<01:35,  1.15s/it, loss=0.805]
Epoch 2/4:  93%|█████████▎| 1167/1250 [30:51<01:35,  1.15s/it, loss=0.708]
Epoch 2/4:  93%|█████████▎| 1168/1250 [30:51<01:34,  1.15s/it, loss=0.708]
Epoch 2/4:  93%|█████████▎| 1168/1250 [30:53<01:34,  1.15s/it, loss=0.476]
Epoch 2/4:  94%|█████████▎| 1169/1250 [30:53<01:32,  1.14s/it, loss=0.476]
Epoch 2/4:  94%|█████████▎| 1169/1250 [30:54<01:32,  1.14s/it, loss=0.657]
Epoch 2/4:  94%|█████████▎| 1170/1250 [30:54<01:31,  1.15s/it, loss=0.657]
Epoch 2/4:  94%|█████████▎| 1170/1250 [30:55<01:31,  1.15s/it, loss=0.739]
Epoch 2/4:  94%|█████████▎| 1171/1250 [30:55<01:32,  1.16s/it, loss=0.739]
Epoch 2/4:  94%|█████████▎| 1171/1250 [30:56<01:32,  1.16s/it, loss=1.12] 
Epoch 2/4:  94%|█████████▍| 1172/1250 [30:56<01:31,  1.17s/it, loss=1.12]
Epoch 2/4:  94%|█████████▍| 1172/1250 [30:57<01:31,  1.17s/it, loss=0.652]
Epoch 2/4:  94%|█████████▍| 1173/1250 [30:57<01:29,  1.16s/it, loss=0.652]
Epoch 2/4:  94%|█████████▍| 1173/1250 [30:58<01:29,  1.16s/it, loss=0.831]
Epoch 2/4:  94%|█████████▍| 1174/1250 [30:58<01:27,  1.15s/it, loss=0.831]
Epoch 2/4:  94%|█████████▍| 1174/1250 [31:00<01:27,  1.15s/it, loss=0.887]
Epoch 2/4:  94%|█████████▍| 1175/1250 [31:00<01:27,  1.16s/it, loss=0.887]
Epoch 2/4:  94%|█████████▍| 1175/1250 [31:01<01:27,  1.16s/it, loss=0.91] 
Epoch 2/4:  94%|█████████▍| 1176/1250 [31:01<01:27,  1.18s/it, loss=0.91]
Epoch 2/4:  94%|█████████▍| 1176/1250 [31:02<01:27,  1.18s/it, loss=0.697]
Epoch 2/4:  94%|█████████▍| 1177/1250 [31:02<01:25,  1.17s/it, loss=0.697]
Epoch 2/4:  94%|█████████▍| 1177/1250 [31:03<01:25,  1.17s/it, loss=0.842]
Epoch 2/4:  94%|█████████▍| 1178/1250 [31:03<01:22,  1.15s/it, loss=0.842]
Epoch 2/4:  94%|█████████▍| 1178/1250 [31:04<01:22,  1.15s/it, loss=0.622]
Epoch 2/4:  94%|█████████▍| 1179/1250 [31:04<01:21,  1.15s/it, loss=0.622]
Epoch 2/4:  94%|█████████▍| 1179/1250 [31:05<01:21,  1.15s/it, loss=0.962]
Epoch 2/4:  94%|█████████▍| 1180/1250 [31:05<01:21,  1.17s/it, loss=0.962]
Epoch 2/4:  94%|█████████▍| 1180/1250 [31:07<01:21,  1.17s/it, loss=0.9]  
Epoch 2/4:  94%|█████████▍| 1181/1250 [31:07<01:21,  1.18s/it, loss=0.9]
Epoch 2/4:  94%|█████████▍| 1181/1250 [31:08<01:21,  1.18s/it, loss=0.416]
Epoch 2/4:  95%|█████████▍| 1182/1250 [31:08<01:19,  1.16s/it, loss=0.416]
Epoch 2/4:  95%|█████████▍| 1182/1250 [31:09<01:19,  1.16s/it, loss=0.661]
Epoch 2/4:  95%|█████████▍| 1183/1250 [31:09<01:17,  1.16s/it, loss=0.661]
Epoch 2/4:  95%|█████████▍| 1183/1250 [31:10<01:17,  1.16s/it, loss=0.98] 
Epoch 2/4:  95%|█████████▍| 1184/1250 [31:10<01:16,  1.16s/it, loss=0.98]
Epoch 2/4:  95%|█████████▍| 1184/1250 [31:11<01:16,  1.16s/it, loss=0.714]
Epoch 2/4:  95%|█████████▍| 1185/1250 [31:11<01:15,  1.16s/it, loss=0.714]
Epoch 2/4:  95%|█████████▍| 1185/1250 [31:12<01:15,  1.16s/it, loss=0.797]
Epoch 2/4:  95%|█████████▍| 1186/1250 [31:12<01:14,  1.16s/it, loss=0.797]
Epoch 2/4:  95%|█████████▍| 1186/1250 [31:13<01:14,  1.16s/it, loss=0.368]
Epoch 2/4:  95%|█████████▍| 1187/1250 [31:13<01:13,  1.16s/it, loss=0.368]
Epoch 2/4:  95%|█████████▍| 1187/1250 [31:15<01:13,  1.16s/it, loss=0.749]
Epoch 2/4:  95%|█████████▌| 1188/1250 [31:15<01:14,  1.21s/it, loss=0.749]
Epoch 2/4:  95%|█████████▌| 1188/1250 [31:16<01:14,  1.21s/it, loss=1.28] 
Epoch 2/4:  95%|█████████▌| 1189/1250 [31:16<01:14,  1.23s/it, loss=1.28]
Epoch 2/4:  95%|█████████▌| 1189/1250 [31:17<01:14,  1.23s/it, loss=0.863]
Epoch 2/4:  95%|█████████▌| 1190/1250 [31:17<01:11,  1.19s/it, loss=0.863]
Epoch 2/4:  95%|█████████▌| 1190/1250 [31:18<01:11,  1.19s/it, loss=0.789]
Epoch 2/4:  95%|█████████▌| 1191/1250 [31:18<01:07,  1.15s/it, loss=0.789]
Epoch 2/4:  95%|█████████▌| 1191/1250 [31:19<01:07,  1.15s/it, loss=0.747]
Epoch 2/4:  95%|█████████▌| 1192/1250 [31:19<01:05,  1.14s/it, loss=0.747]
Epoch 2/4:  95%|█████████▌| 1192/1250 [31:21<01:05,  1.14s/it, loss=0.788]
Epoch 2/4:  95%|█████████▌| 1193/1250 [31:21<01:05,  1.15s/it, loss=0.788]
Epoch 2/4:  95%|█████████▌| 1193/1250 [31:22<01:05,  1.15s/it, loss=0.891]
Epoch 2/4:  96%|█████████▌| 1194/1250 [31:22<01:04,  1.15s/it, loss=0.891]
Epoch 2/4:  96%|█████████▌| 1194/1250 [31:23<01:04,  1.15s/it, loss=0.516]
Epoch 2/4:  96%|█████████▌| 1195/1250 [31:23<01:02,  1.13s/it, loss=0.516]
Epoch 2/4:  96%|█████████▌| 1195/1250 [31:24<01:02,  1.13s/it, loss=0.782]
Epoch 2/4:  96%|█████████▌| 1196/1250 [31:24<01:00,  1.13s/it, loss=0.782]
Epoch 2/4:  96%|█████████▌| 1196/1250 [31:25<01:00,  1.13s/it, loss=0.733]
Epoch 2/4:  96%|█████████▌| 1197/1250 [31:25<01:01,  1.16s/it, loss=0.733]
Epoch 2/4:  96%|█████████▌| 1197/1250 [31:26<01:01,  1.16s/it, loss=0.764]
Epoch 2/4:  96%|█████████▌| 1198/1250 [31:26<01:01,  1.18s/it, loss=0.764]
Epoch 2/4:  96%|█████████▌| 1198/1250 [31:27<01:01,  1.18s/it, loss=0.788]
Epoch 2/4:  96%|█████████▌| 1199/1250 [31:27<00:59,  1.16s/it, loss=0.788]
Epoch 2/4:  96%|█████████▌| 1199/1250 [31:29<00:59,  1.16s/it, loss=0.461]
Epoch 2/4:  96%|█████████▌| 1200/1250 [31:29<00:57,  1.14s/it, loss=0.461]
Epoch 2/4:  96%|█████████▌| 1200/1250 [31:30<00:57,  1.14s/it, loss=0.68] 
Epoch 2/4:  96%|█████████▌| 1201/1250 [31:30<00:56,  1.15s/it, loss=0.68]
Epoch 2/4:  96%|█████████▌| 1201/1250 [31:31<00:56,  1.15s/it, loss=0.661]
Epoch 2/4:  96%|█████████▌| 1202/1250 [31:31<00:55,  1.16s/it, loss=0.661]
Epoch 2/4:  96%|█████████▌| 1202/1250 [31:32<00:55,  1.16s/it, loss=0.448]
Epoch 2/4:  96%|█████████▌| 1203/1250 [31:32<00:54,  1.16s/it, loss=0.448]
Epoch 2/4:  96%|█████████▌| 1203/1250 [31:33<00:54,  1.16s/it, loss=0.646]
Epoch 2/4:  96%|█████████▋| 1204/1250 [31:33<00:53,  1.16s/it, loss=0.646]
Epoch 2/4:  96%|█████████▋| 1204/1250 [31:34<00:53,  1.16s/it, loss=0.827]
Epoch 2/4:  96%|█████████▋| 1205/1250 [31:34<00:52,  1.17s/it, loss=0.827]
Epoch 2/4:  96%|█████████▋| 1205/1250 [31:36<00:52,  1.17s/it, loss=0.591]
Epoch 2/4:  96%|█████████▋| 1206/1250 [31:36<00:51,  1.17s/it, loss=0.591]
Epoch 2/4:  96%|█████████▋| 1206/1250 [31:37<00:51,  1.17s/it, loss=0.676]
Epoch 2/4:  97%|█████████▋| 1207/1250 [31:37<00:50,  1.17s/it, loss=0.676]
Epoch 2/4:  97%|█████████▋| 1207/1250 [31:38<00:50,  1.17s/it, loss=0.689]
Epoch 2/4:  97%|█████████▋| 1208/1250 [31:38<00:48,  1.15s/it, loss=0.689]
Epoch 2/4:  97%|█████████▋| 1208/1250 [31:39<00:48,  1.15s/it, loss=0.92] 
Epoch 2/4:  97%|█████████▋| 1209/1250 [31:39<00:47,  1.15s/it, loss=0.92]
Epoch 2/4:  97%|█████████▋| 1209/1250 [31:40<00:47,  1.15s/it, loss=0.512]
Epoch 2/4:  97%|█████████▋| 1210/1250 [31:40<00:46,  1.16s/it, loss=0.512]
Epoch 2/4:  97%|█████████▋| 1210/1250 [31:41<00:46,  1.16s/it, loss=0.663]
Epoch 2/4:  97%|█████████▋| 1211/1250 [31:41<00:45,  1.17s/it, loss=0.663]
Epoch 2/4:  97%|█████████▋| 1211/1250 [31:43<00:45,  1.17s/it, loss=0.88] 
Epoch 2/4:  97%|█████████▋| 1212/1250 [31:43<00:44,  1.18s/it, loss=0.88]
Epoch 2/4:  97%|█████████▋| 1212/1250 [31:44<00:44,  1.18s/it, loss=0.798]
Epoch 2/4:  97%|█████████▋| 1213/1250 [31:44<00:44,  1.19s/it, loss=0.798]
Epoch 2/4:  97%|█████████▋| 1213/1250 [31:45<00:44,  1.19s/it, loss=0.704]
Epoch 2/4:  97%|█████████▋| 1214/1250 [31:45<00:42,  1.19s/it, loss=0.704]
Epoch 2/4:  97%|█████████▋| 1214/1250 [31:46<00:42,  1.19s/it, loss=0.563]
Epoch 2/4:  97%|█████████▋| 1215/1250 [31:46<00:40,  1.17s/it, loss=0.563]
Epoch 2/4:  97%|█████████▋| 1215/1250 [31:47<00:40,  1.17s/it, loss=0.472]
Epoch 2/4:  97%|█████████▋| 1216/1250 [31:47<00:38,  1.15s/it, loss=0.472]
Epoch 2/4:  97%|█████████▋| 1216/1250 [31:48<00:38,  1.15s/it, loss=0.978]
Epoch 2/4:  97%|█████████▋| 1217/1250 [31:48<00:37,  1.13s/it, loss=0.978]
Epoch 2/4:  97%|█████████▋| 1217/1250 [31:49<00:37,  1.13s/it, loss=0.811]
Epoch 2/4:  97%|█████████▋| 1218/1250 [31:49<00:36,  1.13s/it, loss=0.811]
Epoch 2/4:  97%|█████████▋| 1218/1250 [31:51<00:36,  1.13s/it, loss=0.585]
Epoch 2/4:  98%|█████████▊| 1219/1250 [31:51<00:35,  1.15s/it, loss=0.585]
Epoch 2/4:  98%|█████████▊| 1219/1250 [31:52<00:35,  1.15s/it, loss=0.536]
Epoch 2/4:  98%|█████████▊| 1220/1250 [31:52<00:35,  1.17s/it, loss=0.536]
Epoch 2/4:  98%|█████████▊| 1220/1250 [31:53<00:35,  1.17s/it, loss=0.806]
Epoch 2/4:  98%|█████████▊| 1221/1250 [31:53<00:33,  1.15s/it, loss=0.806]
Epoch 2/4:  98%|█████████▊| 1221/1250 [31:54<00:33,  1.15s/it, loss=0.545]
Epoch 2/4:  98%|█████████▊| 1222/1250 [31:54<00:31,  1.13s/it, loss=0.545]
Epoch 2/4:  98%|█████████▊| 1222/1250 [31:55<00:31,  1.13s/it, loss=0.779]
Epoch 2/4:  98%|█████████▊| 1223/1250 [31:55<00:30,  1.14s/it, loss=0.779]
Epoch 2/4:  98%|█████████▊| 1223/1250 [31:56<00:30,  1.14s/it, loss=0.764]
Epoch 2/4:  98%|█████████▊| 1224/1250 [31:56<00:29,  1.15s/it, loss=0.764]
Epoch 2/4:  98%|█████████▊| 1224/1250 [31:57<00:29,  1.15s/it, loss=0.582]
Epoch 2/4:  98%|█████████▊| 1225/1250 [31:57<00:28,  1.15s/it, loss=0.582]
Epoch 2/4:  98%|█████████▊| 1225/1250 [31:59<00:28,  1.15s/it, loss=0.694]
Epoch 2/4:  98%|█████████▊| 1226/1250 [31:59<00:27,  1.13s/it, loss=0.694]
Epoch 2/4:  98%|█████████▊| 1226/1250 [32:00<00:27,  1.13s/it, loss=0.559]
Epoch 2/4:  98%|█████████▊| 1227/1250 [32:00<00:26,  1.14s/it, loss=0.559]
Epoch 2/4:  98%|█████████▊| 1227/1250 [32:01<00:26,  1.14s/it, loss=0.523]
Epoch 2/4:  98%|█████████▊| 1228/1250 [32:01<00:25,  1.17s/it, loss=0.523]
Epoch 2/4:  98%|█████████▊| 1228/1250 [32:02<00:25,  1.17s/it, loss=0.711]
Epoch 2/4:  98%|█████████▊| 1229/1250 [32:02<00:24,  1.18s/it, loss=0.711]
Epoch 2/4:  98%|█████████▊| 1229/1250 [32:03<00:24,  1.18s/it, loss=0.63] 
Epoch 2/4:  98%|█████████▊| 1230/1250 [32:03<00:23,  1.15s/it, loss=0.63]
Epoch 2/4:  98%|█████████▊| 1230/1250 [32:04<00:23,  1.15s/it, loss=0.666]
Epoch 2/4:  98%|█████████▊| 1231/1250 [32:04<00:21,  1.14s/it, loss=0.666]
Epoch 2/4:  98%|█████████▊| 1231/1250 [32:06<00:21,  1.14s/it, loss=0.595]
Epoch 2/4:  99%|█████████▊| 1232/1250 [32:06<00:20,  1.14s/it, loss=0.595]
Epoch 2/4:  99%|█████████▊| 1232/1250 [32:07<00:20,  1.14s/it, loss=0.778]
Epoch 2/4:  99%|█████████▊| 1233/1250 [32:07<00:19,  1.14s/it, loss=0.778]
Epoch 2/4:  99%|█████████▊| 1233/1250 [32:08<00:19,  1.14s/it, loss=1.05] 
Epoch 2/4:  99%|█████████▊| 1234/1250 [32:08<00:18,  1.13s/it, loss=1.05]
Epoch 2/4:  99%|█████████▊| 1234/1250 [32:09<00:18,  1.13s/it, loss=0.639]
Epoch 2/4:  99%|█████████▉| 1235/1250 [32:09<00:16,  1.12s/it, loss=0.639]
Epoch 2/4:  99%|█████████▉| 1235/1250 [32:10<00:16,  1.12s/it, loss=0.706]
Epoch 2/4:  99%|█████████▉| 1236/1250 [32:10<00:16,  1.14s/it, loss=0.706]
Epoch 2/4:  99%|█████████▉| 1236/1250 [32:11<00:16,  1.14s/it, loss=0.734]
Epoch 2/4:  99%|█████████▉| 1237/1250 [32:11<00:15,  1.20s/it, loss=0.734]
Epoch 2/4:  99%|█████████▉| 1237/1250 [32:13<00:15,  1.20s/it, loss=0.822]
Epoch 2/4:  99%|█████████▉| 1238/1250 [32:13<00:14,  1.21s/it, loss=0.822]
Epoch 2/4:  99%|█████████▉| 1238/1250 [32:14<00:14,  1.21s/it, loss=0.548]
Epoch 2/4:  99%|█████████▉| 1239/1250 [32:14<00:12,  1.17s/it, loss=0.548]
Epoch 2/4:  99%|█████████▉| 1239/1250 [32:15<00:12,  1.17s/it, loss=0.671]
Epoch 2/4:  99%|█████████▉| 1240/1250 [32:15<00:11,  1.14s/it, loss=0.671]
Epoch 2/4:  99%|█████████▉| 1240/1250 [32:16<00:11,  1.14s/it, loss=0.631]
Epoch 2/4:  99%|█████████▉| 1241/1250 [32:16<00:10,  1.14s/it, loss=0.631]
Epoch 2/4:  99%|█████████▉| 1241/1250 [32:17<00:10,  1.14s/it, loss=0.59] 
Epoch 2/4:  99%|█████████▉| 1242/1250 [32:17<00:09,  1.14s/it, loss=0.59]
Epoch 2/4:  99%|█████████▉| 1242/1250 [32:18<00:09,  1.14s/it, loss=0.988]
Epoch 2/4:  99%|█████████▉| 1243/1250 [32:18<00:07,  1.14s/it, loss=0.988]
Epoch 2/4:  99%|█████████▉| 1243/1250 [32:19<00:07,  1.14s/it, loss=0.751]
Epoch 2/4: 100%|█████████▉| 1244/1250 [32:19<00:06,  1.15s/it, loss=0.751]
Epoch 2/4: 100%|█████████▉| 1244/1250 [32:21<00:06,  1.15s/it, loss=0.814]
Epoch 2/4: 100%|█████████▉| 1245/1250 [32:21<00:05,  1.17s/it, loss=0.814]
Epoch 2/4: 100%|█████████▉| 1245/1250 [32:22<00:05,  1.17s/it, loss=0.967]
Epoch 2/4: 100%|█████████▉| 1246/1250 [32:22<00:04,  1.18s/it, loss=0.967]
Epoch 2/4: 100%|█████████▉| 1246/1250 [32:23<00:04,  1.18s/it, loss=0.754]
Epoch 2/4: 100%|█████████▉| 1247/1250 [32:23<00:03,  1.16s/it, loss=0.754]
Epoch 2/4: 100%|█████████▉| 1247/1250 [32:24<00:03,  1.16s/it, loss=0.906]
Epoch 2/4: 100%|█████████▉| 1248/1250 [32:24<00:02,  1.13s/it, loss=0.906]
Epoch 2/4: 100%|█████████▉| 1248/1250 [32:25<00:02,  1.13s/it, loss=0.634]
Epoch 2/4: 100%|█████████▉| 1249/1250 [32:25<00:01,  1.14s/it, loss=0.634]
Epoch 2/4: 100%|█████████▉| 1249/1250 [32:26<00:01,  1.14s/it, loss=0.703]
Epoch 2/4: 100%|██████████| 1250/1250 [32:26<00:00,  1.16s/it, loss=0.703]
Epoch 2/4: 100%|██████████| 1250/1250 [32:26<00:00,  1.56s/it, loss=0.703]
Epoch 2 Avg Loss: 0.7048

Epoch 3/4:   0%|          | 0/1250 [00:00<?, ?it/s]
Epoch 3/4:   0%|          | 0/1250 [00:01<?, ?it/s, loss=0.332]
Epoch 3/4:   0%|          | 1/1250 [00:01<23:34,  1.13s/it, loss=0.332]
Epoch 3/4:   0%|          | 1/1250 [00:02<23:34,  1.13s/it, loss=0.572]
Epoch 3/4:   0%|          | 2/1250 [00:02<22:40,  1.09s/it, loss=0.572]
Epoch 3/4:   0%|          | 2/1250 [00:03<22:40,  1.09s/it, loss=0.455]
Epoch 3/4:   0%|          | 3/1250 [00:03<22:54,  1.10s/it, loss=0.455]
Epoch 3/4:   0%|          | 3/1250 [00:04<22:54,  1.10s/it, loss=0.533]
Epoch 3/4:   0%|          | 4/1250 [00:04<23:38,  1.14s/it, loss=0.533]
Epoch 3/4:   0%|          | 4/1250 [00:05<23:38,  1.14s/it, loss=0.436]
Epoch 3/4:   0%|          | 5/1250 [00:05<23:48,  1.15s/it, loss=0.436]
Epoch 3/4:   0%|          | 5/1250 [00:06<23:48,  1.15s/it, loss=0.573]
Epoch 3/4:   0%|          | 6/1250 [00:06<23:17,  1.12s/it, loss=0.573]
Epoch 3/4:   0%|          | 6/1250 [00:07<23:17,  1.12s/it, loss=0.545]
Epoch 3/4:   1%|          | 7/1250 [00:07<22:58,  1.11s/it, loss=0.545]
Epoch 3/4:   1%|          | 7/1250 [00:09<22:58,  1.11s/it, loss=0.418]
Epoch 3/4:   1%|          | 8/1250 [00:09<23:40,  1.14s/it, loss=0.418]
Epoch 3/4:   1%|          | 8/1250 [00:10<23:40,  1.14s/it, loss=0.715]
Epoch 3/4:   1%|          | 9/1250 [00:10<24:06,  1.17s/it, loss=0.715]
Epoch 3/4:   1%|          | 9/1250 [00:11<24:06,  1.17s/it, loss=0.507]
Epoch 3/4:   1%|          | 10/1250 [00:11<24:24,  1.18s/it, loss=0.507]
Epoch 3/4:   1%|          | 10/1250 [00:12<24:24,  1.18s/it, loss=0.537]
Epoch 3/4:   1%|          | 11/1250 [00:12<24:03,  1.16s/it, loss=0.537]
Epoch 3/4:   1%|          | 11/1250 [00:13<24:03,  1.16s/it, loss=0.473]
Epoch 3/4:   1%|          | 12/1250 [00:13<24:07,  1.17s/it, loss=0.473]
Epoch 3/4:   1%|          | 12/1250 [00:14<24:07,  1.17s/it, loss=0.672]
Epoch 3/4:   1%|          | 13/1250 [00:14<24:10,  1.17s/it, loss=0.672]
Epoch 3/4:   1%|          | 13/1250 [00:16<24:10,  1.17s/it, loss=0.433]
Epoch 3/4:   1%|          | 14/1250 [00:16<24:19,  1.18s/it, loss=0.433]
Epoch 3/4:   1%|          | 14/1250 [00:17<24:19,  1.18s/it, loss=0.679]
Epoch 3/4:   1%|          | 15/1250 [00:17<23:49,  1.16s/it, loss=0.679]
Epoch 3/4:   1%|          | 15/1250 [00:18<23:49,  1.16s/it, loss=0.398]
Epoch 3/4:   1%|▏         | 16/1250 [00:18<23:54,  1.16s/it, loss=0.398]
Epoch 3/4:   1%|▏         | 16/1250 [00:19<23:54,  1.16s/it, loss=0.591]
Epoch 3/4:   1%|▏         | 17/1250 [00:19<23:37,  1.15s/it, loss=0.591]
Epoch 3/4:   1%|▏         | 17/1250 [00:20<23:37,  1.15s/it, loss=0.664]
Epoch 3/4:   1%|▏         | 18/1250 [00:20<23:52,  1.16s/it, loss=0.664]
Epoch 3/4:   1%|▏         | 18/1250 [00:21<23:52,  1.16s/it, loss=0.66] 
Epoch 3/4:   2%|▏         | 19/1250 [00:21<23:37,  1.15s/it, loss=0.66]
Epoch 3/4:   2%|▏         | 19/1250 [00:22<23:37,  1.15s/it, loss=0.595]
Epoch 3/4:   2%|▏         | 20/1250 [00:22<23:10,  1.13s/it, loss=0.595]
Epoch 3/4:   2%|▏         | 20/1250 [00:24<23:10,  1.13s/it, loss=0.639]
Epoch 3/4:   2%|▏         | 21/1250 [00:24<23:10,  1.13s/it, loss=0.639]
Epoch 3/4:   2%|▏         | 21/1250 [00:25<23:10,  1.13s/it, loss=0.52] 
Epoch 3/4:   2%|▏         | 22/1250 [00:25<23:38,  1.16s/it, loss=0.52]
Epoch 3/4:   2%|▏         | 22/1250 [00:26<23:38,  1.16s/it, loss=0.691]
Epoch 3/4:   2%|▏         | 23/1250 [00:26<24:02,  1.18s/it, loss=0.691]
Epoch 3/4:   2%|▏         | 23/1250 [00:27<24:02,  1.18s/it, loss=0.46] 
Epoch 3/4:   2%|▏         | 24/1250 [00:27<23:27,  1.15s/it, loss=0.46]
Epoch 3/4:   2%|▏         | 24/1250 [00:28<23:27,  1.15s/it, loss=0.55]
Epoch 3/4:   2%|▏         | 25/1250 [00:28<23:16,  1.14s/it, loss=0.55]
Epoch 3/4:   2%|▏         | 25/1250 [00:29<23:16,  1.14s/it, loss=0.661]
Epoch 3/4:   2%|▏         | 26/1250 [00:29<23:21,  1.14s/it, loss=0.661]
Epoch 3/4:   2%|▏         | 26/1250 [00:31<23:21,  1.14s/it, loss=0.547]
Epoch 3/4:   2%|▏         | 27/1250 [00:31<23:34,  1.16s/it, loss=0.547]
Epoch 3/4:   2%|▏         | 27/1250 [00:32<23:34,  1.16s/it, loss=0.491]
Epoch 3/4:   2%|▏         | 28/1250 [00:32<23:20,  1.15s/it, loss=0.491]
Epoch 3/4:   2%|▏         | 28/1250 [00:33<23:20,  1.15s/it, loss=0.521]
Epoch 3/4:   2%|▏         | 29/1250 [00:33<23:15,  1.14s/it, loss=0.521]
Epoch 3/4:   2%|▏         | 29/1250 [00:34<23:15,  1.14s/it, loss=0.522]
Epoch 3/4:   2%|▏         | 30/1250 [00:34<23:36,  1.16s/it, loss=0.522]
Epoch 3/4:   2%|▏         | 30/1250 [00:35<23:36,  1.16s/it, loss=0.684]
Epoch 3/4:   2%|▏         | 31/1250 [00:35<23:31,  1.16s/it, loss=0.684]
Epoch 3/4:   2%|▏         | 31/1250 [00:36<23:31,  1.16s/it, loss=0.512]
Epoch 3/4:   3%|▎         | 32/1250 [00:36<23:51,  1.17s/it, loss=0.512]
Epoch 3/4:   3%|▎         | 32/1250 [00:38<23:51,  1.17s/it, loss=0.464]
Epoch 3/4:   3%|▎         | 33/1250 [00:38<23:43,  1.17s/it, loss=0.464]
Epoch 3/4:   3%|▎         | 33/1250 [00:39<23:43,  1.17s/it, loss=0.551]
Epoch 3/4:   3%|▎         | 34/1250 [00:39<24:12,  1.19s/it, loss=0.551]
Epoch 3/4:   3%|▎         | 34/1250 [00:40<24:12,  1.19s/it, loss=0.437]
Epoch 3/4:   3%|▎         | 35/1250 [00:40<23:51,  1.18s/it, loss=0.437]
Epoch 3/4:   3%|▎         | 35/1250 [00:41<23:51,  1.18s/it, loss=0.623]
Epoch 3/4:   3%|▎         | 36/1250 [00:41<23:18,  1.15s/it, loss=0.623]
Epoch 3/4:   3%|▎         | 36/1250 [00:42<23:18,  1.15s/it, loss=0.607]
Epoch 3/4:   3%|▎         | 37/1250 [00:42<22:40,  1.12s/it, loss=0.607]
Epoch 3/4:   3%|▎         | 37/1250 [00:43<22:40,  1.12s/it, loss=0.356]
Epoch 3/4:   3%|▎         | 38/1250 [00:43<22:30,  1.11s/it, loss=0.356]
Epoch 3/4:   3%|▎         | 38/1250 [00:44<22:30,  1.11s/it, loss=0.475]
Epoch 3/4:   3%|▎         | 39/1250 [00:44<23:08,  1.15s/it, loss=0.475]
Epoch 3/4:   3%|▎         | 39/1250 [00:46<23:08,  1.15s/it, loss=0.554]
Epoch 3/4:   3%|▎         | 40/1250 [00:46<24:15,  1.20s/it, loss=0.554]
Epoch 3/4:   3%|▎         | 40/1250 [00:47<24:15,  1.20s/it, loss=0.29] 
Epoch 3/4:   3%|▎         | 41/1250 [00:47<24:03,  1.19s/it, loss=0.29]
Epoch 3/4:   3%|▎         | 41/1250 [00:48<24:03,  1.19s/it, loss=0.377]
Epoch 3/4:   3%|▎         | 42/1250 [00:48<23:18,  1.16s/it, loss=0.377]
Epoch 3/4:   3%|▎         | 42/1250 [00:49<23:18,  1.16s/it, loss=0.459]
Epoch 3/4:   3%|▎         | 43/1250 [00:49<22:43,  1.13s/it, loss=0.459]
Epoch 3/4:   3%|▎         | 43/1250 [00:50<22:43,  1.13s/it, loss=0.576]
Epoch 3/4:   4%|▎         | 44/1250 [00:50<22:29,  1.12s/it, loss=0.576]
Epoch 3/4:   4%|▎         | 44/1250 [00:51<22:29,  1.12s/it, loss=0.673]
Epoch 3/4:   4%|▎         | 45/1250 [00:51<22:26,  1.12s/it, loss=0.673]
Epoch 3/4:   4%|▎         | 45/1250 [00:52<22:26,  1.12s/it, loss=0.595]
Epoch 3/4:   4%|▎         | 46/1250 [00:52<22:13,  1.11s/it, loss=0.595]
Epoch 3/4:   4%|▎         | 46/1250 [00:54<22:13,  1.11s/it, loss=0.39] 
Epoch 3/4:   4%|▍         | 47/1250 [00:54<22:39,  1.13s/it, loss=0.39]
Epoch 3/4:   4%|▍         | 47/1250 [00:55<22:39,  1.13s/it, loss=0.475]
Epoch 3/4:   4%|▍         | 48/1250 [00:55<23:04,  1.15s/it, loss=0.475]
Epoch 3/4:   4%|▍         | 48/1250 [00:56<23:04,  1.15s/it, loss=0.547]
Epoch 3/4:   4%|▍         | 49/1250 [00:56<23:10,  1.16s/it, loss=0.547]
Epoch 3/4:   4%|▍         | 49/1250 [00:57<23:10,  1.16s/it, loss=0.318]
Epoch 3/4:   4%|▍         | 50/1250 [00:57<22:38,  1.13s/it, loss=0.318]
Epoch 3/4:   4%|▍         | 50/1250 [00:58<22:38,  1.13s/it, loss=0.348]
Epoch 3/4:   4%|▍         | 51/1250 [00:58<22:13,  1.11s/it, loss=0.348]
Epoch 3/4:   4%|▍         | 51/1250 [00:59<22:13,  1.11s/it, loss=0.554]
Epoch 3/4:   4%|▍         | 52/1250 [00:59<22:35,  1.13s/it, loss=0.554]
Epoch 3/4:   4%|▍         | 52/1250 [01:00<22:35,  1.13s/it, loss=0.306]
Epoch 3/4:   4%|▍         | 53/1250 [01:00<23:02,  1.15s/it, loss=0.306]
Epoch 3/4:   4%|▍         | 53/1250 [01:02<23:02,  1.15s/it, loss=0.603]
Epoch 3/4:   4%|▍         | 54/1250 [01:02<22:50,  1.15s/it, loss=0.603]
Epoch 3/4:   4%|▍         | 54/1250 [01:03<22:50,  1.15s/it, loss=0.743]
Epoch 3/4:   4%|▍         | 55/1250 [01:03<22:18,  1.12s/it, loss=0.743]
Epoch 3/4:   4%|▍         | 55/1250 [01:04<22:18,  1.12s/it, loss=0.635]
Epoch 3/4:   4%|▍         | 56/1250 [01:04<22:24,  1.13s/it, loss=0.635]
Epoch 3/4:   4%|▍         | 56/1250 [01:05<22:24,  1.13s/it, loss=0.372]
Epoch 3/4:   5%|▍         | 57/1250 [01:05<22:56,  1.15s/it, loss=0.372]
Epoch 3/4:   5%|▍         | 57/1250 [01:06<22:56,  1.15s/it, loss=0.687]
Epoch 3/4:   5%|▍         | 58/1250 [01:06<23:03,  1.16s/it, loss=0.687]
Epoch 3/4:   5%|▍         | 58/1250 [01:07<23:03,  1.16s/it, loss=0.566]
Epoch 3/4:   5%|▍         | 59/1250 [01:07<22:37,  1.14s/it, loss=0.566]
Epoch 3/4:   5%|▍         | 59/1250 [01:08<22:37,  1.14s/it, loss=0.475]
Epoch 3/4:   5%|▍         | 60/1250 [01:08<22:29,  1.13s/it, loss=0.475]
Epoch 3/4:   5%|▍         | 60/1250 [01:10<22:29,  1.13s/it, loss=0.368]
Epoch 3/4:   5%|▍         | 61/1250 [01:10<22:44,  1.15s/it, loss=0.368]
Epoch 3/4:   5%|▍         | 61/1250 [01:11<22:44,  1.15s/it, loss=0.408]
Epoch 3/4:   5%|▍         | 62/1250 [01:11<23:20,  1.18s/it, loss=0.408]
Epoch 3/4:   5%|▍         | 62/1250 [01:12<23:20,  1.18s/it, loss=0.606]
Epoch 3/4:   5%|▌         | 63/1250 [01:12<23:28,  1.19s/it, loss=0.606]
Epoch 3/4:   5%|▌         | 63/1250 [01:13<23:28,  1.19s/it, loss=0.48] 
Epoch 3/4:   5%|▌         | 64/1250 [01:13<23:01,  1.17s/it, loss=0.48]
Epoch 3/4:   5%|▌         | 64/1250 [01:14<23:01,  1.17s/it, loss=0.499]
Epoch 3/4:   5%|▌         | 65/1250 [01:14<22:37,  1.15s/it, loss=0.499]
Epoch 3/4:   5%|▌         | 65/1250 [01:15<22:37,  1.15s/it, loss=0.688]
Epoch 3/4:   5%|▌         | 66/1250 [01:15<22:37,  1.15s/it, loss=0.688]
Epoch 3/4:   5%|▌         | 66/1250 [01:16<22:37,  1.15s/it, loss=0.665]
Epoch 3/4:   5%|▌         | 67/1250 [01:16<22:33,  1.14s/it, loss=0.665]
Epoch 3/4:   5%|▌         | 67/1250 [01:18<22:33,  1.14s/it, loss=0.544]
Epoch 3/4:   5%|▌         | 68/1250 [01:18<22:09,  1.12s/it, loss=0.544]
Epoch 3/4:   5%|▌         | 68/1250 [01:19<22:09,  1.12s/it, loss=0.621]
Epoch 3/4:   6%|▌         | 69/1250 [01:19<21:54,  1.11s/it, loss=0.621]
Epoch 3/4:   6%|▌         | 69/1250 [01:20<21:54,  1.11s/it, loss=0.571]
Epoch 3/4:   6%|▌         | 70/1250 [01:20<22:07,  1.13s/it, loss=0.571]
Epoch 3/4:   6%|▌         | 70/1250 [01:21<22:07,  1.13s/it, loss=0.399]
Epoch 3/4:   6%|▌         | 71/1250 [01:21<22:23,  1.14s/it, loss=0.399]
Epoch 3/4:   6%|▌         | 71/1250 [01:22<22:23,  1.14s/it, loss=0.426]
Epoch 3/4:   6%|▌         | 72/1250 [01:22<22:16,  1.13s/it, loss=0.426]
Epoch 3/4:   6%|▌         | 72/1250 [01:23<22:16,  1.13s/it, loss=0.395]
Epoch 3/4:   6%|▌         | 73/1250 [01:23<21:34,  1.10s/it, loss=0.395]
Epoch 3/4:   6%|▌         | 73/1250 [01:24<21:34,  1.10s/it, loss=0.411]
Epoch 3/4:   6%|▌         | 74/1250 [01:24<21:37,  1.10s/it, loss=0.411]
Epoch 3/4:   6%|▌         | 74/1250 [01:25<21:37,  1.10s/it, loss=0.394]
Epoch 3/4:   6%|▌         | 75/1250 [01:25<22:01,  1.12s/it, loss=0.394]
Epoch 3/4:   6%|▌         | 75/1250 [01:27<22:01,  1.12s/it, loss=0.55] 
Epoch 3/4:   6%|▌         | 76/1250 [01:27<22:06,  1.13s/it, loss=0.55]
Epoch 3/4:   6%|▌         | 76/1250 [01:28<22:06,  1.13s/it, loss=0.992]
Epoch 3/4:   6%|▌         | 77/1250 [01:28<21:45,  1.11s/it, loss=0.992]
Epoch 3/4:   6%|▌         | 77/1250 [01:29<21:45,  1.11s/it, loss=0.494]
Epoch 3/4:   6%|▌         | 78/1250 [01:29<21:36,  1.11s/it, loss=0.494]
Epoch 3/4:   6%|▌         | 78/1250 [01:30<21:36,  1.11s/it, loss=0.541]
Epoch 3/4:   6%|▋         | 79/1250 [01:30<21:55,  1.12s/it, loss=0.541]
Epoch 3/4:   6%|▋         | 79/1250 [01:31<21:55,  1.12s/it, loss=0.505]
Epoch 3/4:   6%|▋         | 80/1250 [01:31<22:14,  1.14s/it, loss=0.505]
Epoch 3/4:   6%|▋         | 80/1250 [01:32<22:14,  1.14s/it, loss=0.57] 
Epoch 3/4:   6%|▋         | 81/1250 [01:32<21:59,  1.13s/it, loss=0.57]
Epoch 3/4:   6%|▋         | 81/1250 [01:33<21:59,  1.13s/it, loss=0.44]
Epoch 3/4:   7%|▋         | 82/1250 [01:33<21:51,  1.12s/it, loss=0.44]
Epoch 3/4:   7%|▋         | 82/1250 [01:34<21:51,  1.12s/it, loss=0.418]
Epoch 3/4:   7%|▋         | 83/1250 [01:34<22:00,  1.13s/it, loss=0.418]
Epoch 3/4:   7%|▋         | 83/1250 [01:36<22:00,  1.13s/it, loss=0.519]
Epoch 3/4:   7%|▋         | 84/1250 [01:36<22:22,  1.15s/it, loss=0.519]
Epoch 3/4:   7%|▋         | 84/1250 [01:37<22:22,  1.15s/it, loss=0.409]
Epoch 3/4:   7%|▋         | 85/1250 [01:37<22:15,  1.15s/it, loss=0.409]
Epoch 3/4:   7%|▋         | 85/1250 [01:38<22:15,  1.15s/it, loss=0.514]
Epoch 3/4:   7%|▋         | 86/1250 [01:38<21:43,  1.12s/it, loss=0.514]
Epoch 3/4:   7%|▋         | 86/1250 [01:39<21:43,  1.12s/it, loss=0.62] 
Epoch 3/4:   7%|▋         | 87/1250 [01:39<21:42,  1.12s/it, loss=0.62]
Epoch 3/4:   7%|▋         | 87/1250 [01:40<21:42,  1.12s/it, loss=0.466]
Epoch 3/4:   7%|▋         | 88/1250 [01:40<22:16,  1.15s/it, loss=0.466]
Epoch 3/4:   7%|▋         | 88/1250 [01:41<22:16,  1.15s/it, loss=0.328]
Epoch 3/4:   7%|▋         | 89/1250 [01:41<22:38,  1.17s/it, loss=0.328]
Epoch 3/4:   7%|▋         | 89/1250 [01:42<22:38,  1.17s/it, loss=0.495]
Epoch 3/4:   7%|▋         | 90/1250 [01:42<22:00,  1.14s/it, loss=0.495]
Epoch 3/4:   7%|▋         | 90/1250 [01:43<22:00,  1.14s/it, loss=0.563]
Epoch 3/4:   7%|▋         | 91/1250 [01:43<21:30,  1.11s/it, loss=0.563]
Epoch 3/4:   7%|▋         | 91/1250 [01:45<21:30,  1.11s/it, loss=0.572]
Epoch 3/4:   7%|▋         | 92/1250 [01:45<21:44,  1.13s/it, loss=0.572]
Epoch 3/4:   7%|▋         | 92/1250 [01:46<21:44,  1.13s/it, loss=0.467]
Epoch 3/4:   7%|▋         | 93/1250 [01:46<22:15,  1.15s/it, loss=0.467]
Epoch 3/4:   7%|▋         | 93/1250 [01:47<22:15,  1.15s/it, loss=0.379]
Epoch 3/4:   8%|▊         | 94/1250 [01:47<22:01,  1.14s/it, loss=0.379]
Epoch 3/4:   8%|▊         | 94/1250 [01:48<22:01,  1.14s/it, loss=0.676]
Epoch 3/4:   8%|▊         | 95/1250 [01:48<21:28,  1.12s/it, loss=0.676]
Epoch 3/4:   8%|▊         | 95/1250 [01:49<21:28,  1.12s/it, loss=0.516]
Epoch 3/4:   8%|▊         | 96/1250 [01:49<21:13,  1.10s/it, loss=0.516]
Epoch 3/4:   8%|▊         | 96/1250 [01:50<21:13,  1.10s/it, loss=0.43] 
Epoch 3/4:   8%|▊         | 97/1250 [01:50<21:26,  1.12s/it, loss=0.43]
Epoch 3/4:   8%|▊         | 97/1250 [01:51<21:26,  1.12s/it, loss=0.65]
Epoch 3/4:   8%|▊         | 98/1250 [01:51<21:29,  1.12s/it, loss=0.65]
Epoch 3/4:   8%|▊         | 98/1250 [01:52<21:29,  1.12s/it, loss=0.521]
Epoch 3/4:   8%|▊         | 99/1250 [01:52<21:19,  1.11s/it, loss=0.521]
Epoch 3/4:   8%|▊         | 99/1250 [01:54<21:19,  1.11s/it, loss=0.405]
Epoch 3/4:   8%|▊         | 100/1250 [01:54<21:16,  1.11s/it, loss=0.405]
Epoch 3/4:   8%|▊         | 100/1250 [01:55<21:16,  1.11s/it, loss=0.538]
Epoch 3/4:   8%|▊         | 101/1250 [01:55<21:35,  1.13s/it, loss=0.538]
Epoch 3/4:   8%|▊         | 101/1250 [01:56<21:35,  1.13s/it, loss=0.379]
Epoch 3/4:   8%|▊         | 102/1250 [01:56<22:00,  1.15s/it, loss=0.379]
Epoch 3/4:   8%|▊         | 102/1250 [01:57<22:00,  1.15s/it, loss=0.529]
Epoch 3/4:   8%|▊         | 103/1250 [01:57<21:47,  1.14s/it, loss=0.529]
Epoch 3/4:   8%|▊         | 103/1250 [01:58<21:47,  1.14s/it, loss=0.596]
Epoch 3/4:   8%|▊         | 104/1250 [01:58<21:15,  1.11s/it, loss=0.596]
Epoch 3/4:   8%|▊         | 104/1250 [01:59<21:15,  1.11s/it, loss=0.63] 
Epoch 3/4:   8%|▊         | 105/1250 [01:59<21:06,  1.11s/it, loss=0.63]
Epoch 3/4:   8%|▊         | 105/1250 [02:00<21:06,  1.11s/it, loss=0.681]
Epoch 3/4:   8%|▊         | 106/1250 [02:00<21:24,  1.12s/it, loss=0.681]
Epoch 3/4:   8%|▊         | 106/1250 [02:02<21:24,  1.12s/it, loss=0.586]
Epoch 3/4:   9%|▊         | 107/1250 [02:02<21:32,  1.13s/it, loss=0.586]
Epoch 3/4:   9%|▊         | 107/1250 [02:03<21:32,  1.13s/it, loss=0.571]
Epoch 3/4:   9%|▊         | 108/1250 [02:03<21:21,  1.12s/it, loss=0.571]
Epoch 3/4:   9%|▊         | 108/1250 [02:04<21:21,  1.12s/it, loss=0.62] 
Epoch 3/4:   9%|▊         | 109/1250 [02:04<20:56,  1.10s/it, loss=0.62]
Epoch 3/4:   9%|▊         | 109/1250 [02:05<20:56,  1.10s/it, loss=0.625]
Epoch 3/4:   9%|▉         | 110/1250 [02:05<21:06,  1.11s/it, loss=0.625]
Epoch 3/4:   9%|▉         | 110/1250 [02:06<21:06,  1.11s/it, loss=0.563]
Epoch 3/4:   9%|▉         | 111/1250 [02:06<21:37,  1.14s/it, loss=0.563]
Epoch 3/4:   9%|▉         | 111/1250 [02:07<21:37,  1.14s/it, loss=0.87] 
Epoch 3/4:   9%|▉         | 112/1250 [02:07<21:50,  1.15s/it, loss=0.87]
Epoch 3/4:   9%|▉         | 112/1250 [02:08<21:50,  1.15s/it, loss=0.413]
Epoch 3/4:   9%|▉         | 113/1250 [02:08<21:24,  1.13s/it, loss=0.413]
Epoch 3/4:   9%|▉         | 113/1250 [02:09<21:24,  1.13s/it, loss=0.406]
Epoch 3/4:   9%|▉         | 114/1250 [02:09<21:04,  1.11s/it, loss=0.406]
Epoch 3/4:   9%|▉         | 114/1250 [02:11<21:04,  1.11s/it, loss=0.413]
Epoch 3/4:   9%|▉         | 115/1250 [02:11<21:30,  1.14s/it, loss=0.413]
Epoch 3/4:   9%|▉         | 115/1250 [02:12<21:30,  1.14s/it, loss=0.434]
Epoch 3/4:   9%|▉         | 116/1250 [02:12<21:46,  1.15s/it, loss=0.434]
Epoch 3/4:   9%|▉         | 116/1250 [02:13<21:46,  1.15s/it, loss=0.621]
Epoch 3/4:   9%|▉         | 117/1250 [02:13<21:19,  1.13s/it, loss=0.621]
Epoch 3/4:   9%|▉         | 117/1250 [02:14<21:19,  1.13s/it, loss=0.458]
Epoch 3/4:   9%|▉         | 118/1250 [02:14<20:44,  1.10s/it, loss=0.458]
Epoch 3/4:   9%|▉         | 118/1250 [02:15<20:44,  1.10s/it, loss=0.585]
Epoch 3/4:  10%|▉         | 119/1250 [02:15<20:57,  1.11s/it, loss=0.585]
Epoch 3/4:  10%|▉         | 119/1250 [02:16<20:57,  1.11s/it, loss=0.636]
Epoch 3/4:  10%|▉         | 120/1250 [02:16<21:23,  1.14s/it, loss=0.636]
Epoch 3/4:  10%|▉         | 120/1250 [02:17<21:23,  1.14s/it, loss=0.778]
Epoch 3/4:  10%|▉         | 121/1250 [02:17<21:21,  1.14s/it, loss=0.778]
Epoch 3/4:  10%|▉         | 121/1250 [02:18<21:21,  1.14s/it, loss=0.614]
Epoch 3/4:  10%|▉         | 122/1250 [02:18<20:52,  1.11s/it, loss=0.614]
Epoch 3/4:  10%|▉         | 122/1250 [02:19<20:52,  1.11s/it, loss=0.537]
Epoch 3/4:  10%|▉         | 123/1250 [02:19<20:42,  1.10s/it, loss=0.537]
Epoch 3/4:  10%|▉         | 123/1250 [02:21<20:42,  1.10s/it, loss=0.367]
Epoch 3/4:  10%|▉         | 124/1250 [02:21<21:02,  1.12s/it, loss=0.367]
Epoch 3/4:  10%|▉         | 124/1250 [02:22<21:02,  1.12s/it, loss=0.526]
Epoch 3/4:  10%|█         | 125/1250 [02:22<21:10,  1.13s/it, loss=0.526]
Epoch 3/4:  10%|█         | 125/1250 [02:23<21:10,  1.13s/it, loss=0.534]
Epoch 3/4:  10%|█         | 126/1250 [02:23<20:50,  1.11s/it, loss=0.534]
Epoch 3/4:  10%|█         | 126/1250 [02:24<20:50,  1.11s/it, loss=1.04] 
Epoch 3/4:  10%|█         | 127/1250 [02:24<20:26,  1.09s/it, loss=1.04]
Epoch 3/4:  10%|█         | 127/1250 [02:25<20:26,  1.09s/it, loss=0.524]
Epoch 3/4:  10%|█         | 128/1250 [02:25<20:42,  1.11s/it, loss=0.524]
Epoch 3/4:  10%|█         | 128/1250 [02:26<20:42,  1.11s/it, loss=0.436]
Epoch 3/4:  10%|█         | 129/1250 [02:26<21:11,  1.13s/it, loss=0.436]
Epoch 3/4:  10%|█         | 129/1250 [02:27<21:11,  1.13s/it, loss=0.616]
Epoch 3/4:  10%|█         | 130/1250 [02:27<21:05,  1.13s/it, loss=0.616]
Epoch 3/4:  10%|█         | 130/1250 [02:28<21:05,  1.13s/it, loss=0.516]
Epoch 3/4:  10%|█         | 131/1250 [02:28<20:30,  1.10s/it, loss=0.516]
Epoch 3/4:  10%|█         | 131/1250 [02:29<20:30,  1.10s/it, loss=0.475]
Epoch 3/4:  11%|█         | 132/1250 [02:29<20:20,  1.09s/it, loss=0.475]
Epoch 3/4:  11%|█         | 132/1250 [02:31<20:20,  1.09s/it, loss=0.609]
Epoch 3/4:  11%|█         | 133/1250 [02:31<20:39,  1.11s/it, loss=0.609]
Epoch 3/4:  11%|█         | 133/1250 [02:32<20:39,  1.11s/it, loss=0.623]
Epoch 3/4:  11%|█         | 134/1250 [02:32<20:49,  1.12s/it, loss=0.623]
Epoch 3/4:  11%|█         | 134/1250 [02:33<20:49,  1.12s/it, loss=0.545]
Epoch 3/4:  11%|█         | 135/1250 [02:33<20:45,  1.12s/it, loss=0.545]
Epoch 3/4:  11%|█         | 135/1250 [02:34<20:45,  1.12s/it, loss=0.473]
Epoch 3/4:  11%|█         | 136/1250 [02:34<20:52,  1.12s/it, loss=0.473]
Epoch 3/4:  11%|█         | 136/1250 [02:35<20:52,  1.12s/it, loss=0.51] 
Epoch 3/4:  11%|█         | 137/1250 [02:35<21:23,  1.15s/it, loss=0.51]
Epoch 3/4:  11%|█         | 137/1250 [02:36<21:23,  1.15s/it, loss=0.472]
Epoch 3/4:  11%|█         | 138/1250 [02:36<21:17,  1.15s/it, loss=0.472]
Epoch 3/4:  11%|█         | 138/1250 [02:37<21:17,  1.15s/it, loss=0.655]
Epoch 3/4:  11%|█         | 139/1250 [02:37<20:54,  1.13s/it, loss=0.655]
Epoch 3/4:  11%|█         | 139/1250 [02:38<20:54,  1.13s/it, loss=0.444]
Epoch 3/4:  11%|█         | 140/1250 [02:38<20:18,  1.10s/it, loss=0.444]
Epoch 3/4:  11%|█         | 140/1250 [02:40<20:18,  1.10s/it, loss=0.585]
Epoch 3/4:  11%|█▏        | 141/1250 [02:40<20:06,  1.09s/it, loss=0.585]
Epoch 3/4:  11%|█▏        | 141/1250 [02:41<20:06,  1.09s/it, loss=0.381]
Epoch 3/4:  11%|█▏        | 142/1250 [02:41<20:41,  1.12s/it, loss=0.381]
Epoch 3/4:  11%|█▏        | 142/1250 [02:42<20:41,  1.12s/it, loss=0.445]
Epoch 3/4:  11%|█▏        | 143/1250 [02:42<20:53,  1.13s/it, loss=0.445]
Epoch 3/4:  11%|█▏        | 143/1250 [02:43<20:53,  1.13s/it, loss=1.05] 
Epoch 3/4:  12%|█▏        | 144/1250 [02:43<20:36,  1.12s/it, loss=1.05]
Epoch 3/4:  12%|█▏        | 144/1250 [02:44<20:36,  1.12s/it, loss=0.593]
Epoch 3/4:  12%|█▏        | 145/1250 [02:44<20:02,  1.09s/it, loss=0.593]
Epoch 3/4:  12%|█▏        | 145/1250 [02:45<20:02,  1.09s/it, loss=0.585]
Epoch 3/4:  12%|█▏        | 146/1250 [02:45<20:32,  1.12s/it, loss=0.585]
Epoch 3/4:  12%|█▏        | 146/1250 [02:46<20:32,  1.12s/it, loss=0.396]
Epoch 3/4:  12%|█▏        | 147/1250 [02:46<20:59,  1.14s/it, loss=0.396]
Epoch 3/4:  12%|█▏        | 147/1250 [02:47<20:59,  1.14s/it, loss=0.471]
Epoch 3/4:  12%|█▏        | 148/1250 [02:47<20:53,  1.14s/it, loss=0.471]
Epoch 3/4:  12%|█▏        | 148/1250 [02:49<20:53,  1.14s/it, loss=0.643]
Epoch 3/4:  12%|█▏        | 149/1250 [02:49<20:27,  1.11s/it, loss=0.643]
Epoch 3/4:  12%|█▏        | 149/1250 [02:50<20:27,  1.11s/it, loss=0.449]
Epoch 3/4:  12%|█▏        | 150/1250 [02:50<20:08,  1.10s/it, loss=0.449]
Epoch 3/4:  12%|█▏        | 150/1250 [02:51<20:08,  1.10s/it, loss=0.758]
Epoch 3/4:  12%|█▏        | 151/1250 [02:51<20:17,  1.11s/it, loss=0.758]
Epoch 3/4:  12%|█▏        | 151/1250 [02:52<20:17,  1.11s/it, loss=0.39] 
Epoch 3/4:  12%|█▏        | 152/1250 [02:52<20:37,  1.13s/it, loss=0.39]
Epoch 3/4:  12%|█▏        | 152/1250 [02:53<20:37,  1.13s/it, loss=0.538]
Epoch 3/4:  12%|█▏        | 153/1250 [02:53<20:31,  1.12s/it, loss=0.538]
Epoch 3/4:  12%|█▏        | 153/1250 [02:54<20:31,  1.12s/it, loss=0.591]
Epoch 3/4:  12%|█▏        | 154/1250 [02:54<20:08,  1.10s/it, loss=0.591]
Epoch 3/4:  12%|█▏        | 154/1250 [02:55<20:08,  1.10s/it, loss=0.721]
Epoch 3/4:  12%|█▏        | 155/1250 [02:55<20:10,  1.11s/it, loss=0.721]
Epoch 3/4:  12%|█▏        | 155/1250 [02:56<20:10,  1.11s/it, loss=0.424]
Epoch 3/4:  12%|█▏        | 156/1250 [02:56<20:23,  1.12s/it, loss=0.424]
Epoch 3/4:  12%|█▏        | 156/1250 [02:57<20:23,  1.12s/it, loss=0.452]
Epoch 3/4:  13%|█▎        | 157/1250 [02:57<20:21,  1.12s/it, loss=0.452]
Epoch 3/4:  13%|█▎        | 157/1250 [02:58<20:21,  1.12s/it, loss=0.46] 
Epoch 3/4:  13%|█▎        | 158/1250 [02:58<19:57,  1.10s/it, loss=0.46]
Epoch 3/4:  13%|█▎        | 158/1250 [03:00<19:57,  1.10s/it, loss=0.422]
Epoch 3/4:  13%|█▎        | 159/1250 [03:00<19:58,  1.10s/it, loss=0.422]
Epoch 3/4:  13%|█▎        | 159/1250 [03:01<19:58,  1.10s/it, loss=0.362]
Epoch 3/4:  13%|█▎        | 160/1250 [03:01<20:35,  1.13s/it, loss=0.362]
Epoch 3/4:  13%|█▎        | 160/1250 [03:02<20:35,  1.13s/it, loss=0.678]
Epoch 3/4:  13%|█▎        | 161/1250 [03:02<20:46,  1.14s/it, loss=0.678]
Epoch 3/4:  13%|█▎        | 161/1250 [03:03<20:46,  1.14s/it, loss=0.522]
Epoch 3/4:  13%|█▎        | 162/1250 [03:03<20:17,  1.12s/it, loss=0.522]
Epoch 3/4:  13%|█▎        | 162/1250 [03:04<20:17,  1.12s/it, loss=0.721]
Epoch 3/4:  13%|█▎        | 163/1250 [03:04<19:39,  1.08s/it, loss=0.721]
Epoch 3/4:  13%|█▎        | 163/1250 [03:05<19:39,  1.08s/it, loss=0.44] 
Epoch 3/4:  13%|█▎        | 164/1250 [03:05<19:36,  1.08s/it, loss=0.44]
Epoch 3/4:  13%|█▎        | 164/1250 [03:06<19:36,  1.08s/it, loss=0.439]
Epoch 3/4:  13%|█▎        | 165/1250 [03:06<20:16,  1.12s/it, loss=0.439]
Epoch 3/4:  13%|█▎        | 165/1250 [03:07<20:16,  1.12s/it, loss=0.471]
Epoch 3/4:  13%|█▎        | 166/1250 [03:07<20:22,  1.13s/it, loss=0.471]
Epoch 3/4:  13%|█▎        | 166/1250 [03:09<20:22,  1.13s/it, loss=0.552]
Epoch 3/4:  13%|█▎        | 167/1250 [03:09<19:54,  1.10s/it, loss=0.552]
Epoch 3/4:  13%|█▎        | 167/1250 [03:10<19:54,  1.10s/it, loss=0.589]
Epoch 3/4:  13%|█▎        | 168/1250 [03:10<19:30,  1.08s/it, loss=0.589]
Epoch 3/4:  13%|█▎        | 168/1250 [03:11<19:30,  1.08s/it, loss=0.985]
Epoch 3/4:  14%|█▎        | 169/1250 [03:11<19:51,  1.10s/it, loss=0.985]
Epoch 3/4:  14%|█▎        | 169/1250 [03:12<19:51,  1.10s/it, loss=0.451]
Epoch 3/4:  14%|█▎        | 170/1250 [03:12<20:18,  1.13s/it, loss=0.451]
Epoch 3/4:  14%|█▎        | 170/1250 [03:13<20:18,  1.13s/it, loss=0.434]
Epoch 3/4:  14%|█▎        | 171/1250 [03:13<20:11,  1.12s/it, loss=0.434]
Epoch 3/4:  14%|█▎        | 171/1250 [03:14<20:11,  1.12s/it, loss=0.4]  
Epoch 3/4:  14%|█▍        | 172/1250 [03:14<19:47,  1.10s/it, loss=0.4]
Epoch 3/4:  14%|█▍        | 172/1250 [03:15<19:47,  1.10s/it, loss=0.611]
Epoch 3/4:  14%|█▍        | 173/1250 [03:15<19:58,  1.11s/it, loss=0.611]
Epoch 3/4:  14%|█▍        | 173/1250 [03:16<19:58,  1.11s/it, loss=0.591]
Epoch 3/4:  14%|█▍        | 174/1250 [03:16<20:30,  1.14s/it, loss=0.591]
Epoch 3/4:  14%|█▍        | 174/1250 [03:18<20:30,  1.14s/it, loss=0.435]
Epoch 3/4:  14%|█▍        | 175/1250 [03:18<20:28,  1.14s/it, loss=0.435]
Epoch 3/4:  14%|█▍        | 175/1250 [03:19<20:28,  1.14s/it, loss=0.483]
Epoch 3/4:  14%|█▍        | 176/1250 [03:19<19:52,  1.11s/it, loss=0.483]
Epoch 3/4:  14%|█▍        | 176/1250 [03:20<19:52,  1.11s/it, loss=0.715]
Epoch 3/4:  14%|█▍        | 177/1250 [03:20<19:24,  1.09s/it, loss=0.715]
Epoch 3/4:  14%|█▍        | 177/1250 [03:21<19:24,  1.09s/it, loss=0.551]
Epoch 3/4:  14%|█▍        | 178/1250 [03:21<19:27,  1.09s/it, loss=0.551]
Epoch 3/4:  14%|█▍        | 178/1250 [03:22<19:27,  1.09s/it, loss=0.49] 
Epoch 3/4:  14%|█▍        | 179/1250 [03:22<19:43,  1.10s/it, loss=0.49]
Epoch 3/4:  14%|█▍        | 179/1250 [03:23<19:43,  1.10s/it, loss=0.471]
Epoch 3/4:  14%|█▍        | 180/1250 [03:23<19:38,  1.10s/it, loss=0.471]
Epoch 3/4:  14%|█▍        | 180/1250 [03:24<19:38,  1.10s/it, loss=0.385]
Epoch 3/4:  14%|█▍        | 181/1250 [03:24<19:21,  1.09s/it, loss=0.385]
Epoch 3/4:  14%|█▍        | 181/1250 [03:25<19:21,  1.09s/it, loss=0.575]
Epoch 3/4:  15%|█▍        | 182/1250 [03:25<19:21,  1.09s/it, loss=0.575]
Epoch 3/4:  15%|█▍        | 182/1250 [03:26<19:21,  1.09s/it, loss=0.471]
Epoch 3/4:  15%|█▍        | 183/1250 [03:26<19:59,  1.12s/it, loss=0.471]
Epoch 3/4:  15%|█▍        | 183/1250 [03:28<19:59,  1.12s/it, loss=0.446]
Epoch 3/4:  15%|█▍        | 184/1250 [03:28<20:59,  1.18s/it, loss=0.446]
Epoch 3/4:  15%|█▍        | 184/1250 [03:29<20:59,  1.18s/it, loss=0.579]
Epoch 3/4:  15%|█▍        | 185/1250 [03:29<20:28,  1.15s/it, loss=0.579]
Epoch 3/4:  15%|█▍        | 185/1250 [03:30<20:28,  1.15s/it, loss=0.438]
Epoch 3/4:  15%|█▍        | 186/1250 [03:30<19:44,  1.11s/it, loss=0.438]
Epoch 3/4:  15%|█▍        | 186/1250 [03:31<19:44,  1.11s/it, loss=0.473]
Epoch 3/4:  15%|█▍        | 187/1250 [03:31<19:31,  1.10s/it, loss=0.473]
Epoch 3/4:  15%|█▍        | 187/1250 [03:32<19:31,  1.10s/it, loss=0.625]
Epoch 3/4:  15%|█▌        | 188/1250 [03:32<19:40,  1.11s/it, loss=0.625]
Epoch 3/4:  15%|█▌        | 188/1250 [03:33<19:40,  1.11s/it, loss=0.57] 
Epoch 3/4:  15%|█▌        | 189/1250 [03:33<19:37,  1.11s/it, loss=0.57]
Epoch 3/4:  15%|█▌        | 189/1250 [03:34<19:37,  1.11s/it, loss=0.64]
Epoch 3/4:  15%|█▌        | 190/1250 [03:34<19:15,  1.09s/it, loss=0.64]
Epoch 3/4:  15%|█▌        | 190/1250 [03:35<19:15,  1.09s/it, loss=0.526]
Epoch 3/4:  15%|█▌        | 191/1250 [03:35<19:08,  1.08s/it, loss=0.526]
Epoch 3/4:  15%|█▌        | 191/1250 [03:36<19:08,  1.08s/it, loss=0.843]
Epoch 3/4:  15%|█▌        | 192/1250 [03:36<19:32,  1.11s/it, loss=0.843]
Epoch 3/4:  15%|█▌        | 192/1250 [03:37<19:32,  1.11s/it, loss=0.427]
Epoch 3/4:  15%|█▌        | 193/1250 [03:37<19:51,  1.13s/it, loss=0.427]
Epoch 3/4:  15%|█▌        | 193/1250 [03:39<19:51,  1.13s/it, loss=0.737]
Epoch 3/4:  16%|█▌        | 194/1250 [03:39<19:35,  1.11s/it, loss=0.737]
Epoch 3/4:  16%|█▌        | 194/1250 [03:40<19:35,  1.11s/it, loss=0.587]
Epoch 3/4:  16%|█▌        | 195/1250 [03:40<19:17,  1.10s/it, loss=0.587]
Epoch 3/4:  16%|█▌        | 195/1250 [03:41<19:17,  1.10s/it, loss=0.641]
Epoch 3/4:  16%|█▌        | 196/1250 [03:41<19:21,  1.10s/it, loss=0.641]
Epoch 3/4:  16%|█▌        | 196/1250 [03:42<19:21,  1.10s/it, loss=0.612]
Epoch 3/4:  16%|█▌        | 197/1250 [03:42<19:28,  1.11s/it, loss=0.612]
Epoch 3/4:  16%|█▌        | 197/1250 [03:43<19:28,  1.11s/it, loss=0.505]
Epoch 3/4:  16%|█▌        | 198/1250 [03:43<19:21,  1.10s/it, loss=0.505]
Epoch 3/4:  16%|█▌        | 198/1250 [03:44<19:21,  1.10s/it, loss=0.56] 
Epoch 3/4:  16%|█▌        | 199/1250 [03:44<18:56,  1.08s/it, loss=0.56]
Epoch 3/4:  16%|█▌        | 199/1250 [03:45<18:56,  1.08s/it, loss=0.773]
Epoch 3/4:  16%|█▌        | 200/1250 [03:45<19:04,  1.09s/it, loss=0.773]
Epoch 3/4:  16%|█▌        | 200/1250 [03:46<19:04,  1.09s/it, loss=0.558]
Epoch 3/4:  16%|█▌        | 201/1250 [03:46<19:41,  1.13s/it, loss=0.558]
Epoch 3/4:  16%|█▌        | 201/1250 [03:47<19:41,  1.13s/it, loss=0.55] 
Epoch 3/4:  16%|█▌        | 202/1250 [03:47<19:43,  1.13s/it, loss=0.55]
Epoch 3/4:  16%|█▌        | 202/1250 [03:48<19:43,  1.13s/it, loss=0.53]
Epoch 3/4:  16%|█▌        | 203/1250 [03:48<19:15,  1.10s/it, loss=0.53]
Epoch 3/4:  16%|█▌        | 203/1250 [03:49<19:15,  1.10s/it, loss=0.454]
Epoch 3/4:  16%|█▋        | 204/1250 [03:49<18:44,  1.08s/it, loss=0.454]
Epoch 3/4:  16%|█▋        | 204/1250 [03:51<18:44,  1.08s/it, loss=0.447]
Epoch 3/4:  16%|█▋        | 205/1250 [03:51<18:45,  1.08s/it, loss=0.447]
Epoch 3/4:  16%|█▋        | 205/1250 [03:52<18:45,  1.08s/it, loss=0.458]
Epoch 3/4:  16%|█▋        | 206/1250 [03:52<19:20,  1.11s/it, loss=0.458]
Epoch 3/4:  16%|█▋        | 206/1250 [03:53<19:20,  1.11s/it, loss=0.556]
Epoch 3/4:  17%|█▋        | 207/1250 [03:53<19:43,  1.14s/it, loss=0.556]
Epoch 3/4:  17%|█▋        | 207/1250 [03:54<19:43,  1.14s/it, loss=0.566]
Epoch 3/4:  17%|█▋        | 208/1250 [03:54<19:24,  1.12s/it, loss=0.566]
Epoch 3/4:  17%|█▋        | 208/1250 [03:55<19:24,  1.12s/it, loss=0.626]
Epoch 3/4:  17%|█▋        | 209/1250 [03:55<19:09,  1.10s/it, loss=0.626]
Epoch 3/4:  17%|█▋        | 209/1250 [03:56<19:09,  1.10s/it, loss=0.637]
Epoch 3/4:  17%|█▋        | 210/1250 [03:56<19:10,  1.11s/it, loss=0.637]
Epoch 3/4:  17%|█▋        | 210/1250 [03:57<19:10,  1.11s/it, loss=0.368]
Epoch 3/4:  17%|█▋        | 211/1250 [03:57<19:13,  1.11s/it, loss=0.368]
Epoch 3/4:  17%|█▋        | 211/1250 [03:58<19:13,  1.11s/it, loss=0.455]
Epoch 3/4:  17%|█▋        | 212/1250 [03:58<18:54,  1.09s/it, loss=0.455]
Epoch 3/4:  17%|█▋        | 212/1250 [03:59<18:54,  1.09s/it, loss=0.618]
Epoch 3/4:  17%|█▋        | 213/1250 [03:59<18:25,  1.07s/it, loss=0.618]
Epoch 3/4:  17%|█▋        | 213/1250 [04:00<18:25,  1.07s/it, loss=0.573]
Epoch 3/4:  17%|█▋        | 214/1250 [04:00<18:21,  1.06s/it, loss=0.573]
Epoch 3/4:  17%|█▋        | 214/1250 [04:02<18:21,  1.06s/it, loss=0.805]
Epoch 3/4:  17%|█▋        | 215/1250 [04:02<18:55,  1.10s/it, loss=0.805]
Epoch 3/4:  17%|█▋        | 215/1250 [04:03<18:55,  1.10s/it, loss=0.486]
Epoch 3/4:  17%|█▋        | 216/1250 [04:03<19:07,  1.11s/it, loss=0.486]
Epoch 3/4:  17%|█▋        | 216/1250 [04:04<19:07,  1.11s/it, loss=0.541]
Epoch 3/4:  17%|█▋        | 217/1250 [04:04<19:00,  1.10s/it, loss=0.541]
Epoch 3/4:  17%|█▋        | 217/1250 [04:05<19:00,  1.10s/it, loss=0.461]
Epoch 3/4:  17%|█▋        | 218/1250 [04:05<18:39,  1.08s/it, loss=0.461]
Epoch 3/4:  17%|█▋        | 218/1250 [04:06<18:39,  1.08s/it, loss=0.58] 
Epoch 3/4:  18%|█▊        | 219/1250 [04:06<18:46,  1.09s/it, loss=0.58]
Epoch 3/4:  18%|█▊        | 219/1250 [04:07<18:46,  1.09s/it, loss=0.481]
Epoch 3/4:  18%|█▊        | 220/1250 [04:07<18:54,  1.10s/it, loss=0.481]
Epoch 3/4:  18%|█▊        | 220/1250 [04:08<18:54,  1.10s/it, loss=0.5]  
Epoch 3/4:  18%|█▊        | 221/1250 [04:08<18:45,  1.09s/it, loss=0.5]
Epoch 3/4:  18%|█▊        | 221/1250 [04:09<18:45,  1.09s/it, loss=0.64]
Epoch 3/4:  18%|█▊        | 222/1250 [04:09<18:23,  1.07s/it, loss=0.64]
Epoch 3/4:  18%|█▊        | 222/1250 [04:10<18:23,  1.07s/it, loss=0.551]
Epoch 3/4:  18%|█▊        | 223/1250 [04:10<18:16,  1.07s/it, loss=0.551]
Epoch 3/4:  18%|█▊        | 223/1250 [04:11<18:16,  1.07s/it, loss=0.414]
Epoch 3/4:  18%|█▊        | 224/1250 [04:11<18:53,  1.10s/it, loss=0.414]
Epoch 3/4:  18%|█▊        | 224/1250 [04:13<18:53,  1.10s/it, loss=0.476]
Epoch 3/4:  18%|█▊        | 225/1250 [04:13<19:17,  1.13s/it, loss=0.476]
Epoch 3/4:  18%|█▊        | 225/1250 [04:14<19:17,  1.13s/it, loss=0.614]
Epoch 3/4:  18%|█▊        | 226/1250 [04:14<18:49,  1.10s/it, loss=0.614]
Epoch 3/4:  18%|█▊        | 226/1250 [04:15<18:49,  1.10s/it, loss=0.461]
Epoch 3/4:  18%|█▊        | 227/1250 [04:15<18:26,  1.08s/it, loss=0.461]
Epoch 3/4:  18%|█▊        | 227/1250 [04:16<18:26,  1.08s/it, loss=0.509]
Epoch 3/4:  18%|█▊        | 228/1250 [04:16<18:41,  1.10s/it, loss=0.509]
Epoch 3/4:  18%|█▊        | 228/1250 [04:17<18:41,  1.10s/it, loss=0.294]
Epoch 3/4:  18%|█▊        | 229/1250 [04:17<19:07,  1.12s/it, loss=0.294]
Epoch 3/4:  18%|█▊        | 229/1250 [04:18<19:07,  1.12s/it, loss=0.388]
Epoch 3/4:  18%|█▊        | 230/1250 [04:18<19:13,  1.13s/it, loss=0.388]
Epoch 3/4:  18%|█▊        | 230/1250 [04:19<19:13,  1.13s/it, loss=0.384]
Epoch 3/4:  18%|█▊        | 231/1250 [04:19<18:53,  1.11s/it, loss=0.384]
Epoch 3/4:  18%|█▊        | 231/1250 [04:20<18:53,  1.11s/it, loss=0.464]
Epoch 3/4:  19%|█▊        | 232/1250 [04:20<18:36,  1.10s/it, loss=0.464]
Epoch 3/4:  19%|█▊        | 232/1250 [04:21<18:36,  1.10s/it, loss=0.732]
Epoch 3/4:  19%|█▊        | 233/1250 [04:21<18:36,  1.10s/it, loss=0.732]
Epoch 3/4:  19%|█▊        | 233/1250 [04:23<18:36,  1.10s/it, loss=0.569]
Epoch 3/4:  19%|█▊        | 234/1250 [04:23<18:43,  1.11s/it, loss=0.569]
Epoch 3/4:  19%|█▊        | 234/1250 [04:24<18:43,  1.11s/it, loss=0.547]
Epoch 3/4:  19%|█▉        | 235/1250 [04:24<18:29,  1.09s/it, loss=0.547]
Epoch 3/4:  19%|█▉        | 235/1250 [04:25<18:29,  1.09s/it, loss=0.64] 
Epoch 3/4:  19%|█▉        | 236/1250 [04:25<18:00,  1.07s/it, loss=0.64]
Epoch 3/4:  19%|█▉        | 236/1250 [04:26<18:00,  1.07s/it, loss=0.537]
Epoch 3/4:  19%|█▉        | 237/1250 [04:26<17:57,  1.06s/it, loss=0.537]
Epoch 3/4:  19%|█▉        | 237/1250 [04:27<17:57,  1.06s/it, loss=0.585]
Epoch 3/4:  19%|█▉        | 238/1250 [04:27<18:23,  1.09s/it, loss=0.585]
Epoch 3/4:  19%|█▉        | 238/1250 [04:28<18:23,  1.09s/it, loss=0.642]
Epoch 3/4:  19%|█▉        | 239/1250 [04:28<18:45,  1.11s/it, loss=0.642]
Epoch 3/4:  19%|█▉        | 239/1250 [04:29<18:45,  1.11s/it, loss=0.456]
Epoch 3/4:  19%|█▉        | 240/1250 [04:29<18:34,  1.10s/it, loss=0.456]
Epoch 3/4:  19%|█▉        | 240/1250 [04:30<18:34,  1.10s/it, loss=0.406]
Epoch 3/4:  19%|█▉        | 241/1250 [04:30<18:19,  1.09s/it, loss=0.406]
Epoch 3/4:  19%|█▉        | 241/1250 [04:31<18:19,  1.09s/it, loss=0.508]
Epoch 3/4:  19%|█▉        | 242/1250 [04:31<18:27,  1.10s/it, loss=0.508]
Epoch 3/4:  19%|█▉        | 242/1250 [04:32<18:27,  1.10s/it, loss=0.404]
Epoch 3/4:  19%|█▉        | 243/1250 [04:32<18:35,  1.11s/it, loss=0.404]
Epoch 3/4:  19%|█▉        | 243/1250 [04:33<18:35,  1.11s/it, loss=0.694]
Epoch 3/4:  20%|█▉        | 244/1250 [04:33<18:33,  1.11s/it, loss=0.694]
Epoch 3/4:  20%|█▉        | 244/1250 [04:35<18:33,  1.11s/it, loss=0.757]
Epoch 3/4:  20%|█▉        | 245/1250 [04:35<18:22,  1.10s/it, loss=0.757]
Epoch 3/4:  20%|█▉        | 245/1250 [04:36<18:22,  1.10s/it, loss=0.537]
Epoch 3/4:  20%|█▉        | 246/1250 [04:36<18:06,  1.08s/it, loss=0.537]
Epoch 3/4:  20%|█▉        | 246/1250 [04:37<18:06,  1.08s/it, loss=0.44] 
Epoch 3/4:  20%|█▉        | 247/1250 [04:37<18:14,  1.09s/it, loss=0.44]
Epoch 3/4:  20%|█▉        | 247/1250 [04:38<18:14,  1.09s/it, loss=0.59]
Epoch 3/4:  20%|█▉        | 248/1250 [04:38<18:23,  1.10s/it, loss=0.59]
Epoch 3/4:  20%|█▉        | 248/1250 [04:39<18:23,  1.10s/it, loss=0.462]
Epoch 3/4:  20%|█▉        | 249/1250 [04:39<18:09,  1.09s/it, loss=0.462]
Epoch 3/4:  20%|█▉        | 249/1250 [04:40<18:09,  1.09s/it, loss=0.639]
Epoch 3/4:  20%|██        | 250/1250 [04:40<17:46,  1.07s/it, loss=0.639]
Epoch 3/4:  20%|██        | 250/1250 [04:41<17:46,  1.07s/it, loss=0.537]
Epoch 3/4:  20%|██        | 251/1250 [04:41<17:44,  1.07s/it, loss=0.537]
Epoch 3/4:  20%|██        | 251/1250 [04:42<17:44,  1.07s/it, loss=0.513]
Epoch 3/4:  20%|██        | 252/1250 [04:42<18:06,  1.09s/it, loss=0.513]
Epoch 3/4:  20%|██        | 252/1250 [04:43<18:06,  1.09s/it, loss=0.488]
Epoch 3/4:  20%|██        | 253/1250 [04:43<18:17,  1.10s/it, loss=0.488]
Epoch 3/4:  20%|██        | 253/1250 [04:44<18:17,  1.10s/it, loss=0.597]
Epoch 3/4:  20%|██        | 254/1250 [04:44<18:02,  1.09s/it, loss=0.597]
Epoch 3/4:  20%|██        | 254/1250 [04:45<18:02,  1.09s/it, loss=0.592]
Epoch 3/4:  20%|██        | 255/1250 [04:45<18:18,  1.10s/it, loss=0.592]
Epoch 3/4:  20%|██        | 255/1250 [04:47<18:18,  1.10s/it, loss=0.519]
Epoch 3/4:  20%|██        | 256/1250 [04:47<18:42,  1.13s/it, loss=0.519]
Epoch 3/4:  20%|██        | 256/1250 [04:48<18:42,  1.13s/it, loss=0.466]
Epoch 3/4:  21%|██        | 257/1250 [04:48<18:42,  1.13s/it, loss=0.466]
Epoch 3/4:  21%|██        | 257/1250 [04:49<18:42,  1.13s/it, loss=0.662]
Epoch 3/4:  21%|██        | 258/1250 [04:49<18:18,  1.11s/it, loss=0.662]
Epoch 3/4:  21%|██        | 258/1250 [04:50<18:18,  1.11s/it, loss=0.671]
Epoch 3/4:  21%|██        | 259/1250 [04:50<17:47,  1.08s/it, loss=0.671]
Epoch 3/4:  21%|██        | 259/1250 [04:51<17:47,  1.08s/it, loss=0.531]
Epoch 3/4:  21%|██        | 260/1250 [04:51<17:36,  1.07s/it, loss=0.531]
Epoch 3/4:  21%|██        | 260/1250 [04:52<17:36,  1.07s/it, loss=0.475]
Epoch 3/4:  21%|██        | 261/1250 [04:52<17:50,  1.08s/it, loss=0.475]
Epoch 3/4:  21%|██        | 261/1250 [04:53<17:50,  1.08s/it, loss=0.528]
Epoch 3/4:  21%|██        | 262/1250 [04:53<18:05,  1.10s/it, loss=0.528]
Epoch 3/4:  21%|██        | 262/1250 [04:54<18:05,  1.10s/it, loss=0.56] 
Epoch 3/4:  21%|██        | 263/1250 [04:54<17:53,  1.09s/it, loss=0.56]
Epoch 3/4:  21%|██        | 263/1250 [04:55<17:53,  1.09s/it, loss=0.545]
Epoch 3/4:  21%|██        | 264/1250 [04:55<17:35,  1.07s/it, loss=0.545]
Epoch 3/4:  21%|██        | 264/1250 [04:56<17:35,  1.07s/it, loss=0.432]
Epoch 3/4:  21%|██        | 265/1250 [04:56<17:37,  1.07s/it, loss=0.432]
Epoch 3/4:  21%|██        | 265/1250 [04:57<17:37,  1.07s/it, loss=0.298]
Epoch 3/4:  21%|██▏       | 266/1250 [04:57<17:49,  1.09s/it, loss=0.298]
Epoch 3/4:  21%|██▏       | 266/1250 [04:59<17:49,  1.09s/it, loss=0.409]
Epoch 3/4:  21%|██▏       | 267/1250 [04:59<17:44,  1.08s/it, loss=0.409]
Epoch 3/4:  21%|██▏       | 267/1250 [05:00<17:44,  1.08s/it, loss=0.627]
Epoch 3/4:  21%|██▏       | 268/1250 [05:00<17:24,  1.06s/it, loss=0.627]
Epoch 3/4:  21%|██▏       | 268/1250 [05:01<17:24,  1.06s/it, loss=0.41] 
Epoch 3/4:  22%|██▏       | 269/1250 [05:01<17:05,  1.05s/it, loss=0.41]
Epoch 3/4:  22%|██▏       | 269/1250 [05:02<17:05,  1.05s/it, loss=0.508]
Epoch 3/4:  22%|██▏       | 270/1250 [05:02<17:19,  1.06s/it, loss=0.508]
Epoch 3/4:  22%|██▏       | 270/1250 [05:03<17:19,  1.06s/it, loss=0.528]
Epoch 3/4:  22%|██▏       | 271/1250 [05:03<17:43,  1.09s/it, loss=0.528]
Epoch 3/4:  22%|██▏       | 271/1250 [05:04<17:43,  1.09s/it, loss=0.806]
Epoch 3/4:  22%|██▏       | 272/1250 [05:04<17:32,  1.08s/it, loss=0.806]
Epoch 3/4:  22%|██▏       | 272/1250 [05:05<17:32,  1.08s/it, loss=0.536]
Epoch 3/4:  22%|██▏       | 273/1250 [05:05<17:01,  1.05s/it, loss=0.536]
Epoch 3/4:  22%|██▏       | 273/1250 [05:06<17:01,  1.05s/it, loss=0.543]
Epoch 3/4:  22%|██▏       | 274/1250 [05:06<16:59,  1.04s/it, loss=0.543]
Epoch 3/4:  22%|██▏       | 274/1250 [05:07<16:59,  1.04s/it, loss=0.593]
Epoch 3/4:  22%|██▏       | 275/1250 [05:07<17:29,  1.08s/it, loss=0.593]
Epoch 3/4:  22%|██▏       | 275/1250 [05:08<17:29,  1.08s/it, loss=0.607]
Epoch 3/4:  22%|██▏       | 276/1250 [05:08<17:48,  1.10s/it, loss=0.607]
Epoch 3/4:  22%|██▏       | 276/1250 [05:09<17:48,  1.10s/it, loss=0.623]
Epoch 3/4:  22%|██▏       | 277/1250 [05:09<17:40,  1.09s/it, loss=0.623]
Epoch 3/4:  22%|██▏       | 277/1250 [05:10<17:40,  1.09s/it, loss=0.644]
Epoch 3/4:  22%|██▏       | 278/1250 [05:10<17:35,  1.09s/it, loss=0.644]
Epoch 3/4:  22%|██▏       | 278/1250 [05:11<17:35,  1.09s/it, loss=0.387]
Epoch 3/4:  22%|██▏       | 279/1250 [05:11<18:07,  1.12s/it, loss=0.387]
Epoch 3/4:  22%|██▏       | 279/1250 [05:13<18:07,  1.12s/it, loss=0.428]
Epoch 3/4:  22%|██▏       | 280/1250 [05:13<18:03,  1.12s/it, loss=0.428]
Epoch 3/4:  22%|██▏       | 280/1250 [05:14<18:03,  1.12s/it, loss=0.484]
Epoch 3/4:  22%|██▏       | 281/1250 [05:14<17:45,  1.10s/it, loss=0.484]
Epoch 3/4:  22%|██▏       | 281/1250 [05:15<17:45,  1.10s/it, loss=0.523]
Epoch 3/4:  23%|██▎       | 282/1250 [05:15<17:15,  1.07s/it, loss=0.523]
Epoch 3/4:  23%|██▎       | 282/1250 [05:16<17:15,  1.07s/it, loss=0.373]
Epoch 3/4:  23%|██▎       | 283/1250 [05:16<17:01,  1.06s/it, loss=0.373]
Epoch 3/4:  23%|██▎       | 283/1250 [05:17<17:01,  1.06s/it, loss=0.424]
Epoch 3/4:  23%|██▎       | 284/1250 [05:17<17:15,  1.07s/it, loss=0.424]
Epoch 3/4:  23%|██▎       | 284/1250 [05:18<17:15,  1.07s/it, loss=0.491]
Epoch 3/4:  23%|██▎       | 285/1250 [05:18<17:29,  1.09s/it, loss=0.491]
Epoch 3/4:  23%|██▎       | 285/1250 [05:19<17:29,  1.09s/it, loss=0.571]
Epoch 3/4:  23%|██▎       | 286/1250 [05:19<17:23,  1.08s/it, loss=0.571]
Epoch 3/4:  23%|██▎       | 286/1250 [05:20<17:23,  1.08s/it, loss=0.526]
Epoch 3/4:  23%|██▎       | 287/1250 [05:20<17:04,  1.06s/it, loss=0.526]
Epoch 3/4:  23%|██▎       | 287/1250 [05:21<17:04,  1.06s/it, loss=0.484]
Epoch 3/4:  23%|██▎       | 288/1250 [05:21<17:29,  1.09s/it, loss=0.484]
Epoch 3/4:  23%|██▎       | 288/1250 [05:22<17:29,  1.09s/it, loss=0.487]
Epoch 3/4:  23%|██▎       | 289/1250 [05:22<18:07,  1.13s/it, loss=0.487]
Epoch 3/4:  23%|██▎       | 289/1250 [05:23<18:07,  1.13s/it, loss=0.469]
Epoch 3/4:  23%|██▎       | 290/1250 [05:23<18:00,  1.13s/it, loss=0.469]
Epoch 3/4:  23%|██▎       | 290/1250 [05:24<18:00,  1.13s/it, loss=0.413]
Epoch 3/4:  23%|██▎       | 291/1250 [05:24<17:21,  1.09s/it, loss=0.413]
Epoch 3/4:  23%|██▎       | 291/1250 [05:25<17:21,  1.09s/it, loss=0.696]
Epoch 3/4:  23%|██▎       | 292/1250 [05:25<16:49,  1.05s/it, loss=0.696]
Epoch 3/4:  23%|██▎       | 292/1250 [05:27<16:49,  1.05s/it, loss=0.474]
Epoch 3/4:  23%|██▎       | 293/1250 [05:27<16:52,  1.06s/it, loss=0.474]
Epoch 3/4:  23%|██▎       | 293/1250 [05:28<16:52,  1.06s/it, loss=0.503]
Epoch 3/4:  24%|██▎       | 294/1250 [05:28<17:15,  1.08s/it, loss=0.503]
Epoch 3/4:  24%|██▎       | 294/1250 [05:29<17:15,  1.08s/it, loss=0.486]
Epoch 3/4:  24%|██▎       | 295/1250 [05:29<17:18,  1.09s/it, loss=0.486]
Epoch 3/4:  24%|██▎       | 295/1250 [05:30<17:18,  1.09s/it, loss=0.736]
Epoch 3/4:  24%|██▎       | 296/1250 [05:30<16:54,  1.06s/it, loss=0.736]
Epoch 3/4:  24%|██▎       | 296/1250 [05:31<16:54,  1.06s/it, loss=0.375]
Epoch 3/4:  24%|██▍       | 297/1250 [05:31<16:33,  1.04s/it, loss=0.375]
Epoch 3/4:  24%|██▍       | 297/1250 [05:32<16:33,  1.04s/it, loss=0.494]
Epoch 3/4:  24%|██▍       | 298/1250 [05:32<16:44,  1.05s/it, loss=0.494]
Epoch 3/4:  24%|██▍       | 298/1250 [05:33<16:44,  1.05s/it, loss=0.625]
Epoch 3/4:  24%|██▍       | 299/1250 [05:33<17:06,  1.08s/it, loss=0.625]
Epoch 3/4:  24%|██▍       | 299/1250 [05:34<17:06,  1.08s/it, loss=0.607]
Epoch 3/4:  24%|██▍       | 300/1250 [05:34<17:06,  1.08s/it, loss=0.607]
Epoch 3/4:  24%|██▍       | 300/1250 [05:35<17:06,  1.08s/it, loss=0.943]
Epoch 3/4:  24%|██▍       | 301/1250 [05:35<17:00,  1.08s/it, loss=0.943]
Epoch 3/4:  24%|██▍       | 301/1250 [05:36<17:00,  1.08s/it, loss=0.599]
Epoch 3/4:  24%|██▍       | 302/1250 [05:36<18:07,  1.15s/it, loss=0.599]
Epoch 3/4:  24%|██▍       | 302/1250 [05:38<18:07,  1.15s/it, loss=0.418]
Epoch 3/4:  24%|██▍       | 303/1250 [05:38<17:40,  1.12s/it, loss=0.418]
Epoch 3/4:  24%|██▍       | 303/1250 [05:39<17:40,  1.12s/it, loss=0.508]
Epoch 3/4:  24%|██▍       | 304/1250 [05:39<17:29,  1.11s/it, loss=0.508]
Epoch 3/4:  24%|██▍       | 304/1250 [05:40<17:29,  1.11s/it, loss=0.504]
Epoch 3/4:  24%|██▍       | 305/1250 [05:40<17:04,  1.08s/it, loss=0.504]
Epoch 3/4:  24%|██▍       | 305/1250 [05:41<17:04,  1.08s/it, loss=0.433]
Epoch 3/4:  24%|██▍       | 306/1250 [05:41<16:31,  1.05s/it, loss=0.433]
Epoch 3/4:  24%|██▍       | 306/1250 [05:42<16:31,  1.05s/it, loss=0.43] 
Epoch 3/4:  25%|██▍       | 307/1250 [05:42<16:23,  1.04s/it, loss=0.43]
Epoch 3/4:  25%|██▍       | 307/1250 [05:43<16:23,  1.04s/it, loss=0.586]
Epoch 3/4:  25%|██▍       | 308/1250 [05:43<16:42,  1.06s/it, loss=0.586]
Epoch 3/4:  25%|██▍       | 308/1250 [05:44<16:42,  1.06s/it, loss=0.485]
Epoch 3/4:  25%|██▍       | 309/1250 [05:44<16:49,  1.07s/it, loss=0.485]
Epoch 3/4:  25%|██▍       | 309/1250 [05:45<16:49,  1.07s/it, loss=0.445]
Epoch 3/4:  25%|██▍       | 310/1250 [05:45<16:41,  1.06s/it, loss=0.445]
Epoch 3/4:  25%|██▍       | 310/1250 [05:46<16:41,  1.06s/it, loss=0.693]
Epoch 3/4:  25%|██▍       | 311/1250 [05:46<16:39,  1.06s/it, loss=0.693]
Epoch 3/4:  25%|██▍       | 311/1250 [05:47<16:39,  1.06s/it, loss=0.493]
Epoch 3/4:  25%|██▍       | 312/1250 [05:47<17:04,  1.09s/it, loss=0.493]
Epoch 3/4:  25%|██▍       | 312/1250 [05:48<17:04,  1.09s/it, loss=0.556]
Epoch 3/4:  25%|██▌       | 313/1250 [05:48<17:30,  1.12s/it, loss=0.556]
Epoch 3/4:  25%|██▌       | 313/1250 [05:49<17:30,  1.12s/it, loss=0.712]
Epoch 3/4:  25%|██▌       | 314/1250 [05:49<17:13,  1.10s/it, loss=0.712]
Epoch 3/4:  25%|██▌       | 314/1250 [05:50<17:13,  1.10s/it, loss=0.658]
Epoch 3/4:  25%|██▌       | 315/1250 [05:50<16:35,  1.06s/it, loss=0.658]
Epoch 3/4:  25%|██▌       | 315/1250 [05:51<16:35,  1.06s/it, loss=0.571]
Epoch 3/4:  25%|██▌       | 316/1250 [05:51<16:10,  1.04s/it, loss=0.571]
Epoch 3/4:  25%|██▌       | 316/1250 [05:52<16:10,  1.04s/it, loss=0.487]
Epoch 3/4:  25%|██▌       | 317/1250 [05:52<16:33,  1.06s/it, loss=0.487]
Epoch 3/4:  25%|██▌       | 317/1250 [05:54<16:33,  1.06s/it, loss=0.579]
Epoch 3/4:  25%|██▌       | 318/1250 [05:54<16:57,  1.09s/it, loss=0.579]
Epoch 3/4:  25%|██▌       | 318/1250 [05:55<16:57,  1.09s/it, loss=0.524]
Epoch 3/4:  26%|██▌       | 319/1250 [05:55<16:36,  1.07s/it, loss=0.524]
Epoch 3/4:  26%|██▌       | 319/1250 [05:56<16:36,  1.07s/it, loss=0.928]
Epoch 3/4:  26%|██▌       | 320/1250 [05:56<16:05,  1.04s/it, loss=0.928]
Epoch 3/4:  26%|██▌       | 320/1250 [05:57<16:05,  1.04s/it, loss=0.632]
Epoch 3/4:  26%|██▌       | 321/1250 [05:57<15:57,  1.03s/it, loss=0.632]
Epoch 3/4:  26%|██▌       | 321/1250 [05:58<15:57,  1.03s/it, loss=0.333]
Epoch 3/4:  26%|██▌       | 322/1250 [05:58<16:38,  1.08s/it, loss=0.333]
Epoch 3/4:  26%|██▌       | 322/1250 [05:59<16:38,  1.08s/it, loss=0.375]
Epoch 3/4:  26%|██▌       | 323/1250 [05:59<17:22,  1.12s/it, loss=0.375]
Epoch 3/4:  26%|██▌       | 323/1250 [06:00<17:22,  1.12s/it, loss=1.13] 
Epoch 3/4:  26%|██▌       | 324/1250 [06:00<16:57,  1.10s/it, loss=1.13]
Epoch 3/4:  26%|██▌       | 324/1250 [06:01<16:57,  1.10s/it, loss=0.657]
Epoch 3/4:  26%|██▌       | 325/1250 [06:01<16:15,  1.05s/it, loss=0.657]
Epoch 3/4:  26%|██▌       | 325/1250 [06:02<16:15,  1.05s/it, loss=0.651]
Epoch 3/4:  26%|██▌       | 326/1250 [06:02<16:00,  1.04s/it, loss=0.651]
Epoch 3/4:  26%|██▌       | 326/1250 [06:03<16:00,  1.04s/it, loss=0.689]
Epoch 3/4:  26%|██▌       | 327/1250 [06:03<16:16,  1.06s/it, loss=0.689]
Epoch 3/4:  26%|██▌       | 327/1250 [06:04<16:16,  1.06s/it, loss=0.538]
Epoch 3/4:  26%|██▌       | 328/1250 [06:04<16:19,  1.06s/it, loss=0.538]
Epoch 3/4:  26%|██▌       | 328/1250 [06:05<16:19,  1.06s/it, loss=0.73] 
Epoch 3/4:  26%|██▋       | 329/1250 [06:05<16:04,  1.05s/it, loss=0.73]
Epoch 3/4:  26%|██▋       | 329/1250 [06:06<16:04,  1.05s/it, loss=0.494]
Epoch 3/4:  26%|██▋       | 330/1250 [06:06<15:53,  1.04s/it, loss=0.494]
Epoch 3/4:  26%|██▋       | 330/1250 [06:07<15:53,  1.04s/it, loss=0.584]
Epoch 3/4:  26%|██▋       | 331/1250 [06:07<16:20,  1.07s/it, loss=0.584]
Epoch 3/4:  26%|██▋       | 331/1250 [06:09<16:20,  1.07s/it, loss=0.378]
Epoch 3/4:  27%|██▋       | 332/1250 [06:09<16:58,  1.11s/it, loss=0.378]
Epoch 3/4:  27%|██▋       | 332/1250 [06:10<16:58,  1.11s/it, loss=0.695]
Epoch 3/4:  27%|██▋       | 333/1250 [06:10<16:50,  1.10s/it, loss=0.695]
Epoch 3/4:  27%|██▋       | 333/1250 [06:11<16:50,  1.10s/it, loss=0.338]
Epoch 3/4:  27%|██▋       | 334/1250 [06:11<16:19,  1.07s/it, loss=0.338]
Epoch 3/4:  27%|██▋       | 334/1250 [06:12<16:19,  1.07s/it, loss=0.481]
Epoch 3/4:  27%|██▋       | 335/1250 [06:12<16:05,  1.06s/it, loss=0.481]
Epoch 3/4:  27%|██▋       | 335/1250 [06:13<16:05,  1.06s/it, loss=0.627]
Epoch 3/4:  27%|██▋       | 336/1250 [06:13<16:20,  1.07s/it, loss=0.627]
Epoch 3/4:  27%|██▋       | 336/1250 [06:14<16:20,  1.07s/it, loss=0.658]
Epoch 3/4:  27%|██▋       | 337/1250 [06:14<16:28,  1.08s/it, loss=0.658]
Epoch 3/4:  27%|██▋       | 337/1250 [06:15<16:28,  1.08s/it, loss=0.479]
Epoch 3/4:  27%|██▋       | 338/1250 [06:15<16:20,  1.08s/it, loss=0.479]
Epoch 3/4:  27%|██▋       | 338/1250 [06:16<16:20,  1.08s/it, loss=0.493]
Epoch 3/4:  27%|██▋       | 339/1250 [06:16<15:51,  1.04s/it, loss=0.493]
Epoch 3/4:  27%|██▋       | 339/1250 [06:17<15:51,  1.04s/it, loss=0.556]
Epoch 3/4:  27%|██▋       | 340/1250 [06:17<15:49,  1.04s/it, loss=0.556]
Epoch 3/4:  27%|██▋       | 340/1250 [06:18<15:49,  1.04s/it, loss=0.573]
Epoch 3/4:  27%|██▋       | 341/1250 [06:18<16:06,  1.06s/it, loss=0.573]
Epoch 3/4:  27%|██▋       | 341/1250 [06:19<16:06,  1.06s/it, loss=0.403]
Epoch 3/4:  27%|██▋       | 342/1250 [06:19<16:17,  1.08s/it, loss=0.403]
Epoch 3/4:  27%|██▋       | 342/1250 [06:20<16:17,  1.08s/it, loss=0.649]
Epoch 3/4:  27%|██▋       | 343/1250 [06:20<15:56,  1.05s/it, loss=0.649]
Epoch 3/4:  27%|██▋       | 343/1250 [06:21<15:56,  1.05s/it, loss=0.554]
Epoch 3/4:  28%|██▊       | 344/1250 [06:21<15:44,  1.04s/it, loss=0.554]
Epoch 3/4:  28%|██▊       | 344/1250 [06:22<15:44,  1.04s/it, loss=0.741]
Epoch 3/4:  28%|██▊       | 345/1250 [06:22<16:28,  1.09s/it, loss=0.741]
Epoch 3/4:  28%|██▊       | 345/1250 [06:24<16:28,  1.09s/it, loss=0.761]
Epoch 3/4:  28%|██▊       | 346/1250 [06:24<16:49,  1.12s/it, loss=0.761]
Epoch 3/4:  28%|██▊       | 346/1250 [06:25<16:49,  1.12s/it, loss=0.627]
Epoch 3/4:  28%|██▊       | 347/1250 [06:25<16:33,  1.10s/it, loss=0.627]
Epoch 3/4:  28%|██▊       | 347/1250 [06:26<16:33,  1.10s/it, loss=0.382]
Epoch 3/4:  28%|██▊       | 348/1250 [06:26<16:02,  1.07s/it, loss=0.382]
Epoch 3/4:  28%|██▊       | 348/1250 [06:27<16:02,  1.07s/it, loss=0.722]
Epoch 3/4:  28%|██▊       | 349/1250 [06:27<15:46,  1.05s/it, loss=0.722]
Epoch 3/4:  28%|██▊       | 349/1250 [06:28<15:46,  1.05s/it, loss=0.41] 
Epoch 3/4:  28%|██▊       | 350/1250 [06:28<15:50,  1.06s/it, loss=0.41]
Epoch 3/4:  28%|██▊       | 350/1250 [06:29<15:50,  1.06s/it, loss=0.425]
Epoch 3/4:  28%|██▊       | 351/1250 [06:29<16:02,  1.07s/it, loss=0.425]
Epoch 3/4:  28%|██▊       | 351/1250 [06:30<16:02,  1.07s/it, loss=0.685]
Epoch 3/4:  28%|██▊       | 352/1250 [06:30<15:54,  1.06s/it, loss=0.685]
Epoch 3/4:  28%|██▊       | 352/1250 [06:31<15:54,  1.06s/it, loss=0.658]
Epoch 3/4:  28%|██▊       | 353/1250 [06:31<15:29,  1.04s/it, loss=0.658]
Epoch 3/4:  28%|██▊       | 353/1250 [06:32<15:29,  1.04s/it, loss=0.546]
Epoch 3/4:  28%|██▊       | 354/1250 [06:32<15:26,  1.03s/it, loss=0.546]
Epoch 3/4:  28%|██▊       | 354/1250 [06:33<15:26,  1.03s/it, loss=0.538]
Epoch 3/4:  28%|██▊       | 355/1250 [06:33<15:50,  1.06s/it, loss=0.538]
Epoch 3/4:  28%|██▊       | 355/1250 [06:34<15:50,  1.06s/it, loss=0.615]
Epoch 3/4:  28%|██▊       | 356/1250 [06:34<16:12,  1.09s/it, loss=0.615]
Epoch 3/4:  28%|██▊       | 356/1250 [06:35<16:12,  1.09s/it, loss=0.804]
Epoch 3/4:  29%|██▊       | 357/1250 [06:35<16:05,  1.08s/it, loss=0.804]
Epoch 3/4:  29%|██▊       | 357/1250 [06:36<16:05,  1.08s/it, loss=0.491]
Epoch 3/4:  29%|██▊       | 358/1250 [06:36<15:56,  1.07s/it, loss=0.491]
Epoch 3/4:  29%|██▊       | 358/1250 [06:37<15:56,  1.07s/it, loss=0.461]
Epoch 3/4:  29%|██▊       | 359/1250 [06:37<15:58,  1.08s/it, loss=0.461]
Epoch 3/4:  29%|██▊       | 359/1250 [06:38<15:58,  1.08s/it, loss=0.503]
Epoch 3/4:  29%|██▉       | 360/1250 [06:38<16:16,  1.10s/it, loss=0.503]
Epoch 3/4:  29%|██▉       | 360/1250 [06:40<16:16,  1.10s/it, loss=0.425]
Epoch 3/4:  29%|██▉       | 361/1250 [06:40<16:15,  1.10s/it, loss=0.425]
Epoch 3/4:  29%|██▉       | 361/1250 [06:41<16:15,  1.10s/it, loss=0.577]
Epoch 3/4:  29%|██▉       | 362/1250 [06:41<15:50,  1.07s/it, loss=0.577]
Epoch 3/4:  29%|██▉       | 362/1250 [06:42<15:50,  1.07s/it, loss=0.846]
Epoch 3/4:  29%|██▉       | 363/1250 [06:42<15:32,  1.05s/it, loss=0.846]
Epoch 3/4:  29%|██▉       | 363/1250 [06:43<15:32,  1.05s/it, loss=0.501]
Epoch 3/4:  29%|██▉       | 364/1250 [06:43<15:43,  1.06s/it, loss=0.501]
Epoch 3/4:  29%|██▉       | 364/1250 [06:44<15:43,  1.06s/it, loss=0.448]
Epoch 3/4:  29%|██▉       | 365/1250 [06:44<16:08,  1.09s/it, loss=0.448]
Epoch 3/4:  29%|██▉       | 365/1250 [06:45<16:08,  1.09s/it, loss=0.53] 
Epoch 3/4:  29%|██▉       | 366/1250 [06:45<16:27,  1.12s/it, loss=0.53]
Epoch 3/4:  29%|██▉       | 366/1250 [06:46<16:27,  1.12s/it, loss=0.513]
Epoch 3/4:  29%|██▉       | 367/1250 [06:46<16:19,  1.11s/it, loss=0.513]
Epoch 3/4:  29%|██▉       | 367/1250 [06:47<16:19,  1.11s/it, loss=0.635]
Epoch 3/4:  29%|██▉       | 368/1250 [06:47<15:58,  1.09s/it, loss=0.635]
Epoch 3/4:  29%|██▉       | 368/1250 [06:48<15:58,  1.09s/it, loss=0.448]
Epoch 3/4:  30%|██▉       | 369/1250 [06:48<15:52,  1.08s/it, loss=0.448]
Epoch 3/4:  30%|██▉       | 369/1250 [06:49<15:52,  1.08s/it, loss=0.535]
Epoch 3/4:  30%|██▉       | 370/1250 [06:49<15:44,  1.07s/it, loss=0.535]
Epoch 3/4:  30%|██▉       | 370/1250 [06:50<15:44,  1.07s/it, loss=0.583]
Epoch 3/4:  30%|██▉       | 371/1250 [06:50<15:34,  1.06s/it, loss=0.583]
Epoch 3/4:  30%|██▉       | 371/1250 [06:51<15:34,  1.06s/it, loss=0.598]
Epoch 3/4:  30%|██▉       | 372/1250 [06:51<15:28,  1.06s/it, loss=0.598]
Epoch 3/4:  30%|██▉       | 372/1250 [06:52<15:28,  1.06s/it, loss=0.82] 
Epoch 3/4:  30%|██▉       | 373/1250 [06:52<15:27,  1.06s/it, loss=0.82]
Epoch 3/4:  30%|██▉       | 373/1250 [06:54<15:27,  1.06s/it, loss=0.663]
Epoch 3/4:  30%|██▉       | 374/1250 [06:54<15:47,  1.08s/it, loss=0.663]
Epoch 3/4:  30%|██▉       | 374/1250 [06:55<15:47,  1.08s/it, loss=0.682]
Epoch 3/4:  30%|███       | 375/1250 [06:55<15:49,  1.08s/it, loss=0.682]
Epoch 3/4:  30%|███       | 375/1250 [06:56<15:49,  1.08s/it, loss=0.763]
Epoch 3/4:  30%|███       | 376/1250 [06:56<15:29,  1.06s/it, loss=0.763]
Epoch 3/4:  30%|███       | 376/1250 [06:57<15:29,  1.06s/it, loss=0.828]
Epoch 3/4:  30%|███       | 377/1250 [06:57<15:11,  1.04s/it, loss=0.828]
Epoch 3/4:  30%|███       | 377/1250 [06:58<15:11,  1.04s/it, loss=0.564]
Epoch 3/4:  30%|███       | 378/1250 [06:58<15:23,  1.06s/it, loss=0.564]
Epoch 3/4:  30%|███       | 378/1250 [06:59<15:23,  1.06s/it, loss=0.746]
Epoch 3/4:  30%|███       | 379/1250 [06:59<16:17,  1.12s/it, loss=0.746]
Epoch 3/4:  30%|███       | 379/1250 [07:00<16:17,  1.12s/it, loss=0.548]
Epoch 3/4:  30%|███       | 380/1250 [07:00<16:26,  1.13s/it, loss=0.548]
Epoch 3/4:  30%|███       | 380/1250 [07:01<16:26,  1.13s/it, loss=0.624]
Epoch 3/4:  30%|███       | 381/1250 [07:01<15:56,  1.10s/it, loss=0.624]
Epoch 3/4:  30%|███       | 381/1250 [07:02<15:56,  1.10s/it, loss=0.59] 
Epoch 3/4:  31%|███       | 382/1250 [07:02<15:25,  1.07s/it, loss=0.59]
Epoch 3/4:  31%|███       | 382/1250 [07:03<15:25,  1.07s/it, loss=0.585]
Epoch 3/4:  31%|███       | 383/1250 [07:03<15:23,  1.06s/it, loss=0.585]
Epoch 3/4:  31%|███       | 383/1250 [07:04<15:23,  1.06s/it, loss=0.538]
Epoch 3/4:  31%|███       | 384/1250 [07:04<15:31,  1.08s/it, loss=0.538]
Epoch 3/4:  31%|███       | 384/1250 [07:05<15:31,  1.08s/it, loss=0.51] 
Epoch 3/4:  31%|███       | 385/1250 [07:05<15:16,  1.06s/it, loss=0.51]
Epoch 3/4:  31%|███       | 385/1250 [07:06<15:16,  1.06s/it, loss=0.559]
Epoch 3/4:  31%|███       | 386/1250 [07:06<14:57,  1.04s/it, loss=0.559]
Epoch 3/4:  31%|███       | 386/1250 [07:07<14:57,  1.04s/it, loss=0.432]
Epoch 3/4:  31%|███       | 387/1250 [07:07<14:54,  1.04s/it, loss=0.432]
Epoch 3/4:  31%|███       | 387/1250 [07:09<14:54,  1.04s/it, loss=0.493]
Epoch 3/4:  31%|███       | 388/1250 [07:09<15:29,  1.08s/it, loss=0.493]
Epoch 3/4:  31%|███       | 388/1250 [07:10<15:29,  1.08s/it, loss=0.402]
Epoch 3/4:  31%|███       | 389/1250 [07:10<15:50,  1.10s/it, loss=0.402]
Epoch 3/4:  31%|███       | 389/1250 [07:11<15:50,  1.10s/it, loss=0.628]
Epoch 3/4:  31%|███       | 390/1250 [07:11<15:40,  1.09s/it, loss=0.628]
Epoch 3/4:  31%|███       | 390/1250 [07:12<15:40,  1.09s/it, loss=0.34] 
Epoch 3/4:  31%|███▏      | 391/1250 [07:12<15:21,  1.07s/it, loss=0.34]
Epoch 3/4:  31%|███▏      | 391/1250 [07:13<15:21,  1.07s/it, loss=0.548]
Epoch 3/4:  31%|███▏      | 392/1250 [07:13<15:24,  1.08s/it, loss=0.548]
Epoch 3/4:  31%|███▏      | 392/1250 [07:14<15:24,  1.08s/it, loss=0.494]
Epoch 3/4:  31%|███▏      | 393/1250 [07:14<15:45,  1.10s/it, loss=0.494]
Epoch 3/4:  31%|███▏      | 393/1250 [07:15<15:45,  1.10s/it, loss=0.641]
Epoch 3/4:  32%|███▏      | 394/1250 [07:15<15:54,  1.12s/it, loss=0.641]
Epoch 3/4:  32%|███▏      | 394/1250 [07:16<15:54,  1.12s/it, loss=0.498]
Epoch 3/4:  32%|███▏      | 395/1250 [07:16<15:33,  1.09s/it, loss=0.498]
Epoch 3/4:  32%|███▏      | 395/1250 [07:17<15:33,  1.09s/it, loss=0.649]
Epoch 3/4:  32%|███▏      | 396/1250 [07:17<15:05,  1.06s/it, loss=0.649]
Epoch 3/4:  32%|███▏      | 396/1250 [07:18<15:05,  1.06s/it, loss=0.548]
Epoch 3/4:  32%|███▏      | 397/1250 [07:18<15:14,  1.07s/it, loss=0.548]
Epoch 3/4:  32%|███▏      | 397/1250 [07:19<15:14,  1.07s/it, loss=0.37] 
Epoch 3/4:  32%|███▏      | 398/1250 [07:19<15:34,  1.10s/it, loss=0.37]
Epoch 3/4:  32%|███▏      | 398/1250 [07:21<15:34,  1.10s/it, loss=0.515]
Epoch 3/4:  32%|███▏      | 399/1250 [07:21<15:28,  1.09s/it, loss=0.515]
Epoch 3/4:  32%|███▏      | 399/1250 [07:22<15:28,  1.09s/it, loss=0.571]
Epoch 3/4:  32%|███▏      | 400/1250 [07:22<14:58,  1.06s/it, loss=0.571]
Epoch 3/4:  32%|███▏      | 400/1250 [07:23<14:58,  1.06s/it, loss=0.624]
Epoch 3/4:  32%|███▏      | 401/1250 [07:23<14:38,  1.03s/it, loss=0.624]
Epoch 3/4:  32%|███▏      | 401/1250 [07:24<14:38,  1.03s/it, loss=0.418]
Epoch 3/4:  32%|███▏      | 402/1250 [07:24<14:50,  1.05s/it, loss=0.418]
Epoch 3/4:  32%|███▏      | 402/1250 [07:25<14:50,  1.05s/it, loss=0.73] 
Epoch 3/4:  32%|███▏      | 403/1250 [07:25<15:06,  1.07s/it, loss=0.73]
Epoch 3/4:  32%|███▏      | 403/1250 [07:26<15:06,  1.07s/it, loss=0.62]
Epoch 3/4:  32%|███▏      | 404/1250 [07:26<15:06,  1.07s/it, loss=0.62]
Epoch 3/4:  32%|███▏      | 404/1250 [07:27<15:06,  1.07s/it, loss=0.377]
Epoch 3/4:  32%|███▏      | 405/1250 [07:27<14:52,  1.06s/it, loss=0.377]
Epoch 3/4:  32%|███▏      | 405/1250 [07:28<14:52,  1.06s/it, loss=0.638]
Epoch 3/4:  32%|███▏      | 406/1250 [07:28<14:48,  1.05s/it, loss=0.638]
Epoch 3/4:  32%|███▏      | 406/1250 [07:29<14:48,  1.05s/it, loss=0.576]
Epoch 3/4:  33%|███▎      | 407/1250 [07:29<15:04,  1.07s/it, loss=0.576]
Epoch 3/4:  33%|███▎      | 407/1250 [07:30<15:04,  1.07s/it, loss=0.684]
Epoch 3/4:  33%|███▎      | 408/1250 [07:30<15:16,  1.09s/it, loss=0.684]
Epoch 3/4:  33%|███▎      | 408/1250 [07:31<15:16,  1.09s/it, loss=0.461]
Epoch 3/4:  33%|███▎      | 409/1250 [07:31<15:04,  1.08s/it, loss=0.461]
Epoch 3/4:  33%|███▎      | 409/1250 [07:32<15:04,  1.08s/it, loss=0.554]
Epoch 3/4:  33%|███▎      | 410/1250 [07:32<14:57,  1.07s/it, loss=0.554]
Epoch 3/4:  33%|███▎      | 410/1250 [07:33<14:57,  1.07s/it, loss=0.629]
Epoch 3/4:  33%|███▎      | 411/1250 [07:33<15:18,  1.09s/it, loss=0.629]
Epoch 3/4:  33%|███▎      | 411/1250 [07:35<15:18,  1.09s/it, loss=0.581]
Epoch 3/4:  33%|███▎      | 412/1250 [07:35<15:35,  1.12s/it, loss=0.581]
Epoch 3/4:  33%|███▎      | 412/1250 [07:36<15:35,  1.12s/it, loss=0.577]
Epoch 3/4:  33%|███▎      | 413/1250 [07:36<15:21,  1.10s/it, loss=0.577]
Epoch 3/4:  33%|███▎      | 413/1250 [07:37<15:21,  1.10s/it, loss=0.407]
Epoch 3/4:  33%|███▎      | 414/1250 [07:37<15:03,  1.08s/it, loss=0.407]
Epoch 3/4:  33%|███▎      | 414/1250 [07:38<15:03,  1.08s/it, loss=0.51] 
Epoch 3/4:  33%|███▎      | 415/1250 [07:38<14:54,  1.07s/it, loss=0.51]
Epoch 3/4:  33%|███▎      | 415/1250 [07:39<14:54,  1.07s/it, loss=0.572]
Epoch 3/4:  33%|███▎      | 416/1250 [07:39<15:00,  1.08s/it, loss=0.572]
Epoch 3/4:  33%|███▎      | 416/1250 [07:40<15:00,  1.08s/it, loss=0.614]
Epoch 3/4:  33%|███▎      | 417/1250 [07:40<15:05,  1.09s/it, loss=0.614]
Epoch 3/4:  33%|███▎      | 417/1250 [07:41<15:05,  1.09s/it, loss=0.462]
Epoch 3/4:  33%|███▎      | 418/1250 [07:41<14:50,  1.07s/it, loss=0.462]
Epoch 3/4:  33%|███▎      | 418/1250 [07:42<14:50,  1.07s/it, loss=0.395]
Epoch 3/4:  34%|███▎      | 419/1250 [07:42<14:37,  1.06s/it, loss=0.395]
Epoch 3/4:  34%|███▎      | 419/1250 [07:43<14:37,  1.06s/it, loss=0.617]
Epoch 3/4:  34%|███▎      | 420/1250 [07:43<14:42,  1.06s/it, loss=0.617]
Epoch 3/4:  34%|███▎      | 420/1250 [07:44<14:42,  1.06s/it, loss=0.482]
Epoch 3/4:  34%|███▎      | 421/1250 [07:44<14:59,  1.08s/it, loss=0.482]
Epoch 3/4:  34%|███▎      | 421/1250 [07:45<14:59,  1.08s/it, loss=0.461]
Epoch 3/4:  34%|███▍      | 422/1250 [07:45<15:24,  1.12s/it, loss=0.461]
Epoch 3/4:  34%|███▍      | 422/1250 [07:46<15:24,  1.12s/it, loss=0.56] 
Epoch 3/4:  34%|███▍      | 423/1250 [07:46<15:19,  1.11s/it, loss=0.56]
Epoch 3/4:  34%|███▍      | 423/1250 [07:47<15:19,  1.11s/it, loss=0.689]
Epoch 3/4:  34%|███▍      | 424/1250 [07:47<15:01,  1.09s/it, loss=0.689]
Epoch 3/4:  34%|███▍      | 424/1250 [07:49<15:01,  1.09s/it, loss=0.631]
Epoch 3/4:  34%|███▍      | 425/1250 [07:49<15:07,  1.10s/it, loss=0.631]
Epoch 3/4:  34%|███▍      | 425/1250 [07:50<15:07,  1.10s/it, loss=0.545]
Epoch 3/4:  34%|███▍      | 426/1250 [07:50<15:11,  1.11s/it, loss=0.545]
Epoch 3/4:  34%|███▍      | 426/1250 [07:51<15:11,  1.11s/it, loss=0.769]
Epoch 3/4:  34%|███▍      | 427/1250 [07:51<15:02,  1.10s/it, loss=0.769]
Epoch 3/4:  34%|███▍      | 427/1250 [07:52<15:02,  1.10s/it, loss=0.71] 
Epoch 3/4:  34%|███▍      | 428/1250 [07:52<14:41,  1.07s/it, loss=0.71]
Epoch 3/4:  34%|███▍      | 428/1250 [07:53<14:41,  1.07s/it, loss=0.533]
Epoch 3/4:  34%|███▍      | 429/1250 [07:53<14:27,  1.06s/it, loss=0.533]
Epoch 3/4:  34%|███▍      | 429/1250 [07:54<14:27,  1.06s/it, loss=0.51] 
Epoch 3/4:  34%|███▍      | 430/1250 [07:54<14:37,  1.07s/it, loss=0.51]
Epoch 3/4:  34%|███▍      | 430/1250 [07:55<14:37,  1.07s/it, loss=0.528]
Epoch 3/4:  34%|███▍      | 431/1250 [07:55<15:03,  1.10s/it, loss=0.528]
Epoch 3/4:  34%|███▍      | 431/1250 [07:56<15:03,  1.10s/it, loss=0.61] 
Epoch 3/4:  35%|███▍      | 432/1250 [07:56<15:04,  1.11s/it, loss=0.61]
Epoch 3/4:  35%|███▍      | 432/1250 [07:57<15:04,  1.11s/it, loss=0.689]
Epoch 3/4:  35%|███▍      | 433/1250 [07:57<14:56,  1.10s/it, loss=0.689]
Epoch 3/4:  35%|███▍      | 433/1250 [07:58<14:56,  1.10s/it, loss=0.777]
Epoch 3/4:  35%|███▍      | 434/1250 [07:58<14:49,  1.09s/it, loss=0.777]
Epoch 3/4:  35%|███▍      | 434/1250 [07:59<14:49,  1.09s/it, loss=0.478]
Epoch 3/4:  35%|███▍      | 435/1250 [07:59<14:45,  1.09s/it, loss=0.478]
Epoch 3/4:  35%|███▍      | 435/1250 [08:01<14:45,  1.09s/it, loss=0.417]
Epoch 3/4:  35%|███▍      | 436/1250 [08:01<14:39,  1.08s/it, loss=0.417]
Epoch 3/4:  35%|███▍      | 436/1250 [08:02<14:39,  1.08s/it, loss=0.421]
Epoch 3/4:  35%|███▍      | 437/1250 [08:02<14:26,  1.07s/it, loss=0.421]
Epoch 3/4:  35%|███▍      | 437/1250 [08:03<14:26,  1.07s/it, loss=0.528]
Epoch 3/4:  35%|███▌      | 438/1250 [08:03<14:14,  1.05s/it, loss=0.528]
Epoch 3/4:  35%|███▌      | 438/1250 [08:04<14:14,  1.05s/it, loss=0.468]
Epoch 3/4:  35%|███▌      | 439/1250 [08:04<14:25,  1.07s/it, loss=0.468]
Epoch 3/4:  35%|███▌      | 439/1250 [08:05<14:25,  1.07s/it, loss=0.604]
Epoch 3/4:  35%|███▌      | 440/1250 [08:05<14:39,  1.09s/it, loss=0.604]
Epoch 3/4:  35%|███▌      | 440/1250 [08:06<14:39,  1.09s/it, loss=0.378]
Epoch 3/4:  35%|███▌      | 441/1250 [08:06<14:35,  1.08s/it, loss=0.378]
Epoch 3/4:  35%|███▌      | 441/1250 [08:07<14:35,  1.08s/it, loss=0.526]
Epoch 3/4:  35%|███▌      | 442/1250 [08:07<14:23,  1.07s/it, loss=0.526]
Epoch 3/4:  35%|███▌      | 442/1250 [08:08<14:23,  1.07s/it, loss=0.524]
Epoch 3/4:  35%|███▌      | 443/1250 [08:08<14:25,  1.07s/it, loss=0.524]
Epoch 3/4:  35%|███▌      | 443/1250 [08:09<14:25,  1.07s/it, loss=0.461]
Epoch 3/4:  36%|███▌      | 444/1250 [08:09<14:36,  1.09s/it, loss=0.461]
Epoch 3/4:  36%|███▌      | 444/1250 [08:10<14:36,  1.09s/it, loss=0.515]
Epoch 3/4:  36%|███▌      | 445/1250 [08:10<14:44,  1.10s/it, loss=0.515]
Epoch 3/4:  36%|███▌      | 445/1250 [08:11<14:44,  1.10s/it, loss=0.566]
Epoch 3/4:  36%|███▌      | 446/1250 [08:11<14:39,  1.09s/it, loss=0.566]
Epoch 3/4:  36%|███▌      | 446/1250 [08:12<14:39,  1.09s/it, loss=0.494]
Epoch 3/4:  36%|███▌      | 447/1250 [08:12<14:21,  1.07s/it, loss=0.494]
Epoch 3/4:  36%|███▌      | 447/1250 [08:13<14:21,  1.07s/it, loss=0.504]
Epoch 3/4:  36%|███▌      | 448/1250 [08:13<14:15,  1.07s/it, loss=0.504]
Epoch 3/4:  36%|███▌      | 448/1250 [08:15<14:15,  1.07s/it, loss=0.463]
Epoch 3/4:  36%|███▌      | 449/1250 [08:15<14:27,  1.08s/it, loss=0.463]
Epoch 3/4:  36%|███▌      | 449/1250 [08:16<14:27,  1.08s/it, loss=0.588]
Epoch 3/4:  36%|███▌      | 450/1250 [08:16<14:41,  1.10s/it, loss=0.588]
Epoch 3/4:  36%|███▌      | 450/1250 [08:17<14:41,  1.10s/it, loss=0.785]
Epoch 3/4:  36%|███▌      | 451/1250 [08:17<14:23,  1.08s/it, loss=0.785]
Epoch 3/4:  36%|███▌      | 451/1250 [08:18<14:23,  1.08s/it, loss=0.478]
Epoch 3/4:  36%|███▌      | 452/1250 [08:18<14:01,  1.05s/it, loss=0.478]
Epoch 3/4:  36%|███▌      | 452/1250 [08:19<14:01,  1.05s/it, loss=0.654]
Epoch 3/4:  36%|███▌      | 453/1250 [08:19<14:12,  1.07s/it, loss=0.654]
Epoch 3/4:  36%|███▌      | 453/1250 [08:20<14:12,  1.07s/it, loss=0.658]
Epoch 3/4:  36%|███▋      | 454/1250 [08:20<14:47,  1.12s/it, loss=0.658]
Epoch 3/4:  36%|███▋      | 454/1250 [08:21<14:47,  1.12s/it, loss=0.561]
Epoch 3/4:  36%|███▋      | 455/1250 [08:21<14:47,  1.12s/it, loss=0.561]
Epoch 3/4:  36%|███▋      | 455/1250 [08:22<14:47,  1.12s/it, loss=0.487]
Epoch 3/4:  36%|███▋      | 456/1250 [08:22<14:15,  1.08s/it, loss=0.487]
Epoch 3/4:  36%|███▋      | 456/1250 [08:23<14:15,  1.08s/it, loss=0.438]
Epoch 3/4:  37%|███▋      | 457/1250 [08:23<13:49,  1.05s/it, loss=0.438]
Epoch 3/4:  37%|███▋      | 457/1250 [08:24<13:49,  1.05s/it, loss=0.665]
Epoch 3/4:  37%|███▋      | 458/1250 [08:24<13:52,  1.05s/it, loss=0.665]
Epoch 3/4:  37%|███▋      | 458/1250 [08:25<13:52,  1.05s/it, loss=0.681]
Epoch 3/4:  37%|███▋      | 459/1250 [08:25<14:04,  1.07s/it, loss=0.681]
Epoch 3/4:  37%|███▋      | 459/1250 [08:26<14:04,  1.07s/it, loss=0.556]
Epoch 3/4:  37%|███▋      | 460/1250 [08:26<14:02,  1.07s/it, loss=0.556]
Epoch 3/4:  37%|███▋      | 460/1250 [08:27<14:02,  1.07s/it, loss=0.451]
Epoch 3/4:  37%|███▋      | 461/1250 [08:27<13:57,  1.06s/it, loss=0.451]
Epoch 3/4:  37%|███▋      | 461/1250 [08:28<13:57,  1.06s/it, loss=0.515]
Epoch 3/4:  37%|███▋      | 462/1250 [08:28<13:52,  1.06s/it, loss=0.515]
Epoch 3/4:  37%|███▋      | 462/1250 [08:30<13:52,  1.06s/it, loss=0.649]
Epoch 3/4:  37%|███▋      | 463/1250 [08:30<14:17,  1.09s/it, loss=0.649]
Epoch 3/4:  37%|███▋      | 463/1250 [08:31<14:17,  1.09s/it, loss=0.514]
Epoch 3/4:  37%|███▋      | 464/1250 [08:31<14:20,  1.09s/it, loss=0.514]
Epoch 3/4:  37%|███▋      | 464/1250 [08:32<14:20,  1.09s/it, loss=0.566]
Epoch 3/4:  37%|███▋      | 465/1250 [08:32<13:59,  1.07s/it, loss=0.566]
Epoch 3/4:  37%|███▋      | 465/1250 [08:33<13:59,  1.07s/it, loss=0.466]
Epoch 3/4:  37%|███▋      | 466/1250 [08:33<13:36,  1.04s/it, loss=0.466]
Epoch 3/4:  37%|███▋      | 466/1250 [08:34<13:36,  1.04s/it, loss=0.542]
Epoch 3/4:  37%|███▋      | 467/1250 [08:34<13:48,  1.06s/it, loss=0.542]
Epoch 3/4:  37%|███▋      | 467/1250 [08:35<13:48,  1.06s/it, loss=0.53] 
Epoch 3/4:  37%|███▋      | 468/1250 [08:35<14:25,  1.11s/it, loss=0.53]
Epoch 3/4:  37%|███▋      | 468/1250 [08:36<14:25,  1.11s/it, loss=0.825]
Epoch 3/4:  38%|███▊      | 469/1250 [08:36<14:29,  1.11s/it, loss=0.825]
Epoch 3/4:  38%|███▊      | 469/1250 [08:37<14:29,  1.11s/it, loss=0.385]
Epoch 3/4:  38%|███▊      | 470/1250 [08:37<14:06,  1.09s/it, loss=0.385]
Epoch 3/4:  38%|███▊      | 470/1250 [08:38<14:06,  1.09s/it, loss=0.645]
Epoch 3/4:  38%|███▊      | 471/1250 [08:38<13:48,  1.06s/it, loss=0.645]
Epoch 3/4:  38%|███▊      | 471/1250 [08:39<13:48,  1.06s/it, loss=0.601]
Epoch 3/4:  38%|███▊      | 472/1250 [08:39<13:54,  1.07s/it, loss=0.601]
Epoch 3/4:  38%|███▊      | 472/1250 [08:40<13:54,  1.07s/it, loss=0.555]
Epoch 3/4:  38%|███▊      | 473/1250 [08:40<14:03,  1.09s/it, loss=0.555]
Epoch 3/4:  38%|███▊      | 473/1250 [08:41<14:03,  1.09s/it, loss=0.432]
Epoch 3/4:  38%|███▊      | 474/1250 [08:41<13:58,  1.08s/it, loss=0.432]
Epoch 3/4:  38%|███▊      | 474/1250 [08:42<13:58,  1.08s/it, loss=0.568]
Epoch 3/4:  38%|███▊      | 475/1250 [08:42<13:49,  1.07s/it, loss=0.568]
Epoch 3/4:  38%|███▊      | 475/1250 [08:44<13:49,  1.07s/it, loss=0.718]
Epoch 3/4:  38%|███▊      | 476/1250 [08:44<13:51,  1.07s/it, loss=0.718]
Epoch 3/4:  38%|███▊      | 476/1250 [08:45<13:51,  1.07s/it, loss=0.522]
Epoch 3/4:  38%|███▊      | 477/1250 [08:45<14:13,  1.10s/it, loss=0.522]
Epoch 3/4:  38%|███▊      | 477/1250 [08:46<14:13,  1.10s/it, loss=0.728]
Epoch 3/4:  38%|███▊      | 478/1250 [08:46<14:29,  1.13s/it, loss=0.728]
Epoch 3/4:  38%|███▊      | 478/1250 [08:47<14:29,  1.13s/it, loss=0.646]
Epoch 3/4:  38%|███▊      | 479/1250 [08:47<14:05,  1.10s/it, loss=0.646]
Epoch 3/4:  38%|███▊      | 479/1250 [08:48<14:05,  1.10s/it, loss=0.567]
Epoch 3/4:  38%|███▊      | 480/1250 [08:48<13:40,  1.07s/it, loss=0.567]
Epoch 3/4:  38%|███▊      | 480/1250 [08:49<13:40,  1.07s/it, loss=0.554]
Epoch 3/4:  38%|███▊      | 481/1250 [08:49<13:41,  1.07s/it, loss=0.554]
Epoch 3/4:  38%|███▊      | 481/1250 [08:50<13:41,  1.07s/it, loss=0.549]
Epoch 3/4:  39%|███▊      | 482/1250 [08:50<13:55,  1.09s/it, loss=0.549]
Epoch 3/4:  39%|███▊      | 482/1250 [08:51<13:55,  1.09s/it, loss=0.652]
Epoch 3/4:  39%|███▊      | 483/1250 [08:51<13:48,  1.08s/it, loss=0.652]
Epoch 3/4:  39%|███▊      | 483/1250 [08:52<13:48,  1.08s/it, loss=0.581]
Epoch 3/4:  39%|███▊      | 484/1250 [08:52<13:33,  1.06s/it, loss=0.581]
Epoch 3/4:  39%|███▊      | 484/1250 [08:53<13:33,  1.06s/it, loss=0.44] 
Epoch 3/4:  39%|███▉      | 485/1250 [08:53<13:39,  1.07s/it, loss=0.44]
Epoch 3/4:  39%|███▉      | 485/1250 [08:54<13:39,  1.07s/it, loss=0.447]
Epoch 3/4:  39%|███▉      | 486/1250 [08:54<14:02,  1.10s/it, loss=0.447]
Epoch 3/4:  39%|███▉      | 486/1250 [08:56<14:02,  1.10s/it, loss=0.657]
Epoch 3/4:  39%|███▉      | 487/1250 [08:56<14:08,  1.11s/it, loss=0.657]
Epoch 3/4:  39%|███▉      | 487/1250 [08:57<14:08,  1.11s/it, loss=0.516]
Epoch 3/4:  39%|███▉      | 488/1250 [08:57<13:48,  1.09s/it, loss=0.516]
Epoch 3/4:  39%|███▉      | 488/1250 [08:58<13:48,  1.09s/it, loss=0.708]
Epoch 3/4:  39%|███▉      | 489/1250 [08:58<13:25,  1.06s/it, loss=0.708]
Epoch 3/4:  39%|███▉      | 489/1250 [08:59<13:25,  1.06s/it, loss=0.792]
Epoch 3/4:  39%|███▉      | 490/1250 [08:59<13:19,  1.05s/it, loss=0.792]
Epoch 3/4:  39%|███▉      | 490/1250 [09:00<13:19,  1.05s/it, loss=0.465]
Epoch 3/4:  39%|███▉      | 491/1250 [09:00<13:33,  1.07s/it, loss=0.465]
Epoch 3/4:  39%|███▉      | 491/1250 [09:01<13:33,  1.07s/it, loss=0.586]
Epoch 3/4:  39%|███▉      | 492/1250 [09:01<13:40,  1.08s/it, loss=0.586]
Epoch 3/4:  39%|███▉      | 492/1250 [09:02<13:40,  1.08s/it, loss=0.617]
Epoch 3/4:  39%|███▉      | 493/1250 [09:02<13:31,  1.07s/it, loss=0.617]
Epoch 3/4:  39%|███▉      | 493/1250 [09:03<13:31,  1.07s/it, loss=0.569]
Epoch 3/4:  40%|███▉      | 494/1250 [09:03<13:23,  1.06s/it, loss=0.569]
Epoch 3/4:  40%|███▉      | 494/1250 [09:04<13:23,  1.06s/it, loss=0.32] 
Epoch 3/4:  40%|███▉      | 495/1250 [09:04<13:38,  1.08s/it, loss=0.32]
Epoch 3/4:  40%|███▉      | 495/1250 [09:05<13:38,  1.08s/it, loss=0.315]
Epoch 3/4:  40%|███▉      | 496/1250 [09:05<14:03,  1.12s/it, loss=0.315]
Epoch 3/4:  40%|███▉      | 496/1250 [09:06<14:03,  1.12s/it, loss=0.424]
Epoch 3/4:  40%|███▉      | 497/1250 [09:06<13:59,  1.12s/it, loss=0.424]
Epoch 3/4:  40%|███▉      | 497/1250 [09:07<13:59,  1.12s/it, loss=0.467]
Epoch 3/4:  40%|███▉      | 498/1250 [09:07<13:33,  1.08s/it, loss=0.467]
Epoch 3/4:  40%|███▉      | 498/1250 [09:08<13:33,  1.08s/it, loss=0.436]
Epoch 3/4:  40%|███▉      | 499/1250 [09:08<13:10,  1.05s/it, loss=0.436]
Epoch 3/4:  40%|███▉      | 499/1250 [09:10<13:10,  1.05s/it, loss=0.659]
Epoch 3/4:  40%|████      | 500/1250 [09:10<13:14,  1.06s/it, loss=0.659]
Epoch 3/4:  40%|████      | 500/1250 [09:11<13:14,  1.06s/it, loss=0.477]
Epoch 3/4:  40%|████      | 501/1250 [09:11<13:44,  1.10s/it, loss=0.477]
Epoch 3/4:  40%|████      | 501/1250 [09:12<13:44,  1.10s/it, loss=0.575]
Epoch 3/4:  40%|████      | 502/1250 [09:12<13:47,  1.11s/it, loss=0.575]
Epoch 3/4:  40%|████      | 502/1250 [09:13<13:47,  1.11s/it, loss=0.373]
Epoch 3/4:  40%|████      | 503/1250 [09:13<13:31,  1.09s/it, loss=0.373]
Epoch 3/4:  40%|████      | 503/1250 [09:14<13:31,  1.09s/it, loss=0.441]
Epoch 3/4:  40%|████      | 504/1250 [09:14<13:20,  1.07s/it, loss=0.441]
Epoch 3/4:  40%|████      | 504/1250 [09:15<13:20,  1.07s/it, loss=0.599]
Epoch 3/4:  40%|████      | 505/1250 [09:15<13:31,  1.09s/it, loss=0.599]
Epoch 3/4:  40%|████      | 505/1250 [09:16<13:31,  1.09s/it, loss=0.451]
Epoch 3/4:  40%|████      | 506/1250 [09:16<13:44,  1.11s/it, loss=0.451]
Epoch 3/4:  40%|████      | 506/1250 [09:17<13:44,  1.11s/it, loss=0.548]
Epoch 3/4:  41%|████      | 507/1250 [09:17<13:22,  1.08s/it, loss=0.548]
Epoch 3/4:  41%|████      | 507/1250 [09:18<13:22,  1.08s/it, loss=0.489]
Epoch 3/4:  41%|████      | 508/1250 [09:18<13:00,  1.05s/it, loss=0.489]
Epoch 3/4:  41%|████      | 508/1250 [09:19<13:00,  1.05s/it, loss=0.656]
Epoch 3/4:  41%|████      | 509/1250 [09:19<13:05,  1.06s/it, loss=0.656]
Epoch 3/4:  41%|████      | 509/1250 [09:20<13:05,  1.06s/it, loss=0.559]
Epoch 3/4:  41%|████      | 510/1250 [09:20<13:21,  1.08s/it, loss=0.559]
Epoch 3/4:  41%|████      | 510/1250 [09:21<13:21,  1.08s/it, loss=0.743]
Epoch 3/4:  41%|████      | 511/1250 [09:21<13:23,  1.09s/it, loss=0.743]
Epoch 3/4:  41%|████      | 511/1250 [09:22<13:23,  1.09s/it, loss=0.592]
Epoch 3/4:  41%|████      | 512/1250 [09:22<13:01,  1.06s/it, loss=0.592]
Epoch 3/4:  41%|████      | 512/1250 [09:24<13:01,  1.06s/it, loss=0.54] 
Epoch 3/4:  41%|████      | 513/1250 [09:24<12:55,  1.05s/it, loss=0.54]
Epoch 3/4:  41%|████      | 513/1250 [09:25<12:55,  1.05s/it, loss=0.579]
Epoch 3/4:  41%|████      | 514/1250 [09:25<13:09,  1.07s/it, loss=0.579]
Epoch 3/4:  41%|████      | 514/1250 [09:26<13:09,  1.07s/it, loss=0.552]
Epoch 3/4:  41%|████      | 515/1250 [09:26<13:33,  1.11s/it, loss=0.552]
Epoch 3/4:  41%|████      | 515/1250 [09:27<13:33,  1.11s/it, loss=0.592]
Epoch 3/4:  41%|████▏     | 516/1250 [09:27<13:23,  1.09s/it, loss=0.592]
Epoch 3/4:  41%|████▏     | 516/1250 [09:28<13:23,  1.09s/it, loss=0.652]
Epoch 3/4:  41%|████▏     | 517/1250 [09:28<13:03,  1.07s/it, loss=0.652]
Epoch 3/4:  41%|████▏     | 517/1250 [09:29<13:03,  1.07s/it, loss=0.704]
Epoch 3/4:  41%|████▏     | 518/1250 [09:29<13:04,  1.07s/it, loss=0.704]
Epoch 3/4:  41%|████▏     | 518/1250 [09:30<13:04,  1.07s/it, loss=0.629]
Epoch 3/4:  42%|████▏     | 519/1250 [09:30<13:22,  1.10s/it, loss=0.629]
Epoch 3/4:  42%|████▏     | 519/1250 [09:31<13:22,  1.10s/it, loss=0.787]
Epoch 3/4:  42%|████▏     | 520/1250 [09:31<13:17,  1.09s/it, loss=0.787]
Epoch 3/4:  42%|████▏     | 520/1250 [09:32<13:17,  1.09s/it, loss=0.752]
Epoch 3/4:  42%|████▏     | 521/1250 [09:32<12:57,  1.07s/it, loss=0.752]
Epoch 3/4:  42%|████▏     | 521/1250 [09:33<12:57,  1.07s/it, loss=0.441]
Epoch 3/4:  42%|████▏     | 522/1250 [09:33<12:42,  1.05s/it, loss=0.441]
Epoch 3/4:  42%|████▏     | 522/1250 [09:34<12:42,  1.05s/it, loss=1.22] 
Epoch 3/4:  42%|████▏     | 523/1250 [09:34<12:46,  1.05s/it, loss=1.22]
Epoch 3/4:  42%|████▏     | 523/1250 [09:35<12:46,  1.05s/it, loss=0.372]
Epoch 3/4:  42%|████▏     | 524/1250 [09:35<13:10,  1.09s/it, loss=0.372]
Epoch 3/4:  42%|████▏     | 524/1250 [09:37<13:10,  1.09s/it, loss=0.65] 
Epoch 3/4:  42%|████▏     | 525/1250 [09:37<13:14,  1.10s/it, loss=0.65]
Epoch 3/4:  42%|████▏     | 525/1250 [09:38<13:14,  1.10s/it, loss=0.448]
Epoch 3/4:  42%|████▏     | 526/1250 [09:38<12:54,  1.07s/it, loss=0.448]
Epoch 3/4:  42%|████▏     | 526/1250 [09:39<12:54,  1.07s/it, loss=0.674]
Epoch 3/4:  42%|████▏     | 527/1250 [09:39<12:39,  1.05s/it, loss=0.674]
Epoch 3/4:  42%|████▏     | 527/1250 [09:40<12:39,  1.05s/it, loss=0.669]
Epoch 3/4:  42%|████▏     | 528/1250 [09:40<12:45,  1.06s/it, loss=0.669]
Epoch 3/4:  42%|████▏     | 528/1250 [09:41<12:45,  1.06s/it, loss=0.528]
Epoch 3/4:  42%|████▏     | 529/1250 [09:41<13:06,  1.09s/it, loss=0.528]
Epoch 3/4:  42%|████▏     | 529/1250 [09:42<13:06,  1.09s/it, loss=0.62] 
Epoch 3/4:  42%|████▏     | 530/1250 [09:42<13:03,  1.09s/it, loss=0.62]
Epoch 3/4:  42%|████▏     | 530/1250 [09:43<13:03,  1.09s/it, loss=0.509]
Epoch 3/4:  42%|████▏     | 531/1250 [09:43<12:41,  1.06s/it, loss=0.509]
Epoch 3/4:  42%|████▏     | 531/1250 [09:44<12:41,  1.06s/it, loss=0.554]
Epoch 3/4:  43%|████▎     | 532/1250 [09:44<12:27,  1.04s/it, loss=0.554]
Epoch 3/4:  43%|████▎     | 532/1250 [09:45<12:27,  1.04s/it, loss=0.497]
Epoch 3/4:  43%|████▎     | 533/1250 [09:45<12:54,  1.08s/it, loss=0.497]
Epoch 3/4:  43%|████▎     | 533/1250 [09:46<12:54,  1.08s/it, loss=0.543]
Epoch 3/4:  43%|████▎     | 534/1250 [09:46<13:20,  1.12s/it, loss=0.543]
Epoch 3/4:  43%|████▎     | 534/1250 [09:47<13:20,  1.12s/it, loss=0.492]
Epoch 3/4:  43%|████▎     | 535/1250 [09:47<13:10,  1.11s/it, loss=0.492]
Epoch 3/4:  43%|████▎     | 535/1250 [09:48<13:10,  1.11s/it, loss=0.614]
Epoch 3/4:  43%|████▎     | 536/1250 [09:48<12:48,  1.08s/it, loss=0.614]
Epoch 3/4:  43%|████▎     | 536/1250 [09:49<12:48,  1.08s/it, loss=0.586]
Epoch 3/4:  43%|████▎     | 537/1250 [09:49<12:43,  1.07s/it, loss=0.586]
Epoch 3/4:  43%|████▎     | 537/1250 [09:51<12:43,  1.07s/it, loss=0.547]
Epoch 3/4:  43%|████▎     | 538/1250 [09:51<12:57,  1.09s/it, loss=0.547]
Epoch 3/4:  43%|████▎     | 538/1250 [09:52<12:57,  1.09s/it, loss=0.546]
Epoch 3/4:  43%|████▎     | 539/1250 [09:52<13:14,  1.12s/it, loss=0.546]
Epoch 3/4:  43%|████▎     | 539/1250 [09:53<13:14,  1.12s/it, loss=0.498]
Epoch 3/4:  43%|████▎     | 540/1250 [09:53<13:07,  1.11s/it, loss=0.498]
Epoch 3/4:  43%|████▎     | 540/1250 [09:54<13:07,  1.11s/it, loss=0.57] 
Epoch 3/4:  43%|████▎     | 541/1250 [09:54<12:50,  1.09s/it, loss=0.57]
Epoch 3/4:  43%|████▎     | 541/1250 [09:55<12:50,  1.09s/it, loss=0.418]
Epoch 3/4:  43%|████▎     | 542/1250 [09:55<12:48,  1.09s/it, loss=0.418]
Epoch 3/4:  43%|████▎     | 542/1250 [09:56<12:48,  1.09s/it, loss=0.76] 
Epoch 3/4:  43%|████▎     | 543/1250 [09:56<12:58,  1.10s/it, loss=0.76]
Epoch 3/4:  43%|████▎     | 543/1250 [09:57<12:58,  1.10s/it, loss=0.367]
Epoch 3/4:  44%|████▎     | 544/1250 [09:57<12:54,  1.10s/it, loss=0.367]
Epoch 3/4:  44%|████▎     | 544/1250 [09:58<12:54,  1.10s/it, loss=0.721]
Epoch 3/4:  44%|████▎     | 545/1250 [09:58<12:34,  1.07s/it, loss=0.721]
Epoch 3/4:  44%|████▎     | 545/1250 [09:59<12:34,  1.07s/it, loss=0.433]
Epoch 3/4:  44%|████▎     | 546/1250 [09:59<12:22,  1.05s/it, loss=0.433]
Epoch 3/4:  44%|████▎     | 546/1250 [10:00<12:22,  1.05s/it, loss=0.685]
Epoch 3/4:  44%|████▍     | 547/1250 [10:00<12:28,  1.06s/it, loss=0.685]
Epoch 3/4:  44%|████▍     | 547/1250 [10:01<12:28,  1.06s/it, loss=0.429]
Epoch 3/4:  44%|████▍     | 548/1250 [10:01<12:45,  1.09s/it, loss=0.429]
Epoch 3/4:  44%|████▍     | 548/1250 [10:03<12:45,  1.09s/it, loss=0.858]
Epoch 3/4:  44%|████▍     | 549/1250 [10:03<12:46,  1.09s/it, loss=0.858]
Epoch 3/4:  44%|████▍     | 549/1250 [10:04<12:46,  1.09s/it, loss=0.887]
Epoch 3/4:  44%|████▍     | 550/1250 [10:04<12:29,  1.07s/it, loss=0.887]
Epoch 3/4:  44%|████▍     | 550/1250 [10:05<12:29,  1.07s/it, loss=0.53] 
Epoch 3/4:  44%|████▍     | 551/1250 [10:05<12:21,  1.06s/it, loss=0.53]
Epoch 3/4:  44%|████▍     | 551/1250 [10:06<12:21,  1.06s/it, loss=0.776]
Epoch 3/4:  44%|████▍     | 552/1250 [10:06<12:34,  1.08s/it, loss=0.776]
Epoch 3/4:  44%|████▍     | 552/1250 [10:07<12:34,  1.08s/it, loss=0.565]
Epoch 3/4:  44%|████▍     | 553/1250 [10:07<12:43,  1.10s/it, loss=0.565]
Epoch 3/4:  44%|████▍     | 553/1250 [10:08<12:43,  1.10s/it, loss=0.497]
Epoch 3/4:  44%|████▍     | 554/1250 [10:08<12:28,  1.08s/it, loss=0.497]
Epoch 3/4:  44%|████▍     | 554/1250 [10:09<12:28,  1.08s/it, loss=0.332]
Epoch 3/4:  44%|████▍     | 555/1250 [10:09<12:13,  1.05s/it, loss=0.332]
Epoch 3/4:  44%|████▍     | 555/1250 [10:10<12:13,  1.05s/it, loss=0.733]
Epoch 3/4:  44%|████▍     | 556/1250 [10:10<12:20,  1.07s/it, loss=0.733]
Epoch 3/4:  44%|████▍     | 556/1250 [10:11<12:20,  1.07s/it, loss=0.703]
Epoch 3/4:  45%|████▍     | 557/1250 [10:11<12:58,  1.12s/it, loss=0.703]
Epoch 3/4:  45%|████▍     | 557/1250 [10:12<12:58,  1.12s/it, loss=0.579]
Epoch 3/4:  45%|████▍     | 558/1250 [10:12<12:57,  1.12s/it, loss=0.579]
Epoch 3/4:  45%|████▍     | 558/1250 [10:13<12:57,  1.12s/it, loss=0.365]
Epoch 3/4:  45%|████▍     | 559/1250 [10:13<12:30,  1.09s/it, loss=0.365]
Epoch 3/4:  45%|████▍     | 559/1250 [10:14<12:30,  1.09s/it, loss=0.538]
Epoch 3/4:  45%|████▍     | 560/1250 [10:14<12:12,  1.06s/it, loss=0.538]
Epoch 3/4:  45%|████▍     | 560/1250 [10:16<12:12,  1.06s/it, loss=0.609]
Epoch 3/4:  45%|████▍     | 561/1250 [10:16<12:34,  1.09s/it, loss=0.609]
Epoch 3/4:  45%|████▍     | 561/1250 [10:17<12:34,  1.09s/it, loss=0.437]
Epoch 3/4:  45%|████▍     | 562/1250 [10:17<12:55,  1.13s/it, loss=0.437]
Epoch 3/4:  45%|████▍     | 562/1250 [10:18<12:55,  1.13s/it, loss=0.577]
Epoch 3/4:  45%|████▌     | 563/1250 [10:18<12:43,  1.11s/it, loss=0.577]
Epoch 3/4:  45%|████▌     | 563/1250 [10:19<12:43,  1.11s/it, loss=0.452]
Epoch 3/4:  45%|████▌     | 564/1250 [10:19<12:21,  1.08s/it, loss=0.452]
Epoch 3/4:  45%|████▌     | 564/1250 [10:20<12:21,  1.08s/it, loss=0.671]
Epoch 3/4:  45%|████▌     | 565/1250 [10:20<12:18,  1.08s/it, loss=0.671]
Epoch 3/4:  45%|████▌     | 565/1250 [10:21<12:18,  1.08s/it, loss=0.603]
Epoch 3/4:  45%|████▌     | 566/1250 [10:21<12:30,  1.10s/it, loss=0.603]
Epoch 3/4:  45%|████▌     | 566/1250 [10:22<12:30,  1.10s/it, loss=0.681]
Epoch 3/4:  45%|████▌     | 567/1250 [10:22<12:34,  1.10s/it, loss=0.681]
Epoch 3/4:  45%|████▌     | 567/1250 [10:23<12:34,  1.10s/it, loss=0.574]
Epoch 3/4:  45%|████▌     | 568/1250 [10:23<12:17,  1.08s/it, loss=0.574]
Epoch 3/4:  45%|████▌     | 568/1250 [10:24<12:17,  1.08s/it, loss=0.559]
Epoch 3/4:  46%|████▌     | 569/1250 [10:24<11:58,  1.06s/it, loss=0.559]
Epoch 3/4:  46%|████▌     | 569/1250 [10:25<11:58,  1.06s/it, loss=0.506]
Epoch 3/4:  46%|████▌     | 570/1250 [10:25<12:01,  1.06s/it, loss=0.506]
Epoch 3/4:  46%|████▌     | 570/1250 [10:26<12:01,  1.06s/it, loss=0.741]
Epoch 3/4:  46%|████▌     | 571/1250 [10:26<12:19,  1.09s/it, loss=0.741]
Epoch 3/4:  46%|████▌     | 571/1250 [10:28<12:19,  1.09s/it, loss=0.527]
Epoch 3/4:  46%|████▌     | 572/1250 [10:28<12:21,  1.09s/it, loss=0.527]
Epoch 3/4:  46%|████▌     | 572/1250 [10:29<12:21,  1.09s/it, loss=0.716]
Epoch 3/4:  46%|████▌     | 573/1250 [10:29<12:07,  1.07s/it, loss=0.716]
Epoch 3/4:  46%|████▌     | 573/1250 [10:30<12:07,  1.07s/it, loss=0.576]
Epoch 3/4:  46%|████▌     | 574/1250 [10:30<11:56,  1.06s/it, loss=0.576]
Epoch 3/4:  46%|████▌     | 574/1250 [10:31<11:56,  1.06s/it, loss=0.595]
Epoch 3/4:  46%|████▌     | 575/1250 [10:31<12:05,  1.07s/it, loss=0.595]
Epoch 3/4:  46%|████▌     | 575/1250 [10:32<12:05,  1.07s/it, loss=0.358]
Epoch 3/4:  46%|████▌     | 576/1250 [10:32<12:24,  1.10s/it, loss=0.358]
Epoch 3/4:  46%|████▌     | 576/1250 [10:33<12:24,  1.10s/it, loss=0.336]
Epoch 3/4:  46%|████▌     | 577/1250 [10:33<12:16,  1.09s/it, loss=0.336]
Epoch 3/4:  46%|████▌     | 577/1250 [10:34<12:16,  1.09s/it, loss=0.582]
Epoch 3/4:  46%|████▌     | 578/1250 [10:34<11:58,  1.07s/it, loss=0.582]
Epoch 3/4:  46%|████▌     | 578/1250 [10:35<11:58,  1.07s/it, loss=0.625]
Epoch 3/4:  46%|████▋     | 579/1250 [10:35<11:57,  1.07s/it, loss=0.625]
Epoch 3/4:  46%|████▋     | 579/1250 [10:36<11:57,  1.07s/it, loss=0.608]
Epoch 3/4:  46%|████▋     | 580/1250 [10:36<12:09,  1.09s/it, loss=0.608]
Epoch 3/4:  46%|████▋     | 580/1250 [10:37<12:09,  1.09s/it, loss=0.857]
Epoch 3/4:  46%|████▋     | 581/1250 [10:37<12:19,  1.11s/it, loss=0.857]
Epoch 3/4:  46%|████▋     | 581/1250 [10:38<12:19,  1.11s/it, loss=0.501]
Epoch 3/4:  47%|████▋     | 582/1250 [10:38<12:08,  1.09s/it, loss=0.501]
Epoch 3/4:  47%|████▋     | 582/1250 [10:39<12:08,  1.09s/it, loss=0.552]
Epoch 3/4:  47%|████▋     | 583/1250 [10:39<11:53,  1.07s/it, loss=0.552]
Epoch 3/4:  47%|████▋     | 583/1250 [10:41<11:53,  1.07s/it, loss=0.471]
Epoch 3/4:  47%|████▋     | 584/1250 [10:41<12:09,  1.10s/it, loss=0.471]
Epoch 3/4:  47%|████▋     | 584/1250 [10:42<12:09,  1.10s/it, loss=0.499]
Epoch 3/4:  47%|████▋     | 585/1250 [10:42<12:34,  1.14s/it, loss=0.499]
Epoch 3/4:  47%|████▋     | 585/1250 [10:43<12:34,  1.14s/it, loss=0.473]
Epoch 3/4:  47%|████▋     | 586/1250 [10:43<12:25,  1.12s/it, loss=0.473]
Epoch 3/4:  47%|████▋     | 586/1250 [10:44<12:25,  1.12s/it, loss=0.563]
Epoch 3/4:  47%|████▋     | 587/1250 [10:44<11:57,  1.08s/it, loss=0.563]
Epoch 3/4:  47%|████▋     | 587/1250 [10:45<11:57,  1.08s/it, loss=0.628]
Epoch 3/4:  47%|████▋     | 588/1250 [10:45<11:47,  1.07s/it, loss=0.628]
Epoch 3/4:  47%|████▋     | 588/1250 [10:46<11:47,  1.07s/it, loss=0.855]
Epoch 3/4:  47%|████▋     | 589/1250 [10:46<12:02,  1.09s/it, loss=0.855]
Epoch 3/4:  47%|████▋     | 589/1250 [10:47<12:02,  1.09s/it, loss=0.623]
Epoch 3/4:  47%|████▋     | 590/1250 [10:47<12:11,  1.11s/it, loss=0.623]
Epoch 3/4:  47%|████▋     | 590/1250 [10:48<12:11,  1.11s/it, loss=0.69] 
Epoch 3/4:  47%|████▋     | 591/1250 [10:48<11:55,  1.09s/it, loss=0.69]
Epoch 3/4:  47%|████▋     | 591/1250 [10:49<11:55,  1.09s/it, loss=0.528]
Epoch 3/4:  47%|████▋     | 592/1250 [10:49<11:35,  1.06s/it, loss=0.528]
Epoch 3/4:  47%|████▋     | 592/1250 [10:50<11:35,  1.06s/it, loss=0.541]
Epoch 3/4:  47%|████▋     | 593/1250 [10:50<11:32,  1.05s/it, loss=0.541]
Epoch 3/4:  47%|████▋     | 593/1250 [10:51<11:32,  1.05s/it, loss=0.649]
Epoch 3/4:  48%|████▊     | 594/1250 [10:51<11:49,  1.08s/it, loss=0.649]
Epoch 3/4:  48%|████▊     | 594/1250 [10:53<11:49,  1.08s/it, loss=0.55] 
Epoch 3/4:  48%|████▊     | 595/1250 [10:53<11:57,  1.10s/it, loss=0.55]
Epoch 3/4:  48%|████▊     | 595/1250 [10:54<11:57,  1.10s/it, loss=0.812]
Epoch 3/4:  48%|████▊     | 596/1250 [10:54<11:40,  1.07s/it, loss=0.812]
Epoch 3/4:  48%|████▊     | 596/1250 [10:55<11:40,  1.07s/it, loss=0.366]
Epoch 3/4:  48%|████▊     | 597/1250 [10:55<11:22,  1.04s/it, loss=0.366]
Epoch 3/4:  48%|████▊     | 597/1250 [10:56<11:22,  1.04s/it, loss=0.78] 
Epoch 3/4:  48%|████▊     | 598/1250 [10:56<11:29,  1.06s/it, loss=0.78]
Epoch 3/4:  48%|████▊     | 598/1250 [10:57<11:29,  1.06s/it, loss=0.517]
Epoch 3/4:  48%|████▊     | 599/1250 [10:57<11:48,  1.09s/it, loss=0.517]
Epoch 3/4:  48%|████▊     | 599/1250 [10:58<11:48,  1.09s/it, loss=0.402]
Epoch 3/4:  48%|████▊     | 600/1250 [10:58<11:41,  1.08s/it, loss=0.402]
Epoch 3/4:  48%|████▊     | 600/1250 [10:59<11:41,  1.08s/it, loss=0.692]
Epoch 3/4:  48%|████▊     | 601/1250 [10:59<11:25,  1.06s/it, loss=0.692]
Epoch 3/4:  48%|████▊     | 601/1250 [11:00<11:25,  1.06s/it, loss=0.898]
Epoch 3/4:  48%|████▊     | 602/1250 [11:00<11:21,  1.05s/it, loss=0.898]
Epoch 3/4:  48%|████▊     | 602/1250 [11:01<11:21,  1.05s/it, loss=0.453]
Epoch 3/4:  48%|████▊     | 603/1250 [11:01<11:44,  1.09s/it, loss=0.453]
Epoch 3/4:  48%|████▊     | 603/1250 [11:02<11:44,  1.09s/it, loss=0.558]
Epoch 3/4:  48%|████▊     | 604/1250 [11:02<11:58,  1.11s/it, loss=0.558]
Epoch 3/4:  48%|████▊     | 604/1250 [11:03<11:58,  1.11s/it, loss=0.524]
Epoch 3/4:  48%|████▊     | 605/1250 [11:03<11:40,  1.09s/it, loss=0.524]
Epoch 3/4:  48%|████▊     | 605/1250 [11:04<11:40,  1.09s/it, loss=0.77] 
Epoch 3/4:  48%|████▊     | 606/1250 [11:04<11:20,  1.06s/it, loss=0.77]
Epoch 3/4:  48%|████▊     | 606/1250 [11:05<11:20,  1.06s/it, loss=0.773]
Epoch 3/4:  49%|████▊     | 607/1250 [11:05<11:20,  1.06s/it, loss=0.773]
Epoch 3/4:  49%|████▊     | 607/1250 [11:07<11:20,  1.06s/it, loss=0.517]
Epoch 3/4:  49%|████▊     | 608/1250 [11:07<11:52,  1.11s/it, loss=0.517]
Epoch 3/4:  49%|████▊     | 608/1250 [11:08<11:52,  1.11s/it, loss=0.525]
Epoch 3/4:  49%|████▊     | 609/1250 [11:08<11:52,  1.11s/it, loss=0.525]
Epoch 3/4:  49%|████▊     | 609/1250 [11:09<11:52,  1.11s/it, loss=0.516]
Epoch 3/4:  49%|████▉     | 610/1250 [11:09<11:27,  1.07s/it, loss=0.516]
Epoch 3/4:  49%|████▉     | 610/1250 [11:10<11:27,  1.07s/it, loss=0.416]
Epoch 3/4:  49%|████▉     | 611/1250 [11:10<11:05,  1.04s/it, loss=0.416]
Epoch 3/4:  49%|████▉     | 611/1250 [11:11<11:05,  1.04s/it, loss=0.566]
Epoch 3/4:  49%|████▉     | 612/1250 [11:11<11:09,  1.05s/it, loss=0.566]
Epoch 3/4:  49%|████▉     | 612/1250 [11:12<11:09,  1.05s/it, loss=0.821]
Epoch 3/4:  49%|████▉     | 613/1250 [11:12<11:36,  1.09s/it, loss=0.821]
Epoch 3/4:  49%|████▉     | 613/1250 [11:13<11:36,  1.09s/it, loss=0.473]
Epoch 3/4:  49%|████▉     | 614/1250 [11:13<11:38,  1.10s/it, loss=0.473]
Epoch 3/4:  49%|████▉     | 614/1250 [11:14<11:38,  1.10s/it, loss=0.548]
Epoch 3/4:  49%|████▉     | 615/1250 [11:14<11:20,  1.07s/it, loss=0.548]
Epoch 3/4:  49%|████▉     | 615/1250 [11:15<11:20,  1.07s/it, loss=0.629]
Epoch 3/4:  49%|████▉     | 616/1250 [11:15<11:21,  1.07s/it, loss=0.629]
Epoch 3/4:  49%|████▉     | 616/1250 [11:16<11:21,  1.07s/it, loss=0.543]
Epoch 3/4:  49%|████▉     | 617/1250 [11:16<11:32,  1.09s/it, loss=0.543]
Epoch 3/4:  49%|████▉     | 617/1250 [11:17<11:32,  1.09s/it, loss=0.552]
Epoch 3/4:  49%|████▉     | 618/1250 [11:17<11:34,  1.10s/it, loss=0.552]
Epoch 3/4:  49%|████▉     | 618/1250 [11:18<11:34,  1.10s/it, loss=0.595]
Epoch 3/4:  50%|████▉     | 619/1250 [11:18<11:17,  1.07s/it, loss=0.595]
Epoch 3/4:  50%|████▉     | 619/1250 [11:19<11:17,  1.07s/it, loss=0.401]
Epoch 3/4:  50%|████▉     | 620/1250 [11:19<10:57,  1.04s/it, loss=0.401]
Epoch 3/4:  50%|████▉     | 620/1250 [11:20<10:57,  1.04s/it, loss=0.673]
Epoch 3/4:  50%|████▉     | 621/1250 [11:20<10:55,  1.04s/it, loss=0.673]
Epoch 3/4:  50%|████▉     | 621/1250 [11:21<10:55,  1.04s/it, loss=0.557]
Epoch 3/4:  50%|████▉     | 622/1250 [11:21<11:10,  1.07s/it, loss=0.557]
Epoch 3/4:  50%|████▉     | 622/1250 [11:23<11:10,  1.07s/it, loss=0.54] 
Epoch 3/4:  50%|████▉     | 623/1250 [11:23<11:17,  1.08s/it, loss=0.54]
Epoch 3/4:  50%|████▉     | 623/1250 [11:24<11:17,  1.08s/it, loss=0.534]
Epoch 3/4:  50%|████▉     | 624/1250 [11:24<11:14,  1.08s/it, loss=0.534]
Epoch 3/4:  50%|████▉     | 624/1250 [11:25<11:14,  1.08s/it, loss=0.406]
Epoch 3/4:  50%|█████     | 625/1250 [11:25<10:59,  1.06s/it, loss=0.406]
Epoch 3/4:  50%|█████     | 625/1250 [11:26<10:59,  1.06s/it, loss=0.554]
Epoch 3/4:  50%|█████     | 626/1250 [11:26<11:00,  1.06s/it, loss=0.554]
Epoch 3/4:  50%|█████     | 626/1250 [11:27<11:00,  1.06s/it, loss=0.466]
Epoch 3/4:  50%|█████     | 627/1250 [11:27<11:07,  1.07s/it, loss=0.466]
Epoch 3/4:  50%|█████     | 627/1250 [11:28<11:07,  1.07s/it, loss=0.588]
Epoch 3/4:  50%|█████     | 628/1250 [11:28<11:14,  1.09s/it, loss=0.588]
Epoch 3/4:  50%|█████     | 628/1250 [11:29<11:14,  1.09s/it, loss=0.541]
Epoch 3/4:  50%|█████     | 629/1250 [11:29<11:07,  1.07s/it, loss=0.541]
Epoch 3/4:  50%|█████     | 629/1250 [11:30<11:07,  1.07s/it, loss=0.527]
Epoch 3/4:  50%|█████     | 630/1250 [11:30<10:53,  1.05s/it, loss=0.527]
Epoch 3/4:  50%|█████     | 630/1250 [11:31<10:53,  1.05s/it, loss=0.512]
Epoch 3/4:  50%|█████     | 631/1250 [11:31<10:52,  1.05s/it, loss=0.512]
Epoch 3/4:  50%|█████     | 631/1250 [11:32<10:52,  1.05s/it, loss=0.426]
Epoch 3/4:  51%|█████     | 632/1250 [11:32<11:02,  1.07s/it, loss=0.426]
Epoch 3/4:  51%|█████     | 632/1250 [11:33<11:02,  1.07s/it, loss=0.472]
Epoch 3/4:  51%|█████     | 633/1250 [11:33<10:58,  1.07s/it, loss=0.472]
Epoch 3/4:  51%|█████     | 633/1250 [11:34<10:58,  1.07s/it, loss=0.595]
Epoch 3/4:  51%|█████     | 634/1250 [11:34<10:47,  1.05s/it, loss=0.595]
Epoch 3/4:  51%|█████     | 634/1250 [11:35<10:47,  1.05s/it, loss=0.525]
Epoch 3/4:  51%|█████     | 635/1250 [11:35<10:47,  1.05s/it, loss=0.525]
Epoch 3/4:  51%|█████     | 635/1250 [11:36<10:47,  1.05s/it, loss=0.535]
Epoch 3/4:  51%|█████     | 636/1250 [11:36<11:04,  1.08s/it, loss=0.535]
Epoch 3/4:  51%|█████     | 636/1250 [11:38<11:04,  1.08s/it, loss=0.439]
Epoch 3/4:  51%|█████     | 637/1250 [11:38<11:20,  1.11s/it, loss=0.439]
Epoch 3/4:  51%|█████     | 637/1250 [11:39<11:20,  1.11s/it, loss=0.521]
Epoch 3/4:  51%|█████     | 638/1250 [11:39<11:10,  1.09s/it, loss=0.521]
Epoch 3/4:  51%|█████     | 638/1250 [11:40<11:10,  1.09s/it, loss=0.517]
Epoch 3/4:  51%|█████     | 639/1250 [11:40<10:51,  1.07s/it, loss=0.517]
Epoch 3/4:  51%|█████     | 639/1250 [11:41<10:51,  1.07s/it, loss=0.697]
Epoch 3/4:  51%|█████     | 640/1250 [11:41<10:48,  1.06s/it, loss=0.697]
Epoch 3/4:  51%|█████     | 640/1250 [11:42<10:48,  1.06s/it, loss=0.523]
Epoch 3/4:  51%|█████▏    | 641/1250 [11:42<10:57,  1.08s/it, loss=0.523]
Epoch 3/4:  51%|█████▏    | 641/1250 [11:43<10:57,  1.08s/it, loss=0.365]
Epoch 3/4:  51%|█████▏    | 642/1250 [11:43<10:56,  1.08s/it, loss=0.365]
Epoch 3/4:  51%|█████▏    | 642/1250 [11:44<10:56,  1.08s/it, loss=0.804]
Epoch 3/4:  51%|█████▏    | 643/1250 [11:44<10:44,  1.06s/it, loss=0.804]
Epoch 3/4:  51%|█████▏    | 643/1250 [11:45<10:44,  1.06s/it, loss=0.362]
Epoch 3/4:  52%|█████▏    | 644/1250 [11:45<10:34,  1.05s/it, loss=0.362]
Epoch 3/4:  52%|█████▏    | 644/1250 [11:46<10:34,  1.05s/it, loss=0.635]
Epoch 3/4:  52%|█████▏    | 645/1250 [11:46<10:51,  1.08s/it, loss=0.635]
Epoch 3/4:  52%|█████▏    | 645/1250 [11:47<10:51,  1.08s/it, loss=0.612]
Epoch 3/4:  52%|█████▏    | 646/1250 [11:47<11:10,  1.11s/it, loss=0.612]
Epoch 3/4:  52%|█████▏    | 646/1250 [11:48<11:10,  1.11s/it, loss=0.48] 
Epoch 3/4:  52%|█████▏    | 647/1250 [11:48<11:01,  1.10s/it, loss=0.48]
Epoch 3/4:  52%|█████▏    | 647/1250 [11:49<11:01,  1.10s/it, loss=0.458]
Epoch 3/4:  52%|█████▏    | 648/1250 [11:49<10:45,  1.07s/it, loss=0.458]
Epoch 3/4:  52%|█████▏    | 648/1250 [11:50<10:45,  1.07s/it, loss=0.428]
Epoch 3/4:  52%|█████▏    | 649/1250 [11:50<10:40,  1.07s/it, loss=0.428]
Epoch 3/4:  52%|█████▏    | 649/1250 [11:52<10:40,  1.07s/it, loss=0.486]
Epoch 3/4:  52%|█████▏    | 650/1250 [11:52<10:53,  1.09s/it, loss=0.486]
Epoch 3/4:  52%|█████▏    | 650/1250 [11:53<10:53,  1.09s/it, loss=0.477]
Epoch 3/4:  52%|█████▏    | 651/1250 [11:53<10:54,  1.09s/it, loss=0.477]
Epoch 3/4:  52%|█████▏    | 651/1250 [11:54<10:54,  1.09s/it, loss=0.657]
Epoch 3/4:  52%|█████▏    | 652/1250 [11:54<10:39,  1.07s/it, loss=0.657]
Epoch 3/4:  52%|█████▏    | 652/1250 [11:55<10:39,  1.07s/it, loss=0.483]
Epoch 3/4:  52%|█████▏    | 653/1250 [11:55<10:24,  1.05s/it, loss=0.483]
Epoch 3/4:  52%|█████▏    | 653/1250 [11:56<10:24,  1.05s/it, loss=0.716]
Epoch 3/4:  52%|█████▏    | 654/1250 [11:56<10:30,  1.06s/it, loss=0.716]
Epoch 3/4:  52%|█████▏    | 654/1250 [11:57<10:30,  1.06s/it, loss=0.704]
Epoch 3/4:  52%|█████▏    | 655/1250 [11:57<10:48,  1.09s/it, loss=0.704]
Epoch 3/4:  52%|█████▏    | 655/1250 [11:58<10:48,  1.09s/it, loss=0.437]
Epoch 3/4:  52%|█████▏    | 656/1250 [11:58<10:48,  1.09s/it, loss=0.437]
Epoch 3/4:  52%|█████▏    | 656/1250 [11:59<10:48,  1.09s/it, loss=0.699]
Epoch 3/4:  53%|█████▎    | 657/1250 [11:59<10:32,  1.07s/it, loss=0.699]
Epoch 3/4:  53%|█████▎    | 657/1250 [12:00<10:32,  1.07s/it, loss=0.856]
Epoch 3/4:  53%|█████▎    | 658/1250 [12:00<10:19,  1.05s/it, loss=0.856]
Epoch 3/4:  53%|█████▎    | 658/1250 [12:01<10:19,  1.05s/it, loss=0.592]
Epoch 3/4:  53%|█████▎    | 659/1250 [12:01<10:25,  1.06s/it, loss=0.592]
Epoch 3/4:  53%|█████▎    | 659/1250 [12:02<10:25,  1.06s/it, loss=0.585]
Epoch 3/4:  53%|█████▎    | 660/1250 [12:02<10:40,  1.09s/it, loss=0.585]
Epoch 3/4:  53%|█████▎    | 660/1250 [12:03<10:40,  1.09s/it, loss=0.488]
Epoch 3/4:  53%|█████▎    | 661/1250 [12:03<10:44,  1.09s/it, loss=0.488]
Epoch 3/4:  53%|█████▎    | 661/1250 [12:04<10:44,  1.09s/it, loss=0.41] 
Epoch 3/4:  53%|█████▎    | 662/1250 [12:04<10:38,  1.09s/it, loss=0.41]
Epoch 3/4:  53%|█████▎    | 662/1250 [12:05<10:38,  1.09s/it, loss=0.686]
Epoch 3/4:  53%|█████▎    | 663/1250 [12:05<10:27,  1.07s/it, loss=0.686]
Epoch 3/4:  53%|█████▎    | 663/1250 [12:07<10:27,  1.07s/it, loss=0.461]
Epoch 3/4:  53%|█████▎    | 664/1250 [12:07<10:23,  1.06s/it, loss=0.461]
Epoch 3/4:  53%|█████▎    | 664/1250 [12:08<10:23,  1.06s/it, loss=0.633]
Epoch 3/4:  53%|█████▎    | 665/1250 [12:08<10:26,  1.07s/it, loss=0.633]
Epoch 3/4:  53%|█████▎    | 665/1250 [12:09<10:26,  1.07s/it, loss=0.437]
Epoch 3/4:  53%|█████▎    | 666/1250 [12:09<10:20,  1.06s/it, loss=0.437]
Epoch 3/4:  53%|█████▎    | 666/1250 [12:10<10:20,  1.06s/it, loss=0.675]
Epoch 3/4:  53%|█████▎    | 667/1250 [12:10<10:11,  1.05s/it, loss=0.675]
Epoch 3/4:  53%|█████▎    | 667/1250 [12:11<10:11,  1.05s/it, loss=0.472]
Epoch 3/4:  53%|█████▎    | 668/1250 [12:11<10:23,  1.07s/it, loss=0.472]
Epoch 3/4:  53%|█████▎    | 668/1250 [12:12<10:23,  1.07s/it, loss=0.547]
Epoch 3/4:  54%|█████▎    | 669/1250 [12:12<11:02,  1.14s/it, loss=0.547]
Epoch 3/4:  54%|█████▎    | 669/1250 [12:13<11:02,  1.14s/it, loss=0.385]
Epoch 3/4:  54%|█████▎    | 670/1250 [12:13<11:03,  1.14s/it, loss=0.385]
Epoch 3/4:  54%|█████▎    | 670/1250 [12:14<11:03,  1.14s/it, loss=0.802]
Epoch 3/4:  54%|█████▎    | 671/1250 [12:14<10:38,  1.10s/it, loss=0.802]
Epoch 3/4:  54%|█████▎    | 671/1250 [12:15<10:38,  1.10s/it, loss=0.56] 
Epoch 3/4:  54%|█████▍    | 672/1250 [12:15<10:18,  1.07s/it, loss=0.56]
Epoch 3/4:  54%|█████▍    | 672/1250 [12:16<10:18,  1.07s/it, loss=0.74]
Epoch 3/4:  54%|█████▍    | 673/1250 [12:16<10:15,  1.07s/it, loss=0.74]
Epoch 3/4:  54%|█████▍    | 673/1250 [12:17<10:15,  1.07s/it, loss=0.567]
Epoch 3/4:  54%|█████▍    | 674/1250 [12:17<10:24,  1.08s/it, loss=0.567]
Epoch 3/4:  54%|█████▍    | 674/1250 [12:18<10:24,  1.08s/it, loss=0.595]
Epoch 3/4:  54%|█████▍    | 675/1250 [12:18<10:21,  1.08s/it, loss=0.595]
Epoch 3/4:  54%|█████▍    | 675/1250 [12:20<10:21,  1.08s/it, loss=0.531]
Epoch 3/4:  54%|█████▍    | 676/1250 [12:20<10:13,  1.07s/it, loss=0.531]
Epoch 3/4:  54%|█████▍    | 676/1250 [12:21<10:13,  1.07s/it, loss=0.585]
Epoch 3/4:  54%|█████▍    | 677/1250 [12:21<10:08,  1.06s/it, loss=0.585]
Epoch 3/4:  54%|█████▍    | 677/1250 [12:22<10:08,  1.06s/it, loss=0.46] 
Epoch 3/4:  54%|█████▍    | 678/1250 [12:22<10:15,  1.08s/it, loss=0.46]
Epoch 3/4:  54%|█████▍    | 678/1250 [12:23<10:15,  1.08s/it, loss=0.673]
Epoch 3/4:  54%|█████▍    | 679/1250 [12:23<10:19,  1.09s/it, loss=0.673]
Epoch 3/4:  54%|█████▍    | 679/1250 [12:24<10:19,  1.09s/it, loss=0.37] 
Epoch 3/4:  54%|█████▍    | 680/1250 [12:24<10:11,  1.07s/it, loss=0.37]
Epoch 3/4:  54%|█████▍    | 680/1250 [12:25<10:11,  1.07s/it, loss=0.633]
Epoch 3/4:  54%|█████▍    | 681/1250 [12:25<09:59,  1.05s/it, loss=0.633]
Epoch 3/4:  54%|█████▍    | 681/1250 [12:26<09:59,  1.05s/it, loss=0.491]
Epoch 3/4:  55%|█████▍    | 682/1250 [12:26<10:01,  1.06s/it, loss=0.491]
Epoch 3/4:  55%|█████▍    | 682/1250 [12:27<10:01,  1.06s/it, loss=0.666]
Epoch 3/4:  55%|█████▍    | 683/1250 [12:27<10:15,  1.09s/it, loss=0.666]
Epoch 3/4:  55%|█████▍    | 683/1250 [12:28<10:15,  1.09s/it, loss=0.588]
Epoch 3/4:  55%|█████▍    | 684/1250 [12:28<10:20,  1.10s/it, loss=0.588]
Epoch 3/4:  55%|█████▍    | 684/1250 [12:29<10:20,  1.10s/it, loss=0.518]
Epoch 3/4:  55%|█████▍    | 685/1250 [12:29<10:11,  1.08s/it, loss=0.518]
Epoch 3/4:  55%|█████▍    | 685/1250 [12:30<10:11,  1.08s/it, loss=0.485]
Epoch 3/4:  55%|█████▍    | 686/1250 [12:30<10:07,  1.08s/it, loss=0.485]
Epoch 3/4:  55%|█████▍    | 686/1250 [12:31<10:07,  1.08s/it, loss=0.622]
Epoch 3/4:  55%|█████▍    | 687/1250 [12:31<10:08,  1.08s/it, loss=0.622]
Epoch 3/4:  55%|█████▍    | 687/1250 [12:33<10:08,  1.08s/it, loss=0.907]
Epoch 3/4:  55%|█████▌    | 688/1250 [12:33<10:17,  1.10s/it, loss=0.907]
Epoch 3/4:  55%|█████▌    | 688/1250 [12:34<10:17,  1.10s/it, loss=0.647]
Epoch 3/4:  55%|█████▌    | 689/1250 [12:34<10:12,  1.09s/it, loss=0.647]
Epoch 3/4:  55%|█████▌    | 689/1250 [12:35<10:12,  1.09s/it, loss=0.548]
Epoch 3/4:  55%|█████▌    | 690/1250 [12:35<10:03,  1.08s/it, loss=0.548]
Epoch 3/4:  55%|█████▌    | 690/1250 [12:36<10:03,  1.08s/it, loss=0.504]
Epoch 3/4:  55%|█████▌    | 691/1250 [12:36<10:04,  1.08s/it, loss=0.504]
Epoch 3/4:  55%|█████▌    | 691/1250 [12:37<10:04,  1.08s/it, loss=0.591]
Epoch 3/4:  55%|█████▌    | 692/1250 [12:37<10:13,  1.10s/it, loss=0.591]
Epoch 3/4:  55%|█████▌    | 692/1250 [12:38<10:13,  1.10s/it, loss=0.642]
Epoch 3/4:  55%|█████▌    | 693/1250 [12:38<10:11,  1.10s/it, loss=0.642]
Epoch 3/4:  55%|█████▌    | 693/1250 [12:39<10:11,  1.10s/it, loss=0.487]
Epoch 3/4:  56%|█████▌    | 694/1250 [12:39<09:58,  1.08s/it, loss=0.487]
Epoch 3/4:  56%|█████▌    | 694/1250 [12:40<09:58,  1.08s/it, loss=0.371]
Epoch 3/4:  56%|█████▌    | 695/1250 [12:40<09:48,  1.06s/it, loss=0.371]
Epoch 3/4:  56%|█████▌    | 695/1250 [12:41<09:48,  1.06s/it, loss=0.677]
Epoch 3/4:  56%|█████▌    | 696/1250 [12:41<09:49,  1.06s/it, loss=0.677]
Epoch 3/4:  56%|█████▌    | 696/1250 [12:42<09:49,  1.06s/it, loss=0.476]
Epoch 3/4:  56%|█████▌    | 697/1250 [12:42<10:00,  1.09s/it, loss=0.476]
Epoch 3/4:  56%|█████▌    | 697/1250 [12:43<10:00,  1.09s/it, loss=0.738]
Epoch 3/4:  56%|█████▌    | 698/1250 [12:43<10:01,  1.09s/it, loss=0.738]
Epoch 3/4:  56%|█████▌    | 698/1250 [12:44<10:01,  1.09s/it, loss=0.756]
Epoch 3/4:  56%|█████▌    | 699/1250 [12:44<09:50,  1.07s/it, loss=0.756]
Epoch 3/4:  56%|█████▌    | 699/1250 [12:45<09:50,  1.07s/it, loss=0.576]
Epoch 3/4:  56%|█████▌    | 700/1250 [12:45<09:44,  1.06s/it, loss=0.576]
Epoch 3/4:  56%|█████▌    | 700/1250 [12:47<09:44,  1.06s/it, loss=0.683]
Epoch 3/4:  56%|█████▌    | 701/1250 [12:47<09:52,  1.08s/it, loss=0.683]
Epoch 3/4:  56%|█████▌    | 701/1250 [12:48<09:52,  1.08s/it, loss=0.521]
Epoch 3/4:  56%|█████▌    | 702/1250 [12:48<10:01,  1.10s/it, loss=0.521]
Epoch 3/4:  56%|█████▌    | 702/1250 [12:49<10:01,  1.10s/it, loss=0.483]
Epoch 3/4:  56%|█████▌    | 703/1250 [12:49<09:58,  1.09s/it, loss=0.483]
Epoch 3/4:  56%|█████▌    | 703/1250 [12:50<09:58,  1.09s/it, loss=0.617]
Epoch 3/4:  56%|█████▋    | 704/1250 [12:50<09:45,  1.07s/it, loss=0.617]
Epoch 3/4:  56%|█████▋    | 704/1250 [12:51<09:45,  1.07s/it, loss=0.792]
Epoch 3/4:  56%|█████▋    | 705/1250 [12:51<09:34,  1.05s/it, loss=0.792]
Epoch 3/4:  56%|█████▋    | 705/1250 [12:52<09:34,  1.05s/it, loss=0.56] 
Epoch 3/4:  56%|█████▋    | 706/1250 [12:52<09:38,  1.06s/it, loss=0.56]
Epoch 3/4:  56%|█████▋    | 706/1250 [12:53<09:38,  1.06s/it, loss=0.813]
Epoch 3/4:  57%|█████▋    | 707/1250 [12:53<09:45,  1.08s/it, loss=0.813]
Epoch 3/4:  57%|█████▋    | 707/1250 [12:54<09:45,  1.08s/it, loss=0.579]
Epoch 3/4:  57%|█████▋    | 708/1250 [12:54<09:42,  1.07s/it, loss=0.579]
Epoch 3/4:  57%|█████▋    | 708/1250 [12:55<09:42,  1.07s/it, loss=0.825]
Epoch 3/4:  57%|█████▋    | 709/1250 [12:55<09:33,  1.06s/it, loss=0.825]
Epoch 3/4:  57%|█████▋    | 709/1250 [12:56<09:33,  1.06s/it, loss=0.642]
Epoch 3/4:  57%|█████▋    | 710/1250 [12:56<09:32,  1.06s/it, loss=0.642]
Epoch 3/4:  57%|█████▋    | 710/1250 [12:57<09:32,  1.06s/it, loss=0.464]
Epoch 3/4:  57%|█████▋    | 711/1250 [12:57<09:47,  1.09s/it, loss=0.464]
Epoch 3/4:  57%|█████▋    | 711/1250 [12:59<09:47,  1.09s/it, loss=0.563]
Epoch 3/4:  57%|█████▋    | 712/1250 [12:59<10:06,  1.13s/it, loss=0.563]
Epoch 3/4:  57%|█████▋    | 712/1250 [13:00<10:06,  1.13s/it, loss=0.435]
Epoch 3/4:  57%|█████▋    | 713/1250 [13:00<09:56,  1.11s/it, loss=0.435]
Epoch 3/4:  57%|█████▋    | 713/1250 [13:01<09:56,  1.11s/it, loss=0.909]
Epoch 3/4:  57%|█████▋    | 714/1250 [13:01<09:36,  1.08s/it, loss=0.909]
Epoch 3/4:  57%|█████▋    | 714/1250 [13:02<09:36,  1.08s/it, loss=0.58] 
Epoch 3/4:  57%|█████▋    | 715/1250 [13:02<09:27,  1.06s/it, loss=0.58]
Epoch 3/4:  57%|█████▋    | 715/1250 [13:03<09:27,  1.06s/it, loss=0.565]
Epoch 3/4:  57%|█████▋    | 716/1250 [13:03<09:30,  1.07s/it, loss=0.565]
Epoch 3/4:  57%|█████▋    | 716/1250 [13:04<09:30,  1.07s/it, loss=0.601]
Epoch 3/4:  57%|█████▋    | 717/1250 [13:04<09:29,  1.07s/it, loss=0.601]
Epoch 3/4:  57%|█████▋    | 717/1250 [13:05<09:29,  1.07s/it, loss=0.644]
Epoch 3/4:  57%|█████▋    | 718/1250 [13:05<09:22,  1.06s/it, loss=0.644]
Epoch 3/4:  57%|█████▋    | 718/1250 [13:06<09:22,  1.06s/it, loss=0.474]
Epoch 3/4:  58%|█████▊    | 719/1250 [13:06<09:15,  1.05s/it, loss=0.474]
Epoch 3/4:  58%|█████▊    | 719/1250 [13:07<09:15,  1.05s/it, loss=0.5]  
Epoch 3/4:  58%|█████▊    | 720/1250 [13:07<09:23,  1.06s/it, loss=0.5]
Epoch 3/4:  58%|█████▊    | 720/1250 [13:08<09:23,  1.06s/it, loss=0.62]
Epoch 3/4:  58%|█████▊    | 721/1250 [13:08<09:33,  1.08s/it, loss=0.62]
Epoch 3/4:  58%|█████▊    | 721/1250 [13:09<09:33,  1.08s/it, loss=0.554]
Epoch 3/4:  58%|█████▊    | 722/1250 [13:09<09:32,  1.08s/it, loss=0.554]
Epoch 3/4:  58%|█████▊    | 722/1250 [13:10<09:32,  1.08s/it, loss=0.524]
Epoch 3/4:  58%|█████▊    | 723/1250 [13:10<09:27,  1.08s/it, loss=0.524]
Epoch 3/4:  58%|█████▊    | 723/1250 [13:11<09:27,  1.08s/it, loss=0.577]
Epoch 3/4:  58%|█████▊    | 724/1250 [13:11<09:27,  1.08s/it, loss=0.577]
Epoch 3/4:  58%|█████▊    | 724/1250 [13:12<09:27,  1.08s/it, loss=0.547]
Epoch 3/4:  58%|█████▊    | 725/1250 [13:12<09:33,  1.09s/it, loss=0.547]
Epoch 3/4:  58%|█████▊    | 725/1250 [13:13<09:33,  1.09s/it, loss=0.673]
Epoch 3/4:  58%|█████▊    | 726/1250 [13:13<09:34,  1.10s/it, loss=0.673]
Epoch 3/4:  58%|█████▊    | 726/1250 [13:15<09:34,  1.10s/it, loss=0.362]
Epoch 3/4:  58%|█████▊    | 727/1250 [13:15<09:23,  1.08s/it, loss=0.362]
Epoch 3/4:  58%|█████▊    | 727/1250 [13:16<09:23,  1.08s/it, loss=0.759]
Epoch 3/4:  58%|█████▊    | 728/1250 [13:16<09:17,  1.07s/it, loss=0.759]
Epoch 3/4:  58%|█████▊    | 728/1250 [13:17<09:17,  1.07s/it, loss=0.399]
Epoch 3/4:  58%|█████▊    | 729/1250 [13:17<09:33,  1.10s/it, loss=0.399]
Epoch 3/4:  58%|█████▊    | 729/1250 [13:18<09:33,  1.10s/it, loss=0.601]
Epoch 3/4:  58%|█████▊    | 730/1250 [13:18<09:44,  1.12s/it, loss=0.601]
Epoch 3/4:  58%|█████▊    | 730/1250 [13:19<09:44,  1.12s/it, loss=0.655]
Epoch 3/4:  58%|█████▊    | 731/1250 [13:19<09:37,  1.11s/it, loss=0.655]
Epoch 3/4:  58%|█████▊    | 731/1250 [13:20<09:37,  1.11s/it, loss=0.316]
Epoch 3/4:  59%|█████▊    | 732/1250 [13:20<09:21,  1.08s/it, loss=0.316]
Epoch 3/4:  59%|█████▊    | 732/1250 [13:21<09:21,  1.08s/it, loss=0.495]
Epoch 3/4:  59%|█████▊    | 733/1250 [13:21<09:20,  1.08s/it, loss=0.495]
Epoch 3/4:  59%|█████▊    | 733/1250 [13:22<09:20,  1.08s/it, loss=0.676]
Epoch 3/4:  59%|█████▊    | 734/1250 [13:22<09:32,  1.11s/it, loss=0.676]
Epoch 3/4:  59%|█████▊    | 734/1250 [13:23<09:32,  1.11s/it, loss=0.423]
Epoch 3/4:  59%|█████▉    | 735/1250 [13:23<09:38,  1.12s/it, loss=0.423]
Epoch 3/4:  59%|█████▉    | 735/1250 [13:25<09:38,  1.12s/it, loss=0.483]
Epoch 3/4:  59%|█████▉    | 736/1250 [13:25<09:28,  1.11s/it, loss=0.483]
Epoch 3/4:  59%|█████▉    | 736/1250 [13:26<09:28,  1.11s/it, loss=0.519]
Epoch 3/4:  59%|█████▉    | 737/1250 [13:26<09:12,  1.08s/it, loss=0.519]
Epoch 3/4:  59%|█████▉    | 737/1250 [13:27<09:12,  1.08s/it, loss=0.399]
Epoch 3/4:  59%|█████▉    | 738/1250 [13:27<09:04,  1.06s/it, loss=0.399]
Epoch 3/4:  59%|█████▉    | 738/1250 [13:28<09:04,  1.06s/it, loss=0.519]
Epoch 3/4:  59%|█████▉    | 739/1250 [13:28<09:07,  1.07s/it, loss=0.519]
Epoch 3/4:  59%|█████▉    | 739/1250 [13:29<09:07,  1.07s/it, loss=0.567]
Epoch 3/4:  59%|█████▉    | 740/1250 [13:29<09:06,  1.07s/it, loss=0.567]
Epoch 3/4:  59%|█████▉    | 740/1250 [13:30<09:06,  1.07s/it, loss=0.419]
Epoch 3/4:  59%|█████▉    | 741/1250 [13:30<08:55,  1.05s/it, loss=0.419]
Epoch 3/4:  59%|█████▉    | 741/1250 [13:31<08:55,  1.05s/it, loss=0.685]
Epoch 3/4:  59%|█████▉    | 742/1250 [13:31<08:47,  1.04s/it, loss=0.685]
Epoch 3/4:  59%|█████▉    | 742/1250 [13:32<08:47,  1.04s/it, loss=0.544]
Epoch 3/4:  59%|█████▉    | 743/1250 [13:32<08:51,  1.05s/it, loss=0.544]
Epoch 3/4:  59%|█████▉    | 743/1250 [13:33<08:51,  1.05s/it, loss=0.443]
Epoch 3/4:  60%|█████▉    | 744/1250 [13:33<09:10,  1.09s/it, loss=0.443]
Epoch 3/4:  60%|█████▉    | 744/1250 [13:34<09:10,  1.09s/it, loss=0.797]
Epoch 3/4:  60%|█████▉    | 745/1250 [13:34<09:20,  1.11s/it, loss=0.797]
Epoch 3/4:  60%|█████▉    | 745/1250 [13:35<09:20,  1.11s/it, loss=0.658]
Epoch 3/4:  60%|█████▉    | 746/1250 [13:35<09:08,  1.09s/it, loss=0.658]
Epoch 3/4:  60%|█████▉    | 746/1250 [13:36<09:08,  1.09s/it, loss=0.891]
Epoch 3/4:  60%|█████▉    | 747/1250 [13:36<08:58,  1.07s/it, loss=0.891]
Epoch 3/4:  60%|█████▉    | 747/1250 [13:37<08:58,  1.07s/it, loss=0.685]
Epoch 3/4:  60%|█████▉    | 748/1250 [13:37<09:04,  1.08s/it, loss=0.685]
Epoch 3/4:  60%|█████▉    | 748/1250 [13:39<09:04,  1.08s/it, loss=0.439]
Epoch 3/4:  60%|█████▉    | 749/1250 [13:39<09:21,  1.12s/it, loss=0.439]
Epoch 3/4:  60%|█████▉    | 749/1250 [13:40<09:21,  1.12s/it, loss=0.801]
Epoch 3/4:  60%|██████    | 750/1250 [13:40<09:13,  1.11s/it, loss=0.801]
Epoch 3/4:  60%|██████    | 750/1250 [13:41<09:13,  1.11s/it, loss=0.649]
Epoch 3/4:  60%|██████    | 751/1250 [13:41<08:55,  1.07s/it, loss=0.649]
Epoch 3/4:  60%|██████    | 751/1250 [13:42<08:55,  1.07s/it, loss=0.62] 
Epoch 3/4:  60%|██████    | 752/1250 [13:42<08:44,  1.05s/it, loss=0.62]
Epoch 3/4:  60%|██████    | 752/1250 [13:43<08:44,  1.05s/it, loss=0.614]
Epoch 3/4:  60%|██████    | 753/1250 [13:43<08:50,  1.07s/it, loss=0.614]
Epoch 3/4:  60%|██████    | 753/1250 [13:44<08:50,  1.07s/it, loss=0.445]
Epoch 3/4:  60%|██████    | 754/1250 [13:44<09:04,  1.10s/it, loss=0.445]
Epoch 3/4:  60%|██████    | 754/1250 [13:45<09:04,  1.10s/it, loss=0.492]
Epoch 3/4:  60%|██████    | 755/1250 [13:45<09:09,  1.11s/it, loss=0.492]
Epoch 3/4:  60%|██████    | 755/1250 [13:46<09:09,  1.11s/it, loss=0.537]
Epoch 3/4:  60%|██████    | 756/1250 [13:46<09:02,  1.10s/it, loss=0.537]
Epoch 3/4:  60%|██████    | 756/1250 [13:47<09:02,  1.10s/it, loss=0.445]
Epoch 3/4:  61%|██████    | 757/1250 [13:47<08:54,  1.08s/it, loss=0.445]
Epoch 3/4:  61%|██████    | 757/1250 [13:48<08:54,  1.08s/it, loss=0.697]
Epoch 3/4:  61%|██████    | 758/1250 [13:48<08:55,  1.09s/it, loss=0.697]
Epoch 3/4:  61%|██████    | 758/1250 [13:49<08:55,  1.09s/it, loss=0.495]
Epoch 3/4:  61%|██████    | 759/1250 [13:49<08:54,  1.09s/it, loss=0.495]
Epoch 3/4:  61%|██████    | 759/1250 [13:50<08:54,  1.09s/it, loss=0.421]
Epoch 3/4:  61%|██████    | 760/1250 [13:50<08:41,  1.06s/it, loss=0.421]
Epoch 3/4:  61%|██████    | 760/1250 [13:51<08:41,  1.06s/it, loss=0.603]
Epoch 3/4:  61%|██████    | 761/1250 [13:51<08:32,  1.05s/it, loss=0.603]
Epoch 3/4:  61%|██████    | 761/1250 [13:52<08:32,  1.05s/it, loss=0.591]
Epoch 3/4:  61%|██████    | 762/1250 [13:52<08:43,  1.07s/it, loss=0.591]
Epoch 3/4:  61%|██████    | 762/1250 [13:54<08:43,  1.07s/it, loss=0.418]
Epoch 3/4:  61%|██████    | 763/1250 [13:54<08:51,  1.09s/it, loss=0.418]
Epoch 3/4:  61%|██████    | 763/1250 [13:55<08:51,  1.09s/it, loss=0.52] 
Epoch 3/4:  61%|██████    | 764/1250 [13:55<08:46,  1.08s/it, loss=0.52]
Epoch 3/4:  61%|██████    | 764/1250 [13:56<08:46,  1.08s/it, loss=0.439]
Epoch 3/4:  61%|██████    | 765/1250 [13:56<08:33,  1.06s/it, loss=0.439]
Epoch 3/4:  61%|██████    | 765/1250 [13:57<08:33,  1.06s/it, loss=0.442]
Epoch 3/4:  61%|██████▏   | 766/1250 [13:57<08:26,  1.05s/it, loss=0.442]
Epoch 3/4:  61%|██████▏   | 766/1250 [13:58<08:26,  1.05s/it, loss=0.512]
Epoch 3/4:  61%|██████▏   | 767/1250 [13:58<08:35,  1.07s/it, loss=0.512]
Epoch 3/4:  61%|██████▏   | 767/1250 [13:59<08:35,  1.07s/it, loss=0.566]
Epoch 3/4:  61%|██████▏   | 768/1250 [13:59<08:43,  1.09s/it, loss=0.566]
Epoch 3/4:  61%|██████▏   | 768/1250 [14:00<08:43,  1.09s/it, loss=0.543]
Epoch 3/4:  62%|██████▏   | 769/1250 [14:00<08:39,  1.08s/it, loss=0.543]
Epoch 3/4:  62%|██████▏   | 769/1250 [14:01<08:39,  1.08s/it, loss=0.724]
Epoch 3/4:  62%|██████▏   | 770/1250 [14:01<08:28,  1.06s/it, loss=0.724]
Epoch 3/4:  62%|██████▏   | 770/1250 [14:02<08:28,  1.06s/it, loss=0.87] 
Epoch 3/4:  62%|██████▏   | 771/1250 [14:02<08:26,  1.06s/it, loss=0.87]
Epoch 3/4:  62%|██████▏   | 771/1250 [14:03<08:26,  1.06s/it, loss=0.585]
Epoch 3/4:  62%|██████▏   | 772/1250 [14:03<08:42,  1.09s/it, loss=0.585]
Epoch 3/4:  62%|██████▏   | 772/1250 [14:04<08:42,  1.09s/it, loss=0.509]
Epoch 3/4:  62%|██████▏   | 773/1250 [14:04<08:44,  1.10s/it, loss=0.509]
Epoch 3/4:  62%|██████▏   | 773/1250 [14:05<08:44,  1.10s/it, loss=0.74] 
Epoch 3/4:  62%|██████▏   | 774/1250 [14:05<08:29,  1.07s/it, loss=0.74]
Epoch 3/4:  62%|██████▏   | 774/1250 [14:06<08:29,  1.07s/it, loss=0.843]
Epoch 3/4:  62%|██████▏   | 775/1250 [14:06<08:19,  1.05s/it, loss=0.843]
Epoch 3/4:  62%|██████▏   | 775/1250 [14:07<08:19,  1.05s/it, loss=0.514]
Epoch 3/4:  62%|██████▏   | 776/1250 [14:07<08:29,  1.07s/it, loss=0.514]
Epoch 3/4:  62%|██████▏   | 776/1250 [14:09<08:29,  1.07s/it, loss=0.601]
Epoch 3/4:  62%|██████▏   | 777/1250 [14:09<08:43,  1.11s/it, loss=0.601]
Epoch 3/4:  62%|██████▏   | 777/1250 [14:10<08:43,  1.11s/it, loss=0.771]
Epoch 3/4:  62%|██████▏   | 778/1250 [14:10<08:35,  1.09s/it, loss=0.771]
Epoch 3/4:  62%|██████▏   | 778/1250 [14:11<08:35,  1.09s/it, loss=0.503]
Epoch 3/4:  62%|██████▏   | 779/1250 [14:11<08:24,  1.07s/it, loss=0.503]
Epoch 3/4:  62%|██████▏   | 779/1250 [14:12<08:24,  1.07s/it, loss=0.63] 
Epoch 3/4:  62%|██████▏   | 780/1250 [14:12<08:22,  1.07s/it, loss=0.63]
Epoch 3/4:  62%|██████▏   | 780/1250 [14:13<08:22,  1.07s/it, loss=0.592]
Epoch 3/4:  62%|██████▏   | 781/1250 [14:13<08:32,  1.09s/it, loss=0.592]
Epoch 3/4:  62%|██████▏   | 781/1250 [14:14<08:32,  1.09s/it, loss=0.448]
Epoch 3/4:  63%|██████▎   | 782/1250 [14:14<08:36,  1.10s/it, loss=0.448]
Epoch 3/4:  63%|██████▎   | 782/1250 [14:15<08:36,  1.10s/it, loss=0.491]
Epoch 3/4:  63%|██████▎   | 783/1250 [14:15<08:30,  1.09s/it, loss=0.491]
Epoch 3/4:  63%|██████▎   | 783/1250 [14:16<08:30,  1.09s/it, loss=0.755]
Epoch 3/4:  63%|██████▎   | 784/1250 [14:16<08:20,  1.07s/it, loss=0.755]
Epoch 3/4:  63%|██████▎   | 784/1250 [14:17<08:20,  1.07s/it, loss=0.545]
Epoch 3/4:  63%|██████▎   | 785/1250 [14:17<08:14,  1.06s/it, loss=0.545]
Epoch 3/4:  63%|██████▎   | 785/1250 [14:18<08:14,  1.06s/it, loss=0.36] 
Epoch 3/4:  63%|██████▎   | 786/1250 [14:18<08:17,  1.07s/it, loss=0.36]
Epoch 3/4:  63%|██████▎   | 786/1250 [14:19<08:17,  1.07s/it, loss=0.521]
Epoch 3/4:  63%|██████▎   | 787/1250 [14:19<08:22,  1.08s/it, loss=0.521]
Epoch 3/4:  63%|██████▎   | 787/1250 [14:20<08:22,  1.08s/it, loss=0.492]
Epoch 3/4:  63%|██████▎   | 788/1250 [14:20<08:14,  1.07s/it, loss=0.492]
Epoch 3/4:  63%|██████▎   | 788/1250 [14:21<08:14,  1.07s/it, loss=0.655]
Epoch 3/4:  63%|██████▎   | 789/1250 [14:21<08:03,  1.05s/it, loss=0.655]
Epoch 3/4:  63%|██████▎   | 789/1250 [14:23<08:03,  1.05s/it, loss=0.663]
Epoch 3/4:  63%|██████▎   | 790/1250 [14:23<08:02,  1.05s/it, loss=0.663]
Epoch 3/4:  63%|██████▎   | 790/1250 [14:24<08:02,  1.05s/it, loss=0.7]  
Epoch 3/4:  63%|██████▎   | 791/1250 [14:24<08:15,  1.08s/it, loss=0.7]
Epoch 3/4:  63%|██████▎   | 791/1250 [14:25<08:15,  1.08s/it, loss=0.723]
Epoch 3/4:  63%|██████▎   | 792/1250 [14:25<08:15,  1.08s/it, loss=0.723]
Epoch 3/4:  63%|██████▎   | 792/1250 [14:26<08:15,  1.08s/it, loss=0.521]
Epoch 3/4:  63%|██████▎   | 793/1250 [14:26<08:03,  1.06s/it, loss=0.521]
Epoch 3/4:  63%|██████▎   | 793/1250 [14:27<08:03,  1.06s/it, loss=0.534]
Epoch 3/4:  64%|██████▎   | 794/1250 [14:27<07:53,  1.04s/it, loss=0.534]
Epoch 3/4:  64%|██████▎   | 794/1250 [14:28<07:53,  1.04s/it, loss=0.692]
Epoch 3/4:  64%|██████▎   | 795/1250 [14:28<08:02,  1.06s/it, loss=0.692]
Epoch 3/4:  64%|██████▎   | 795/1250 [14:29<08:02,  1.06s/it, loss=0.498]
Epoch 3/4:  64%|██████▎   | 796/1250 [14:29<08:18,  1.10s/it, loss=0.498]
Epoch 3/4:  64%|██████▎   | 796/1250 [14:30<08:18,  1.10s/it, loss=0.774]
Epoch 3/4:  64%|██████▍   | 797/1250 [14:30<08:18,  1.10s/it, loss=0.774]
Epoch 3/4:  64%|██████▍   | 797/1250 [14:31<08:18,  1.10s/it, loss=0.543]
Epoch 3/4:  64%|██████▍   | 798/1250 [14:31<08:07,  1.08s/it, loss=0.543]
Epoch 3/4:  64%|██████▍   | 798/1250 [14:32<08:07,  1.08s/it, loss=0.637]
Epoch 3/4:  64%|██████▍   | 799/1250 [14:32<08:00,  1.06s/it, loss=0.637]
Epoch 3/4:  64%|██████▍   | 799/1250 [14:33<08:00,  1.06s/it, loss=0.536]
Epoch 3/4:  64%|██████▍   | 800/1250 [14:33<08:01,  1.07s/it, loss=0.536]
Epoch 3/4:  64%|██████▍   | 800/1250 [14:34<08:01,  1.07s/it, loss=0.571]
Epoch 3/4:  64%|██████▍   | 801/1250 [14:34<08:02,  1.07s/it, loss=0.571]
Epoch 3/4:  64%|██████▍   | 801/1250 [14:35<08:02,  1.07s/it, loss=0.504]
Epoch 3/4:  64%|██████▍   | 802/1250 [14:35<07:56,  1.06s/it, loss=0.504]
Epoch 3/4:  64%|██████▍   | 802/1250 [14:36<07:56,  1.06s/it, loss=0.336]
Epoch 3/4:  64%|██████▍   | 803/1250 [14:36<07:46,  1.04s/it, loss=0.336]
Epoch 3/4:  64%|██████▍   | 803/1250 [14:37<07:46,  1.04s/it, loss=0.64] 
Epoch 3/4:  64%|██████▍   | 804/1250 [14:37<07:47,  1.05s/it, loss=0.64]
Epoch 3/4:  64%|██████▍   | 804/1250 [14:39<07:47,  1.05s/it, loss=0.633]
Epoch 3/4:  64%|██████▍   | 805/1250 [14:39<07:58,  1.08s/it, loss=0.633]
Epoch 3/4:  64%|██████▍   | 805/1250 [14:40<07:58,  1.08s/it, loss=0.483]
Epoch 3/4:  64%|██████▍   | 806/1250 [14:40<08:02,  1.09s/it, loss=0.483]
Epoch 3/4:  64%|██████▍   | 806/1250 [14:41<08:02,  1.09s/it, loss=0.442]
Epoch 3/4:  65%|██████▍   | 807/1250 [14:41<07:51,  1.07s/it, loss=0.442]
Epoch 3/4:  65%|██████▍   | 807/1250 [14:42<07:51,  1.07s/it, loss=0.676]
Epoch 3/4:  65%|██████▍   | 808/1250 [14:42<07:42,  1.05s/it, loss=0.676]
Epoch 3/4:  65%|██████▍   | 808/1250 [14:43<07:42,  1.05s/it, loss=0.464]
Epoch 3/4:  65%|██████▍   | 809/1250 [14:43<07:49,  1.06s/it, loss=0.464]
Epoch 3/4:  65%|██████▍   | 809/1250 [14:44<07:49,  1.06s/it, loss=0.623]
Epoch 3/4:  65%|██████▍   | 810/1250 [14:44<07:58,  1.09s/it, loss=0.623]
Epoch 3/4:  65%|██████▍   | 810/1250 [14:45<07:58,  1.09s/it, loss=0.583]
Epoch 3/4:  65%|██████▍   | 811/1250 [14:45<07:59,  1.09s/it, loss=0.583]
Epoch 3/4:  65%|██████▍   | 811/1250 [14:46<07:59,  1.09s/it, loss=0.681]
Epoch 3/4:  65%|██████▍   | 812/1250 [14:46<07:49,  1.07s/it, loss=0.681]
Epoch 3/4:  65%|██████▍   | 812/1250 [14:47<07:49,  1.07s/it, loss=0.792]
Epoch 3/4:  65%|██████▌   | 813/1250 [14:47<07:38,  1.05s/it, loss=0.792]
Epoch 3/4:  65%|██████▌   | 813/1250 [14:48<07:38,  1.05s/it, loss=0.691]
Epoch 3/4:  65%|██████▌   | 814/1250 [14:48<07:42,  1.06s/it, loss=0.691]
Epoch 3/4:  65%|██████▌   | 814/1250 [14:49<07:42,  1.06s/it, loss=0.628]
Epoch 3/4:  65%|██████▌   | 815/1250 [14:49<07:50,  1.08s/it, loss=0.628]
Epoch 3/4:  65%|██████▌   | 815/1250 [14:50<07:50,  1.08s/it, loss=0.389]
Epoch 3/4:  65%|██████▌   | 816/1250 [14:50<07:50,  1.08s/it, loss=0.389]
Epoch 3/4:  65%|██████▌   | 816/1250 [14:52<07:50,  1.08s/it, loss=0.603]
Epoch 3/4:  65%|██████▌   | 817/1250 [14:52<07:50,  1.09s/it, loss=0.603]
Epoch 3/4:  65%|██████▌   | 817/1250 [14:53<07:50,  1.09s/it, loss=0.36] 
Epoch 3/4:  65%|██████▌   | 818/1250 [14:53<07:48,  1.09s/it, loss=0.36]
Epoch 3/4:  65%|██████▌   | 818/1250 [14:54<07:48,  1.09s/it, loss=0.47]
Epoch 3/4:  66%|██████▌   | 819/1250 [14:54<07:45,  1.08s/it, loss=0.47]
Epoch 3/4:  66%|██████▌   | 819/1250 [14:55<07:45,  1.08s/it, loss=0.532]
Epoch 3/4:  66%|██████▌   | 820/1250 [14:55<07:41,  1.07s/it, loss=0.532]
Epoch 3/4:  66%|██████▌   | 820/1250 [14:56<07:41,  1.07s/it, loss=0.599]
Epoch 3/4:  66%|██████▌   | 821/1250 [14:56<07:31,  1.05s/it, loss=0.599]
Epoch 3/4:  66%|██████▌   | 821/1250 [14:57<07:31,  1.05s/it, loss=0.733]
Epoch 3/4:  66%|██████▌   | 822/1250 [14:57<07:20,  1.03s/it, loss=0.733]
Epoch 3/4:  66%|██████▌   | 822/1250 [14:58<07:20,  1.03s/it, loss=0.699]
Epoch 3/4:  66%|██████▌   | 823/1250 [14:58<07:23,  1.04s/it, loss=0.699]
Epoch 3/4:  66%|██████▌   | 823/1250 [14:59<07:23,  1.04s/it, loss=0.557]
Epoch 3/4:  66%|██████▌   | 824/1250 [14:59<07:33,  1.07s/it, loss=0.557]
Epoch 3/4:  66%|██████▌   | 824/1250 [15:00<07:33,  1.07s/it, loss=0.335]
Epoch 3/4:  66%|██████▌   | 825/1250 [15:00<07:35,  1.07s/it, loss=0.335]
Epoch 3/4:  66%|██████▌   | 825/1250 [15:01<07:35,  1.07s/it, loss=0.45] 
Epoch 3/4:  66%|██████▌   | 826/1250 [15:01<07:28,  1.06s/it, loss=0.45]
Epoch 3/4:  66%|██████▌   | 826/1250 [15:02<07:28,  1.06s/it, loss=0.705]
Epoch 3/4:  66%|██████▌   | 827/1250 [15:02<07:19,  1.04s/it, loss=0.705]
Epoch 3/4:  66%|██████▌   | 827/1250 [15:03<07:19,  1.04s/it, loss=0.42] 
Epoch 3/4:  66%|██████▌   | 828/1250 [15:03<07:25,  1.06s/it, loss=0.42]
Epoch 3/4:  66%|██████▌   | 828/1250 [15:04<07:25,  1.06s/it, loss=0.673]
Epoch 3/4:  66%|██████▋   | 829/1250 [15:04<07:35,  1.08s/it, loss=0.673]
Epoch 3/4:  66%|██████▋   | 829/1250 [15:05<07:35,  1.08s/it, loss=0.608]
Epoch 3/4:  66%|██████▋   | 830/1250 [15:05<07:42,  1.10s/it, loss=0.608]
Epoch 3/4:  66%|██████▋   | 830/1250 [15:06<07:42,  1.10s/it, loss=0.425]
Epoch 3/4:  66%|██████▋   | 831/1250 [15:06<07:31,  1.08s/it, loss=0.425]
Epoch 3/4:  66%|██████▋   | 831/1250 [15:07<07:31,  1.08s/it, loss=0.366]
Epoch 3/4:  67%|██████▋   | 832/1250 [15:07<07:20,  1.05s/it, loss=0.366]
Epoch 3/4:  67%|██████▋   | 832/1250 [15:08<07:20,  1.05s/it, loss=0.489]
Epoch 3/4:  67%|██████▋   | 833/1250 [15:08<07:22,  1.06s/it, loss=0.489]
Epoch 3/4:  67%|██████▋   | 833/1250 [15:10<07:22,  1.06s/it, loss=0.482]
Epoch 3/4:  67%|██████▋   | 834/1250 [15:10<07:27,  1.08s/it, loss=0.482]
Epoch 3/4:  67%|██████▋   | 834/1250 [15:11<07:27,  1.08s/it, loss=0.52] 
Epoch 3/4:  67%|██████▋   | 835/1250 [15:11<07:26,  1.08s/it, loss=0.52]
Epoch 3/4:  67%|██████▋   | 835/1250 [15:12<07:26,  1.08s/it, loss=0.614]
Epoch 3/4:  67%|██████▋   | 836/1250 [15:12<07:19,  1.06s/it, loss=0.614]
Epoch 3/4:  67%|██████▋   | 836/1250 [15:13<07:19,  1.06s/it, loss=0.505]
Epoch 3/4:  67%|██████▋   | 837/1250 [15:13<07:24,  1.08s/it, loss=0.505]
Epoch 3/4:  67%|██████▋   | 837/1250 [15:14<07:24,  1.08s/it, loss=0.46] 
Epoch 3/4:  67%|██████▋   | 838/1250 [15:14<07:36,  1.11s/it, loss=0.46]
Epoch 3/4:  67%|██████▋   | 838/1250 [15:15<07:36,  1.11s/it, loss=0.369]
Epoch 3/4:  67%|██████▋   | 839/1250 [15:15<07:38,  1.12s/it, loss=0.369]
Epoch 3/4:  67%|██████▋   | 839/1250 [15:16<07:38,  1.12s/it, loss=0.612]
Epoch 3/4:  67%|██████▋   | 840/1250 [15:16<07:24,  1.09s/it, loss=0.612]
Epoch 3/4:  67%|██████▋   | 840/1250 [15:17<07:24,  1.09s/it, loss=0.682]
Epoch 3/4:  67%|██████▋   | 841/1250 [15:17<07:10,  1.05s/it, loss=0.682]
Epoch 3/4:  67%|██████▋   | 841/1250 [15:18<07:10,  1.05s/it, loss=0.5]  
Epoch 3/4:  67%|██████▋   | 842/1250 [15:18<07:05,  1.04s/it, loss=0.5]
Epoch 3/4:  67%|██████▋   | 842/1250 [15:19<07:05,  1.04s/it, loss=0.568]
Epoch 3/4:  67%|██████▋   | 843/1250 [15:19<07:08,  1.05s/it, loss=0.568]
Epoch 3/4:  67%|██████▋   | 843/1250 [15:20<07:08,  1.05s/it, loss=0.731]
Epoch 3/4:  68%|██████▊   | 844/1250 [15:20<07:11,  1.06s/it, loss=0.731]
Epoch 3/4:  68%|██████▊   | 844/1250 [15:21<07:11,  1.06s/it, loss=0.688]
Epoch 3/4:  68%|██████▊   | 845/1250 [15:21<07:04,  1.05s/it, loss=0.688]
Epoch 3/4:  68%|██████▊   | 845/1250 [15:22<07:04,  1.05s/it, loss=0.562]
Epoch 3/4:  68%|██████▊   | 846/1250 [15:22<06:57,  1.03s/it, loss=0.562]
Epoch 3/4:  68%|██████▊   | 846/1250 [15:23<06:57,  1.03s/it, loss=0.56] 
Epoch 3/4:  68%|██████▊   | 847/1250 [15:23<07:06,  1.06s/it, loss=0.56]
Epoch 3/4:  68%|██████▊   | 847/1250 [15:25<07:06,  1.06s/it, loss=0.559]
Epoch 3/4:  68%|██████▊   | 848/1250 [15:25<07:17,  1.09s/it, loss=0.559]
Epoch 3/4:  68%|██████▊   | 848/1250 [15:26<07:17,  1.09s/it, loss=0.782]
Epoch 3/4:  68%|██████▊   | 849/1250 [15:26<07:18,  1.09s/it, loss=0.782]
Epoch 3/4:  68%|██████▊   | 849/1250 [15:27<07:18,  1.09s/it, loss=0.618]
Epoch 3/4:  68%|██████▊   | 850/1250 [15:27<07:07,  1.07s/it, loss=0.618]
Epoch 3/4:  68%|██████▊   | 850/1250 [15:28<07:07,  1.07s/it, loss=0.466]
Epoch 3/4:  68%|██████▊   | 851/1250 [15:28<06:58,  1.05s/it, loss=0.466]
Epoch 3/4:  68%|██████▊   | 851/1250 [15:29<06:58,  1.05s/it, loss=0.547]
Epoch 3/4:  68%|██████▊   | 852/1250 [15:29<07:00,  1.06s/it, loss=0.547]
Epoch 3/4:  68%|██████▊   | 852/1250 [15:30<07:00,  1.06s/it, loss=0.587]
Epoch 3/4:  68%|██████▊   | 853/1250 [15:30<07:05,  1.07s/it, loss=0.587]
Epoch 3/4:  68%|██████▊   | 853/1250 [15:31<07:05,  1.07s/it, loss=0.527]
Epoch 3/4:  68%|██████▊   | 854/1250 [15:31<06:58,  1.06s/it, loss=0.527]
Epoch 3/4:  68%|██████▊   | 854/1250 [15:32<06:58,  1.06s/it, loss=0.51] 
Epoch 3/4:  68%|██████▊   | 855/1250 [15:32<06:50,  1.04s/it, loss=0.51]
Epoch 3/4:  68%|██████▊   | 855/1250 [15:33<06:50,  1.04s/it, loss=0.493]
Epoch 3/4:  68%|██████▊   | 856/1250 [15:33<06:51,  1.04s/it, loss=0.493]
Epoch 3/4:  68%|██████▊   | 856/1250 [15:34<06:51,  1.04s/it, loss=0.746]
Epoch 3/4:  69%|██████▊   | 857/1250 [15:34<07:02,  1.07s/it, loss=0.746]
Epoch 3/4:  69%|██████▊   | 857/1250 [15:35<07:02,  1.07s/it, loss=0.388]
Epoch 3/4:  69%|██████▊   | 858/1250 [15:35<07:14,  1.11s/it, loss=0.388]
Epoch 3/4:  69%|██████▊   | 858/1250 [15:36<07:14,  1.11s/it, loss=0.517]
Epoch 3/4:  69%|██████▊   | 859/1250 [15:36<07:08,  1.10s/it, loss=0.517]
Epoch 3/4:  69%|██████▊   | 859/1250 [15:37<07:08,  1.10s/it, loss=0.53] 
Epoch 3/4:  69%|██████▉   | 860/1250 [15:37<06:58,  1.07s/it, loss=0.53]
Epoch 3/4:  69%|██████▉   | 860/1250 [15:38<06:58,  1.07s/it, loss=0.308]
Epoch 3/4:  69%|██████▉   | 861/1250 [15:38<06:54,  1.07s/it, loss=0.308]
Epoch 3/4:  69%|██████▉   | 861/1250 [15:40<06:54,  1.07s/it, loss=0.367]
Epoch 3/4:  69%|██████▉   | 862/1250 [15:40<06:54,  1.07s/it, loss=0.367]
Epoch 3/4:  69%|██████▉   | 862/1250 [15:41<06:54,  1.07s/it, loss=0.84] 
Epoch 3/4:  69%|██████▉   | 863/1250 [15:41<06:49,  1.06s/it, loss=0.84]
Epoch 3/4:  69%|██████▉   | 863/1250 [15:42<06:49,  1.06s/it, loss=0.262]
Epoch 3/4:  69%|██████▉   | 864/1250 [15:42<06:39,  1.03s/it, loss=0.262]
Epoch 3/4:  69%|██████▉   | 864/1250 [15:43<06:39,  1.03s/it, loss=0.433]
Epoch 3/4:  69%|██████▉   | 865/1250 [15:43<06:32,  1.02s/it, loss=0.433]
Epoch 3/4:  69%|██████▉   | 865/1250 [15:44<06:32,  1.02s/it, loss=0.695]
Epoch 3/4:  69%|██████▉   | 866/1250 [15:44<06:38,  1.04s/it, loss=0.695]
Epoch 3/4:  69%|██████▉   | 866/1250 [15:45<06:38,  1.04s/it, loss=0.496]
Epoch 3/4:  69%|██████▉   | 867/1250 [15:45<06:51,  1.07s/it, loss=0.496]
Epoch 3/4:  69%|██████▉   | 867/1250 [15:46<06:51,  1.07s/it, loss=0.484]
Epoch 3/4:  69%|██████▉   | 868/1250 [15:46<06:59,  1.10s/it, loss=0.484]
Epoch 3/4:  69%|██████▉   | 868/1250 [15:47<06:59,  1.10s/it, loss=0.478]
Epoch 3/4:  70%|██████▉   | 869/1250 [15:47<06:50,  1.08s/it, loss=0.478]
Epoch 3/4:  70%|██████▉   | 869/1250 [15:48<06:50,  1.08s/it, loss=0.614]
Epoch 3/4:  70%|██████▉   | 870/1250 [15:48<06:44,  1.06s/it, loss=0.614]
Epoch 3/4:  70%|██████▉   | 870/1250 [15:49<06:44,  1.06s/it, loss=0.53] 
Epoch 3/4:  70%|██████▉   | 871/1250 [15:49<06:41,  1.06s/it, loss=0.53]
Epoch 3/4:  70%|██████▉   | 871/1250 [15:50<06:41,  1.06s/it, loss=0.689]
Epoch 3/4:  70%|██████▉   | 872/1250 [15:50<06:44,  1.07s/it, loss=0.689]
Epoch 3/4:  70%|██████▉   | 872/1250 [15:51<06:44,  1.07s/it, loss=0.736]
Epoch 3/4:  70%|██████▉   | 873/1250 [15:51<06:42,  1.07s/it, loss=0.736]
Epoch 3/4:  70%|██████▉   | 873/1250 [15:52<06:42,  1.07s/it, loss=0.499]
Epoch 3/4:  70%|██████▉   | 874/1250 [15:52<06:33,  1.05s/it, loss=0.499]
Epoch 3/4:  70%|██████▉   | 874/1250 [15:53<06:33,  1.05s/it, loss=0.515]
Epoch 3/4:  70%|███████   | 875/1250 [15:53<06:35,  1.06s/it, loss=0.515]
Epoch 3/4:  70%|███████   | 875/1250 [15:54<06:35,  1.06s/it, loss=0.608]
Epoch 3/4:  70%|███████   | 876/1250 [15:54<06:47,  1.09s/it, loss=0.608]
Epoch 3/4:  70%|███████   | 876/1250 [15:56<06:47,  1.09s/it, loss=0.601]
Epoch 3/4:  70%|███████   | 877/1250 [15:56<06:50,  1.10s/it, loss=0.601]
Epoch 3/4:  70%|███████   | 877/1250 [15:57<06:50,  1.10s/it, loss=0.387]
Epoch 3/4:  70%|███████   | 878/1250 [15:57<06:40,  1.08s/it, loss=0.387]
Epoch 3/4:  70%|███████   | 878/1250 [15:58<06:40,  1.08s/it, loss=0.659]
Epoch 3/4:  70%|███████   | 879/1250 [15:58<06:27,  1.04s/it, loss=0.659]
Epoch 3/4:  70%|███████   | 879/1250 [15:59<06:27,  1.04s/it, loss=0.616]
Epoch 3/4:  70%|███████   | 880/1250 [15:59<06:28,  1.05s/it, loss=0.616]
Epoch 3/4:  70%|███████   | 880/1250 [16:00<06:28,  1.05s/it, loss=0.476]
Epoch 3/4:  70%|███████   | 881/1250 [16:00<06:40,  1.09s/it, loss=0.476]
Epoch 3/4:  70%|███████   | 881/1250 [16:01<06:40,  1.09s/it, loss=0.556]
Epoch 3/4:  71%|███████   | 882/1250 [16:01<06:40,  1.09s/it, loss=0.556]
Epoch 3/4:  71%|███████   | 882/1250 [16:02<06:40,  1.09s/it, loss=0.518]
Epoch 3/4:  71%|███████   | 883/1250 [16:02<06:27,  1.06s/it, loss=0.518]
Epoch 3/4:  71%|███████   | 883/1250 [16:03<06:27,  1.06s/it, loss=0.75] 
Epoch 3/4:  71%|███████   | 884/1250 [16:03<06:17,  1.03s/it, loss=0.75]
Epoch 3/4:  71%|███████   | 884/1250 [16:04<06:17,  1.03s/it, loss=0.561]
Epoch 3/4:  71%|███████   | 885/1250 [16:04<06:19,  1.04s/it, loss=0.561]
Epoch 3/4:  71%|███████   | 885/1250 [16:05<06:19,  1.04s/it, loss=0.522]
Epoch 3/4:  71%|███████   | 886/1250 [16:05<06:25,  1.06s/it, loss=0.522]
Epoch 3/4:  71%|███████   | 886/1250 [16:06<06:25,  1.06s/it, loss=0.512]
Epoch 3/4:  71%|███████   | 887/1250 [16:06<06:30,  1.08s/it, loss=0.512]
Epoch 3/4:  71%|███████   | 887/1250 [16:07<06:30,  1.08s/it, loss=0.479]
Epoch 3/4:  71%|███████   | 888/1250 [16:07<06:22,  1.06s/it, loss=0.479]
Epoch 3/4:  71%|███████   | 888/1250 [16:08<06:22,  1.06s/it, loss=0.53] 
Epoch 3/4:  71%|███████   | 889/1250 [16:08<06:14,  1.04s/it, loss=0.53]
Epoch 3/4:  71%|███████   | 889/1250 [16:09<06:14,  1.04s/it, loss=0.694]
Epoch 3/4:  71%|███████   | 890/1250 [16:09<06:16,  1.05s/it, loss=0.694]
Epoch 3/4:  71%|███████   | 890/1250 [16:10<06:16,  1.05s/it, loss=0.5]  
Epoch 3/4:  71%|███████▏  | 891/1250 [16:10<06:22,  1.07s/it, loss=0.5]
Epoch 3/4:  71%|███████▏  | 891/1250 [16:11<06:22,  1.07s/it, loss=0.833]
Epoch 3/4:  71%|███████▏  | 892/1250 [16:11<06:23,  1.07s/it, loss=0.833]
Epoch 3/4:  71%|███████▏  | 892/1250 [16:12<06:23,  1.07s/it, loss=0.453]
Epoch 3/4:  71%|███████▏  | 893/1250 [16:12<06:16,  1.05s/it, loss=0.453]
Epoch 3/4:  71%|███████▏  | 893/1250 [16:13<06:16,  1.05s/it, loss=0.644]
Epoch 3/4:  72%|███████▏  | 894/1250 [16:13<06:10,  1.04s/it, loss=0.644]
Epoch 3/4:  72%|███████▏  | 894/1250 [16:14<06:10,  1.04s/it, loss=0.45] 
Epoch 3/4:  72%|███████▏  | 895/1250 [16:14<06:14,  1.05s/it, loss=0.45]
Epoch 3/4:  72%|███████▏  | 895/1250 [16:16<06:14,  1.05s/it, loss=0.444]
Epoch 3/4:  72%|███████▏  | 896/1250 [16:16<06:24,  1.09s/it, loss=0.444]
Epoch 3/4:  72%|███████▏  | 896/1250 [16:17<06:24,  1.09s/it, loss=0.522]
Epoch 3/4:  72%|███████▏  | 897/1250 [16:17<06:21,  1.08s/it, loss=0.522]
Epoch 3/4:  72%|███████▏  | 897/1250 [16:18<06:21,  1.08s/it, loss=0.608]
Epoch 3/4:  72%|███████▏  | 898/1250 [16:18<06:10,  1.05s/it, loss=0.608]
Epoch 3/4:  72%|███████▏  | 898/1250 [16:19<06:10,  1.05s/it, loss=0.776]
Epoch 3/4:  72%|███████▏  | 899/1250 [16:19<06:06,  1.04s/it, loss=0.776]
Epoch 3/4:  72%|███████▏  | 899/1250 [16:20<06:06,  1.04s/it, loss=0.577]
Epoch 3/4:  72%|███████▏  | 900/1250 [16:20<06:13,  1.07s/it, loss=0.577]
Epoch 3/4:  72%|███████▏  | 900/1250 [16:21<06:13,  1.07s/it, loss=0.626]
Epoch 3/4:  72%|███████▏  | 901/1250 [16:21<06:23,  1.10s/it, loss=0.626]
Epoch 3/4:  72%|███████▏  | 901/1250 [16:22<06:23,  1.10s/it, loss=0.64] 
Epoch 3/4:  72%|███████▏  | 902/1250 [16:22<06:18,  1.09s/it, loss=0.64]
Epoch 3/4:  72%|███████▏  | 902/1250 [16:23<06:18,  1.09s/it, loss=0.476]
Epoch 3/4:  72%|███████▏  | 903/1250 [16:23<06:08,  1.06s/it, loss=0.476]
Epoch 3/4:  72%|███████▏  | 903/1250 [16:24<06:08,  1.06s/it, loss=0.678]
Epoch 3/4:  72%|███████▏  | 904/1250 [16:24<06:04,  1.05s/it, loss=0.678]
Epoch 3/4:  72%|███████▏  | 904/1250 [16:25<06:04,  1.05s/it, loss=0.722]
Epoch 3/4:  72%|███████▏  | 905/1250 [16:25<06:07,  1.06s/it, loss=0.722]
Epoch 3/4:  72%|███████▏  | 905/1250 [16:26<06:07,  1.06s/it, loss=0.579]
Epoch 3/4:  72%|███████▏  | 906/1250 [16:26<06:06,  1.07s/it, loss=0.579]
Epoch 3/4:  72%|███████▏  | 906/1250 [16:27<06:06,  1.07s/it, loss=0.434]
Epoch 3/4:  73%|███████▎  | 907/1250 [16:27<05:59,  1.05s/it, loss=0.434]
Epoch 3/4:  73%|███████▎  | 907/1250 [16:28<05:59,  1.05s/it, loss=0.504]
Epoch 3/4:  73%|███████▎  | 908/1250 [16:28<05:56,  1.04s/it, loss=0.504]
Epoch 3/4:  73%|███████▎  | 908/1250 [16:29<05:56,  1.04s/it, loss=0.72] 
Epoch 3/4:  73%|███████▎  | 909/1250 [16:29<06:02,  1.06s/it, loss=0.72]
Epoch 3/4:  73%|███████▎  | 909/1250 [16:30<06:02,  1.06s/it, loss=0.536]
Epoch 3/4:  73%|███████▎  | 910/1250 [16:30<06:05,  1.08s/it, loss=0.536]
Epoch 3/4:  73%|███████▎  | 910/1250 [16:32<06:05,  1.08s/it, loss=0.71] 
Epoch 3/4:  73%|███████▎  | 911/1250 [16:32<06:01,  1.07s/it, loss=0.71]
Epoch 3/4:  73%|███████▎  | 911/1250 [16:33<06:01,  1.07s/it, loss=0.553]
Epoch 3/4:  73%|███████▎  | 912/1250 [16:33<05:52,  1.04s/it, loss=0.553]
Epoch 3/4:  73%|███████▎  | 912/1250 [16:34<05:52,  1.04s/it, loss=0.552]
Epoch 3/4:  73%|███████▎  | 913/1250 [16:34<05:49,  1.04s/it, loss=0.552]
Epoch 3/4:  73%|███████▎  | 913/1250 [16:35<05:49,  1.04s/it, loss=0.624]
Epoch 3/4:  73%|███████▎  | 914/1250 [16:35<06:01,  1.08s/it, loss=0.624]
Epoch 3/4:  73%|███████▎  | 914/1250 [16:36<06:01,  1.08s/it, loss=0.501]
Epoch 3/4:  73%|███████▎  | 915/1250 [16:36<06:11,  1.11s/it, loss=0.501]
Epoch 3/4:  73%|███████▎  | 915/1250 [16:37<06:11,  1.11s/it, loss=1]    
Epoch 3/4:  73%|███████▎  | 916/1250 [16:37<06:05,  1.10s/it, loss=1]
Epoch 3/4:  73%|███████▎  | 916/1250 [16:38<06:05,  1.10s/it, loss=0.767]
Epoch 3/4:  73%|███████▎  | 917/1250 [16:38<05:53,  1.06s/it, loss=0.767]
Epoch 3/4:  73%|███████▎  | 917/1250 [16:39<05:53,  1.06s/it, loss=0.804]
Epoch 3/4:  73%|███████▎  | 918/1250 [16:39<05:48,  1.05s/it, loss=0.804]
Epoch 3/4:  73%|███████▎  | 918/1250 [16:40<05:48,  1.05s/it, loss=0.611]
Epoch 3/4:  74%|███████▎  | 919/1250 [16:40<05:55,  1.07s/it, loss=0.611]
Epoch 3/4:  74%|███████▎  | 919/1250 [16:41<05:55,  1.07s/it, loss=0.404]
Epoch 3/4:  74%|███████▎  | 920/1250 [16:41<05:56,  1.08s/it, loss=0.404]
Epoch 3/4:  74%|███████▎  | 920/1250 [16:42<05:56,  1.08s/it, loss=0.582]
Epoch 3/4:  74%|███████▎  | 921/1250 [16:42<05:48,  1.06s/it, loss=0.582]
Epoch 3/4:  74%|███████▎  | 921/1250 [16:43<05:48,  1.06s/it, loss=0.482]
Epoch 3/4:  74%|███████▍  | 922/1250 [16:43<05:43,  1.05s/it, loss=0.482]
Epoch 3/4:  74%|███████▍  | 922/1250 [16:44<05:43,  1.05s/it, loss=0.916]
Epoch 3/4:  74%|███████▍  | 923/1250 [16:44<05:49,  1.07s/it, loss=0.916]
Epoch 3/4:  74%|███████▍  | 923/1250 [16:46<05:49,  1.07s/it, loss=0.853]
Epoch 3/4:  74%|███████▍  | 924/1250 [16:46<05:58,  1.10s/it, loss=0.853]
Epoch 3/4:  74%|███████▍  | 924/1250 [16:47<05:58,  1.10s/it, loss=0.583]
Epoch 3/4:  74%|███████▍  | 925/1250 [16:47<05:54,  1.09s/it, loss=0.583]
Epoch 3/4:  74%|███████▍  | 925/1250 [16:48<05:54,  1.09s/it, loss=0.722]
Epoch 3/4:  74%|███████▍  | 926/1250 [16:48<05:45,  1.07s/it, loss=0.722]
Epoch 3/4:  74%|███████▍  | 926/1250 [16:49<05:45,  1.07s/it, loss=0.478]
Epoch 3/4:  74%|███████▍  | 927/1250 [16:49<05:40,  1.05s/it, loss=0.478]
Epoch 3/4:  74%|███████▍  | 927/1250 [16:50<05:40,  1.05s/it, loss=0.464]
Epoch 3/4:  74%|███████▍  | 928/1250 [16:50<05:40,  1.06s/it, loss=0.464]
Epoch 3/4:  74%|███████▍  | 928/1250 [16:51<05:40,  1.06s/it, loss=0.673]
Epoch 3/4:  74%|███████▍  | 929/1250 [16:51<05:41,  1.06s/it, loss=0.673]
Epoch 3/4:  74%|███████▍  | 929/1250 [16:52<05:41,  1.06s/it, loss=0.512]
Epoch 3/4:  74%|███████▍  | 930/1250 [16:52<05:39,  1.06s/it, loss=0.512]
Epoch 3/4:  74%|███████▍  | 930/1250 [16:53<05:39,  1.06s/it, loss=0.664]
Epoch 3/4:  74%|███████▍  | 931/1250 [16:53<05:33,  1.05s/it, loss=0.664]
Epoch 3/4:  74%|███████▍  | 931/1250 [16:54<05:33,  1.05s/it, loss=0.576]
Epoch 3/4:  75%|███████▍  | 932/1250 [16:54<05:32,  1.05s/it, loss=0.576]
Epoch 3/4:  75%|███████▍  | 932/1250 [16:55<05:32,  1.05s/it, loss=0.55] 
Epoch 3/4:  75%|███████▍  | 933/1250 [16:55<05:38,  1.07s/it, loss=0.55]
Epoch 3/4:  75%|███████▍  | 933/1250 [16:56<05:38,  1.07s/it, loss=0.485]
Epoch 3/4:  75%|███████▍  | 934/1250 [16:56<05:39,  1.08s/it, loss=0.485]
Epoch 3/4:  75%|███████▍  | 934/1250 [16:57<05:39,  1.08s/it, loss=0.764]
Epoch 3/4:  75%|███████▍  | 935/1250 [16:57<05:37,  1.07s/it, loss=0.764]
Epoch 3/4:  75%|███████▍  | 935/1250 [16:58<05:37,  1.07s/it, loss=0.704]
Epoch 3/4:  75%|███████▍  | 936/1250 [16:58<05:30,  1.05s/it, loss=0.704]
Epoch 3/4:  75%|███████▍  | 936/1250 [16:59<05:30,  1.05s/it, loss=0.533]
Epoch 3/4:  75%|███████▍  | 937/1250 [16:59<05:28,  1.05s/it, loss=0.533]
Epoch 3/4:  75%|███████▍  | 937/1250 [17:00<05:28,  1.05s/it, loss=0.625]
Epoch 3/4:  75%|███████▌  | 938/1250 [17:00<05:33,  1.07s/it, loss=0.625]
Epoch 3/4:  75%|███████▌  | 938/1250 [17:01<05:33,  1.07s/it, loss=0.503]
Epoch 3/4:  75%|███████▌  | 939/1250 [17:01<05:33,  1.07s/it, loss=0.503]
Epoch 3/4:  75%|███████▌  | 939/1250 [17:02<05:33,  1.07s/it, loss=0.679]
Epoch 3/4:  75%|███████▌  | 940/1250 [17:02<05:28,  1.06s/it, loss=0.679]
Epoch 3/4:  75%|███████▌  | 940/1250 [17:03<05:28,  1.06s/it, loss=0.45] 
Epoch 3/4:  75%|███████▌  | 941/1250 [17:03<05:23,  1.05s/it, loss=0.45]
Epoch 3/4:  75%|███████▌  | 941/1250 [17:05<05:23,  1.05s/it, loss=0.606]
Epoch 3/4:  75%|███████▌  | 942/1250 [17:05<05:27,  1.06s/it, loss=0.606]
Epoch 3/4:  75%|███████▌  | 942/1250 [17:06<05:27,  1.06s/it, loss=0.749]
Epoch 3/4:  75%|███████▌  | 943/1250 [17:06<05:37,  1.10s/it, loss=0.749]
Epoch 3/4:  75%|███████▌  | 943/1250 [17:07<05:37,  1.10s/it, loss=0.543]
Epoch 3/4:  76%|███████▌  | 944/1250 [17:07<05:37,  1.10s/it, loss=0.543]
Epoch 3/4:  76%|███████▌  | 944/1250 [17:08<05:37,  1.10s/it, loss=0.569]
Epoch 3/4:  76%|███████▌  | 945/1250 [17:08<05:26,  1.07s/it, loss=0.569]
Epoch 3/4:  76%|███████▌  | 945/1250 [17:09<05:26,  1.07s/it, loss=0.522]
Epoch 3/4:  76%|███████▌  | 946/1250 [17:09<05:17,  1.05s/it, loss=0.522]
Epoch 3/4:  76%|███████▌  | 946/1250 [17:10<05:17,  1.05s/it, loss=0.668]
Epoch 3/4:  76%|███████▌  | 947/1250 [17:10<05:18,  1.05s/it, loss=0.668]
Epoch 3/4:  76%|███████▌  | 947/1250 [17:11<05:18,  1.05s/it, loss=0.669]
Epoch 3/4:  76%|███████▌  | 948/1250 [17:11<05:26,  1.08s/it, loss=0.669]
Epoch 3/4:  76%|███████▌  | 948/1250 [17:12<05:26,  1.08s/it, loss=0.553]
Epoch 3/4:  76%|███████▌  | 949/1250 [17:12<05:23,  1.07s/it, loss=0.553]
Epoch 3/4:  76%|███████▌  | 949/1250 [17:13<05:23,  1.07s/it, loss=0.833]
Epoch 3/4:  76%|███████▌  | 950/1250 [17:13<05:16,  1.06s/it, loss=0.833]
Epoch 3/4:  76%|███████▌  | 950/1250 [17:14<05:16,  1.06s/it, loss=0.719]
Epoch 3/4:  76%|███████▌  | 951/1250 [17:14<05:17,  1.06s/it, loss=0.719]
Epoch 3/4:  76%|███████▌  | 951/1250 [17:15<05:17,  1.06s/it, loss=0.357]
Epoch 3/4:  76%|███████▌  | 952/1250 [17:15<05:25,  1.09s/it, loss=0.357]
Epoch 3/4:  76%|███████▌  | 952/1250 [17:16<05:25,  1.09s/it, loss=0.496]
Epoch 3/4:  76%|███████▌  | 953/1250 [17:16<05:25,  1.10s/it, loss=0.496]
Epoch 3/4:  76%|███████▌  | 953/1250 [17:17<05:25,  1.10s/it, loss=0.409]
Epoch 3/4:  76%|███████▋  | 954/1250 [17:17<05:15,  1.07s/it, loss=0.409]
Epoch 3/4:  76%|███████▋  | 954/1250 [17:18<05:15,  1.07s/it, loss=0.818]
Epoch 3/4:  76%|███████▋  | 955/1250 [17:18<05:06,  1.04s/it, loss=0.818]
Epoch 3/4:  76%|███████▋  | 955/1250 [17:19<05:06,  1.04s/it, loss=0.399]
Epoch 3/4:  76%|███████▋  | 956/1250 [17:19<05:06,  1.04s/it, loss=0.399]
Epoch 3/4:  76%|███████▋  | 956/1250 [17:21<05:06,  1.04s/it, loss=0.489]
Epoch 3/4:  77%|███████▋  | 957/1250 [17:21<05:10,  1.06s/it, loss=0.489]
Epoch 3/4:  77%|███████▋  | 957/1250 [17:22<05:10,  1.06s/it, loss=0.374]
Epoch 3/4:  77%|███████▋  | 958/1250 [17:22<05:11,  1.07s/it, loss=0.374]
Epoch 3/4:  77%|███████▋  | 958/1250 [17:23<05:11,  1.07s/it, loss=0.599]
Epoch 3/4:  77%|███████▋  | 959/1250 [17:23<05:08,  1.06s/it, loss=0.599]
Epoch 3/4:  77%|███████▋  | 959/1250 [17:24<05:08,  1.06s/it, loss=0.608]
Epoch 3/4:  77%|███████▋  | 960/1250 [17:24<05:04,  1.05s/it, loss=0.608]
Epoch 3/4:  77%|███████▋  | 960/1250 [17:25<05:04,  1.05s/it, loss=0.842]
Epoch 3/4:  77%|███████▋  | 961/1250 [17:25<05:04,  1.05s/it, loss=0.842]
Epoch 3/4:  77%|███████▋  | 961/1250 [17:26<05:04,  1.05s/it, loss=0.521]
Epoch 3/4:  77%|███████▋  | 962/1250 [17:26<05:06,  1.07s/it, loss=0.521]
Epoch 3/4:  77%|███████▋  | 962/1250 [17:27<05:06,  1.07s/it, loss=0.527]
Epoch 3/4:  77%|███████▋  | 963/1250 [17:27<05:03,  1.06s/it, loss=0.527]
Epoch 3/4:  77%|███████▋  | 963/1250 [17:28<05:03,  1.06s/it, loss=0.485]
Epoch 3/4:  77%|███████▋  | 964/1250 [17:28<04:56,  1.04s/it, loss=0.485]
Epoch 3/4:  77%|███████▋  | 964/1250 [17:29<04:56,  1.04s/it, loss=0.507]
Epoch 3/4:  77%|███████▋  | 965/1250 [17:29<04:56,  1.04s/it, loss=0.507]
Epoch 3/4:  77%|███████▋  | 965/1250 [17:30<04:56,  1.04s/it, loss=0.661]
Epoch 3/4:  77%|███████▋  | 966/1250 [17:30<05:07,  1.08s/it, loss=0.661]
Epoch 3/4:  77%|███████▋  | 966/1250 [17:31<05:07,  1.08s/it, loss=0.564]
Epoch 3/4:  77%|███████▋  | 967/1250 [17:31<05:09,  1.09s/it, loss=0.564]
Epoch 3/4:  77%|███████▋  | 967/1250 [17:32<05:09,  1.09s/it, loss=0.721]
Epoch 3/4:  77%|███████▋  | 968/1250 [17:32<05:01,  1.07s/it, loss=0.721]
Epoch 3/4:  77%|███████▋  | 968/1250 [17:33<05:01,  1.07s/it, loss=0.74] 
Epoch 3/4:  78%|███████▊  | 969/1250 [17:33<04:52,  1.04s/it, loss=0.74]
Epoch 3/4:  78%|███████▊  | 969/1250 [17:34<04:52,  1.04s/it, loss=0.367]
Epoch 3/4:  78%|███████▊  | 970/1250 [17:34<04:49,  1.03s/it, loss=0.367]
Epoch 3/4:  78%|███████▊  | 970/1250 [17:35<04:49,  1.03s/it, loss=0.749]
Epoch 3/4:  78%|███████▊  | 971/1250 [17:35<05:00,  1.08s/it, loss=0.749]
Epoch 3/4:  78%|███████▊  | 971/1250 [17:37<05:00,  1.08s/it, loss=0.532]
Epoch 3/4:  78%|███████▊  | 972/1250 [17:37<05:04,  1.10s/it, loss=0.532]
Epoch 3/4:  78%|███████▊  | 972/1250 [17:38<05:04,  1.10s/it, loss=0.605]
Epoch 3/4:  78%|███████▊  | 973/1250 [17:38<04:57,  1.07s/it, loss=0.605]
Epoch 3/4:  78%|███████▊  | 973/1250 [17:39<04:57,  1.07s/it, loss=0.584]
Epoch 3/4:  78%|███████▊  | 974/1250 [17:39<04:48,  1.05s/it, loss=0.584]
Epoch 3/4:  78%|███████▊  | 974/1250 [17:40<04:48,  1.05s/it, loss=0.471]
Epoch 3/4:  78%|███████▊  | 975/1250 [17:40<04:48,  1.05s/it, loss=0.471]
Epoch 3/4:  78%|███████▊  | 975/1250 [17:41<04:48,  1.05s/it, loss=0.666]
Epoch 3/4:  78%|███████▊  | 976/1250 [17:41<04:52,  1.07s/it, loss=0.666]
Epoch 3/4:  78%|███████▊  | 976/1250 [17:42<04:52,  1.07s/it, loss=1.03] 
Epoch 3/4:  78%|███████▊  | 977/1250 [17:42<04:53,  1.07s/it, loss=1.03]
Epoch 3/4:  78%|███████▊  | 977/1250 [17:43<04:53,  1.07s/it, loss=0.528]
Epoch 3/4:  78%|███████▊  | 978/1250 [17:43<04:49,  1.06s/it, loss=0.528]
Epoch 3/4:  78%|███████▊  | 978/1250 [17:44<04:49,  1.06s/it, loss=0.614]
Epoch 3/4:  78%|███████▊  | 979/1250 [17:44<04:44,  1.05s/it, loss=0.614]
Epoch 3/4:  78%|███████▊  | 979/1250 [17:45<04:44,  1.05s/it, loss=0.58] 
Epoch 3/4:  78%|███████▊  | 980/1250 [17:45<04:48,  1.07s/it, loss=0.58]
Epoch 3/4:  78%|███████▊  | 980/1250 [17:46<04:48,  1.07s/it, loss=0.649]
Epoch 3/4:  78%|███████▊  | 981/1250 [17:46<04:54,  1.10s/it, loss=0.649]
Epoch 3/4:  78%|███████▊  | 981/1250 [17:47<04:54,  1.10s/it, loss=0.713]
Epoch 3/4:  79%|███████▊  | 982/1250 [17:47<04:54,  1.10s/it, loss=0.713]
Epoch 3/4:  79%|███████▊  | 982/1250 [17:48<04:54,  1.10s/it, loss=0.609]
Epoch 3/4:  79%|███████▊  | 983/1250 [17:48<04:45,  1.07s/it, loss=0.609]
Epoch 3/4:  79%|███████▊  | 983/1250 [17:49<04:45,  1.07s/it, loss=0.628]
Epoch 3/4:  79%|███████▊  | 984/1250 [17:49<04:39,  1.05s/it, loss=0.628]
Epoch 3/4:  79%|███████▊  | 984/1250 [17:50<04:39,  1.05s/it, loss=0.863]
Epoch 3/4:  79%|███████▉  | 985/1250 [17:50<04:39,  1.05s/it, loss=0.863]
Epoch 3/4:  79%|███████▉  | 985/1250 [17:51<04:39,  1.05s/it, loss=0.506]
Epoch 3/4:  79%|███████▉  | 986/1250 [17:51<04:41,  1.07s/it, loss=0.506]
Epoch 3/4:  79%|███████▉  | 986/1250 [17:53<04:41,  1.07s/it, loss=0.473]
Epoch 3/4:  79%|███████▉  | 987/1250 [17:53<04:41,  1.07s/it, loss=0.473]
Epoch 3/4:  79%|███████▉  | 987/1250 [17:54<04:41,  1.07s/it, loss=0.535]
Epoch 3/4:  79%|███████▉  | 988/1250 [17:54<04:38,  1.06s/it, loss=0.535]
Epoch 3/4:  79%|███████▉  | 988/1250 [17:55<04:38,  1.06s/it, loss=0.49] 
Epoch 3/4:  79%|███████▉  | 989/1250 [17:55<04:36,  1.06s/it, loss=0.49]
Epoch 3/4:  79%|███████▉  | 989/1250 [17:56<04:36,  1.06s/it, loss=0.443]
Epoch 3/4:  79%|███████▉  | 990/1250 [17:56<04:37,  1.07s/it, loss=0.443]
Epoch 3/4:  79%|███████▉  | 990/1250 [17:57<04:37,  1.07s/it, loss=0.548]
Epoch 3/4:  79%|███████▉  | 991/1250 [17:57<04:36,  1.07s/it, loss=0.548]
Epoch 3/4:  79%|███████▉  | 991/1250 [17:58<04:36,  1.07s/it, loss=0.425]
Epoch 3/4:  79%|███████▉  | 992/1250 [17:58<04:34,  1.06s/it, loss=0.425]
Epoch 3/4:  79%|███████▉  | 992/1250 [17:59<04:34,  1.06s/it, loss=0.618]
Epoch 3/4:  79%|███████▉  | 993/1250 [17:59<04:29,  1.05s/it, loss=0.618]
Epoch 3/4:  79%|███████▉  | 993/1250 [18:00<04:29,  1.05s/it, loss=0.616]
Epoch 3/4:  80%|███████▉  | 994/1250 [18:00<04:28,  1.05s/it, loss=0.616]
Epoch 3/4:  80%|███████▉  | 994/1250 [18:01<04:28,  1.05s/it, loss=0.693]
Epoch 3/4:  80%|███████▉  | 995/1250 [18:01<04:30,  1.06s/it, loss=0.693]
Epoch 3/4:  80%|███████▉  | 995/1250 [18:02<04:30,  1.06s/it, loss=0.522]
Epoch 3/4:  80%|███████▉  | 996/1250 [18:02<04:28,  1.06s/it, loss=0.522]
Epoch 3/4:  80%|███████▉  | 996/1250 [18:03<04:28,  1.06s/it, loss=0.623]
Epoch 3/4:  80%|███████▉  | 997/1250 [18:03<04:22,  1.04s/it, loss=0.623]
Epoch 3/4:  80%|███████▉  | 997/1250 [18:04<04:22,  1.04s/it, loss=0.691]
Epoch 3/4:  80%|███████▉  | 998/1250 [18:04<04:18,  1.03s/it, loss=0.691]
Epoch 3/4:  80%|███████▉  | 998/1250 [18:05<04:18,  1.03s/it, loss=0.618]
Epoch 3/4:  80%|███████▉  | 999/1250 [18:05<04:25,  1.06s/it, loss=0.618]
Epoch 3/4:  80%|███████▉  | 999/1250 [18:06<04:25,  1.06s/it, loss=0.989]
Epoch 3/4:  80%|████████  | 1000/1250 [18:06<04:31,  1.09s/it, loss=0.989]
Epoch 3/4:  80%|████████  | 1000/1250 [18:07<04:31,  1.09s/it, loss=0.674]
Epoch 3/4:  80%|████████  | 1001/1250 [18:07<04:27,  1.07s/it, loss=0.674]
Epoch 3/4:  80%|████████  | 1001/1250 [18:08<04:27,  1.07s/it, loss=0.479]
Epoch 3/4:  80%|████████  | 1002/1250 [18:08<04:19,  1.05s/it, loss=0.479]
Epoch 3/4:  80%|████████  | 1002/1250 [18:09<04:19,  1.05s/it, loss=0.383]
Epoch 3/4:  80%|████████  | 1003/1250 [18:09<04:13,  1.03s/it, loss=0.383]
Epoch 3/4:  80%|████████  | 1003/1250 [18:10<04:13,  1.03s/it, loss=0.541]
Epoch 3/4:  80%|████████  | 1004/1250 [18:10<04:16,  1.04s/it, loss=0.541]
Epoch 3/4:  80%|████████  | 1004/1250 [18:12<04:16,  1.04s/it, loss=0.52] 
Epoch 3/4:  80%|████████  | 1005/1250 [18:12<04:29,  1.10s/it, loss=0.52]
Epoch 3/4:  80%|████████  | 1005/1250 [18:13<04:29,  1.10s/it, loss=0.646]
Epoch 3/4:  80%|████████  | 1006/1250 [18:13<04:27,  1.10s/it, loss=0.646]
Epoch 3/4:  80%|████████  | 1006/1250 [18:14<04:27,  1.10s/it, loss=0.981]
Epoch 3/4:  81%|████████  | 1007/1250 [18:14<04:22,  1.08s/it, loss=0.981]
Epoch 3/4:  81%|████████  | 1007/1250 [18:15<04:22,  1.08s/it, loss=0.391]
Epoch 3/4:  81%|████████  | 1008/1250 [18:15<04:22,  1.09s/it, loss=0.391]
Epoch 3/4:  81%|████████  | 1008/1250 [18:16<04:22,  1.09s/it, loss=0.606]
Epoch 3/4:  81%|████████  | 1009/1250 [18:16<04:28,  1.12s/it, loss=0.606]
Epoch 3/4:  81%|████████  | 1009/1250 [18:17<04:28,  1.12s/it, loss=0.488]
Epoch 3/4:  81%|████████  | 1010/1250 [18:17<04:28,  1.12s/it, loss=0.488]
Epoch 3/4:  81%|████████  | 1010/1250 [18:18<04:28,  1.12s/it, loss=0.676]
Epoch 3/4:  81%|████████  | 1011/1250 [18:18<04:19,  1.08s/it, loss=0.676]
Epoch 3/4:  81%|████████  | 1011/1250 [18:19<04:19,  1.08s/it, loss=0.581]
Epoch 3/4:  81%|████████  | 1012/1250 [18:19<04:08,  1.04s/it, loss=0.581]
Epoch 3/4:  81%|████████  | 1012/1250 [18:20<04:08,  1.04s/it, loss=0.588]
Epoch 3/4:  81%|████████  | 1013/1250 [18:20<04:04,  1.03s/it, loss=0.588]
Epoch 3/4:  81%|████████  | 1013/1250 [18:21<04:04,  1.03s/it, loss=0.65] 
Epoch 3/4:  81%|████████  | 1014/1250 [18:21<04:08,  1.05s/it, loss=0.65]
Epoch 3/4:  81%|████████  | 1014/1250 [18:22<04:08,  1.05s/it, loss=0.628]
Epoch 3/4:  81%|████████  | 1015/1250 [18:22<04:09,  1.06s/it, loss=0.628]
Epoch 3/4:  81%|████████  | 1015/1250 [18:23<04:09,  1.06s/it, loss=0.639]
Epoch 3/4:  81%|████████▏ | 1016/1250 [18:23<04:04,  1.04s/it, loss=0.639]
Epoch 3/4:  81%|████████▏ | 1016/1250 [18:24<04:04,  1.04s/it, loss=0.364]
Epoch 3/4:  81%|████████▏ | 1017/1250 [18:24<03:59,  1.03s/it, loss=0.364]
Epoch 3/4:  81%|████████▏ | 1017/1250 [18:25<03:59,  1.03s/it, loss=0.754]
Epoch 3/4:  81%|████████▏ | 1018/1250 [18:25<04:01,  1.04s/it, loss=0.754]
Epoch 3/4:  81%|████████▏ | 1018/1250 [18:27<04:01,  1.04s/it, loss=0.675]
Epoch 3/4:  82%|████████▏ | 1019/1250 [18:27<04:07,  1.07s/it, loss=0.675]
Epoch 3/4:  82%|████████▏ | 1019/1250 [18:28<04:07,  1.07s/it, loss=0.683]
Epoch 3/4:  82%|████████▏ | 1020/1250 [18:28<04:05,  1.07s/it, loss=0.683]
Epoch 3/4:  82%|████████▏ | 1020/1250 [18:29<04:05,  1.07s/it, loss=0.486]
Epoch 3/4:  82%|████████▏ | 1021/1250 [18:29<03:59,  1.05s/it, loss=0.486]
Epoch 3/4:  82%|████████▏ | 1021/1250 [18:30<03:59,  1.05s/it, loss=0.43] 
Epoch 3/4:  82%|████████▏ | 1022/1250 [18:30<03:54,  1.03s/it, loss=0.43]
Epoch 3/4:  82%|████████▏ | 1022/1250 [18:31<03:54,  1.03s/it, loss=0.681]
Epoch 3/4:  82%|████████▏ | 1023/1250 [18:31<03:57,  1.05s/it, loss=0.681]
Epoch 3/4:  82%|████████▏ | 1023/1250 [18:32<03:57,  1.05s/it, loss=0.461]
Epoch 3/4:  82%|████████▏ | 1024/1250 [18:32<04:00,  1.06s/it, loss=0.461]
Epoch 3/4:  82%|████████▏ | 1024/1250 [18:33<04:00,  1.06s/it, loss=0.559]
Epoch 3/4:  82%|████████▏ | 1025/1250 [18:33<03:58,  1.06s/it, loss=0.559]
Epoch 3/4:  82%|████████▏ | 1025/1250 [18:34<03:58,  1.06s/it, loss=0.497]
Epoch 3/4:  82%|████████▏ | 1026/1250 [18:34<03:53,  1.04s/it, loss=0.497]
Epoch 3/4:  82%|████████▏ | 1026/1250 [18:35<03:53,  1.04s/it, loss=0.43] 
Epoch 3/4:  82%|████████▏ | 1027/1250 [18:35<03:51,  1.04s/it, loss=0.43]
Epoch 3/4:  82%|████████▏ | 1027/1250 [18:36<03:51,  1.04s/it, loss=0.446]
Epoch 3/4:  82%|████████▏ | 1028/1250 [18:36<03:56,  1.06s/it, loss=0.446]
Epoch 3/4:  82%|████████▏ | 1028/1250 [18:37<03:56,  1.06s/it, loss=0.603]
Epoch 3/4:  82%|████████▏ | 1029/1250 [18:37<03:59,  1.08s/it, loss=0.603]
Epoch 3/4:  82%|████████▏ | 1029/1250 [18:38<03:59,  1.08s/it, loss=0.685]
Epoch 3/4:  82%|████████▏ | 1030/1250 [18:38<03:58,  1.08s/it, loss=0.685]
Epoch 3/4:  82%|████████▏ | 1030/1250 [18:39<03:58,  1.08s/it, loss=0.476]
Epoch 3/4:  82%|████████▏ | 1031/1250 [18:39<03:54,  1.07s/it, loss=0.476]
Epoch 3/4:  82%|████████▏ | 1031/1250 [18:40<03:54,  1.07s/it, loss=0.527]
Epoch 3/4:  83%|████████▎ | 1032/1250 [18:40<03:50,  1.06s/it, loss=0.527]
Epoch 3/4:  83%|████████▎ | 1032/1250 [18:41<03:50,  1.06s/it, loss=0.523]
Epoch 3/4:  83%|████████▎ | 1033/1250 [18:41<03:48,  1.06s/it, loss=0.523]
Epoch 3/4:  83%|████████▎ | 1033/1250 [18:42<03:48,  1.06s/it, loss=0.435]
Epoch 3/4:  83%|████████▎ | 1034/1250 [18:42<03:47,  1.05s/it, loss=0.435]
Epoch 3/4:  83%|████████▎ | 1034/1250 [18:43<03:47,  1.05s/it, loss=0.564]
Epoch 3/4:  83%|████████▎ | 1035/1250 [18:43<03:44,  1.04s/it, loss=0.564]
Epoch 3/4:  83%|████████▎ | 1035/1250 [18:44<03:44,  1.04s/it, loss=0.639]
Epoch 3/4:  83%|████████▎ | 1036/1250 [18:44<03:41,  1.03s/it, loss=0.639]
Epoch 3/4:  83%|████████▎ | 1036/1250 [18:45<03:41,  1.03s/it, loss=0.545]
Epoch 3/4:  83%|████████▎ | 1037/1250 [18:45<03:45,  1.06s/it, loss=0.545]
Epoch 3/4:  83%|████████▎ | 1037/1250 [18:47<03:45,  1.06s/it, loss=0.539]
Epoch 3/4:  83%|████████▎ | 1038/1250 [18:47<03:48,  1.08s/it, loss=0.539]
Epoch 3/4:  83%|████████▎ | 1038/1250 [18:48<03:48,  1.08s/it, loss=0.469]
Epoch 3/4:  83%|████████▎ | 1039/1250 [18:48<03:44,  1.06s/it, loss=0.469]
Epoch 3/4:  83%|████████▎ | 1039/1250 [18:49<03:44,  1.06s/it, loss=0.429]
Epoch 3/4:  83%|████████▎ | 1040/1250 [18:49<03:37,  1.04s/it, loss=0.429]
Epoch 3/4:  83%|████████▎ | 1040/1250 [18:50<03:37,  1.04s/it, loss=0.515]
Epoch 3/4:  83%|████████▎ | 1041/1250 [18:50<03:33,  1.02s/it, loss=0.515]
Epoch 3/4:  83%|████████▎ | 1041/1250 [18:51<03:33,  1.02s/it, loss=1.04] 
Epoch 3/4:  83%|████████▎ | 1042/1250 [18:51<03:36,  1.04s/it, loss=1.04]
Epoch 3/4:  83%|████████▎ | 1042/1250 [18:52<03:36,  1.04s/it, loss=0.523]
Epoch 3/4:  83%|████████▎ | 1043/1250 [18:52<03:42,  1.08s/it, loss=0.523]
Epoch 3/4:  83%|████████▎ | 1043/1250 [18:53<03:42,  1.08s/it, loss=0.551]
Epoch 3/4:  84%|████████▎ | 1044/1250 [18:53<03:41,  1.07s/it, loss=0.551]
Epoch 3/4:  84%|████████▎ | 1044/1250 [18:54<03:41,  1.07s/it, loss=0.805]
Epoch 3/4:  84%|████████▎ | 1045/1250 [18:54<03:34,  1.05s/it, loss=0.805]
Epoch 3/4:  84%|████████▎ | 1045/1250 [18:55<03:34,  1.05s/it, loss=0.576]
Epoch 3/4:  84%|████████▎ | 1046/1250 [18:55<03:30,  1.03s/it, loss=0.576]
Epoch 3/4:  84%|████████▎ | 1046/1250 [18:56<03:30,  1.03s/it, loss=0.606]
Epoch 3/4:  84%|████████▍ | 1047/1250 [18:56<03:38,  1.08s/it, loss=0.606]
Epoch 3/4:  84%|████████▍ | 1047/1250 [18:57<03:38,  1.08s/it, loss=0.637]
Epoch 3/4:  84%|████████▍ | 1048/1250 [18:57<03:39,  1.09s/it, loss=0.637]
Epoch 3/4:  84%|████████▍ | 1048/1250 [18:58<03:39,  1.09s/it, loss=0.431]
Epoch 3/4:  84%|████████▍ | 1049/1250 [18:58<03:35,  1.07s/it, loss=0.431]
Epoch 3/4:  84%|████████▍ | 1049/1250 [18:59<03:35,  1.07s/it, loss=0.403]
Epoch 3/4:  84%|████████▍ | 1050/1250 [18:59<03:29,  1.05s/it, loss=0.403]
Epoch 3/4:  84%|████████▍ | 1050/1250 [19:00<03:29,  1.05s/it, loss=0.649]
Epoch 3/4:  84%|████████▍ | 1051/1250 [19:00<03:26,  1.04s/it, loss=0.649]
Epoch 3/4:  84%|████████▍ | 1051/1250 [19:01<03:26,  1.04s/it, loss=0.437]
Epoch 3/4:  84%|████████▍ | 1052/1250 [19:01<03:33,  1.08s/it, loss=0.437]
Epoch 3/4:  84%|████████▍ | 1052/1250 [19:03<03:33,  1.08s/it, loss=0.393]
Epoch 3/4:  84%|████████▍ | 1053/1250 [19:03<03:37,  1.10s/it, loss=0.393]
Epoch 3/4:  84%|████████▍ | 1053/1250 [19:04<03:37,  1.10s/it, loss=0.457]
Epoch 3/4:  84%|████████▍ | 1054/1250 [19:04<03:30,  1.07s/it, loss=0.457]
Epoch 3/4:  84%|████████▍ | 1054/1250 [19:05<03:30,  1.07s/it, loss=0.613]
Epoch 3/4:  84%|████████▍ | 1055/1250 [19:05<03:22,  1.04s/it, loss=0.613]
Epoch 3/4:  84%|████████▍ | 1055/1250 [19:06<03:22,  1.04s/it, loss=0.898]
Epoch 3/4:  84%|████████▍ | 1056/1250 [19:06<03:20,  1.03s/it, loss=0.898]
Epoch 3/4:  84%|████████▍ | 1056/1250 [19:07<03:20,  1.03s/it, loss=0.579]
Epoch 3/4:  85%|████████▍ | 1057/1250 [19:07<03:25,  1.07s/it, loss=0.579]
Epoch 3/4:  85%|████████▍ | 1057/1250 [19:08<03:25,  1.07s/it, loss=0.549]
Epoch 3/4:  85%|████████▍ | 1058/1250 [19:08<03:26,  1.08s/it, loss=0.549]
Epoch 3/4:  85%|████████▍ | 1058/1250 [19:09<03:26,  1.08s/it, loss=0.739]
Epoch 3/4:  85%|████████▍ | 1059/1250 [19:09<03:20,  1.05s/it, loss=0.739]
Epoch 3/4:  85%|████████▍ | 1059/1250 [19:10<03:20,  1.05s/it, loss=0.879]
Epoch 3/4:  85%|████████▍ | 1060/1250 [19:10<03:14,  1.02s/it, loss=0.879]
Epoch 3/4:  85%|████████▍ | 1060/1250 [19:11<03:14,  1.02s/it, loss=0.542]
Epoch 3/4:  85%|████████▍ | 1061/1250 [19:11<03:16,  1.04s/it, loss=0.542]
Epoch 3/4:  85%|████████▍ | 1061/1250 [19:12<03:16,  1.04s/it, loss=0.65] 
Epoch 3/4:  85%|████████▍ | 1062/1250 [19:12<03:24,  1.09s/it, loss=0.65]
Epoch 3/4:  85%|████████▍ | 1062/1250 [19:13<03:24,  1.09s/it, loss=0.789]
Epoch 3/4:  85%|████████▌ | 1063/1250 [19:13<03:24,  1.09s/it, loss=0.789]
Epoch 3/4:  85%|████████▌ | 1063/1250 [19:14<03:24,  1.09s/it, loss=0.46] 
Epoch 3/4:  85%|████████▌ | 1064/1250 [19:14<03:17,  1.06s/it, loss=0.46]
Epoch 3/4:  85%|████████▌ | 1064/1250 [19:15<03:17,  1.06s/it, loss=0.444]
Epoch 3/4:  85%|████████▌ | 1065/1250 [19:15<03:12,  1.04s/it, loss=0.444]
Epoch 3/4:  85%|████████▌ | 1065/1250 [19:16<03:12,  1.04s/it, loss=0.61] 
Epoch 3/4:  85%|████████▌ | 1066/1250 [19:16<03:15,  1.06s/it, loss=0.61]
Epoch 3/4:  85%|████████▌ | 1066/1250 [19:17<03:15,  1.06s/it, loss=0.55]
Epoch 3/4:  85%|████████▌ | 1067/1250 [19:17<03:17,  1.08s/it, loss=0.55]
Epoch 3/4:  85%|████████▌ | 1067/1250 [19:18<03:17,  1.08s/it, loss=0.644]
Epoch 3/4:  85%|████████▌ | 1068/1250 [19:18<03:17,  1.08s/it, loss=0.644]
Epoch 3/4:  85%|████████▌ | 1068/1250 [19:19<03:17,  1.08s/it, loss=0.52] 
Epoch 3/4:  86%|████████▌ | 1069/1250 [19:19<03:11,  1.06s/it, loss=0.52]
Epoch 3/4:  86%|████████▌ | 1069/1250 [19:20<03:11,  1.06s/it, loss=0.467]
Epoch 3/4:  86%|████████▌ | 1070/1250 [19:20<03:06,  1.03s/it, loss=0.467]
Epoch 3/4:  86%|████████▌ | 1070/1250 [19:21<03:06,  1.03s/it, loss=0.405]
Epoch 3/4:  86%|████████▌ | 1071/1250 [19:21<03:05,  1.04s/it, loss=0.405]
Epoch 3/4:  86%|████████▌ | 1071/1250 [19:22<03:05,  1.04s/it, loss=0.684]
Epoch 3/4:  86%|████████▌ | 1072/1250 [19:22<03:05,  1.04s/it, loss=0.684]
Epoch 3/4:  86%|████████▌ | 1072/1250 [19:23<03:05,  1.04s/it, loss=0.42] 
Epoch 3/4:  86%|████████▌ | 1073/1250 [19:23<03:02,  1.03s/it, loss=0.42]
Epoch 3/4:  86%|████████▌ | 1073/1250 [19:24<03:02,  1.03s/it, loss=0.681]
Epoch 3/4:  86%|████████▌ | 1074/1250 [19:24<02:59,  1.02s/it, loss=0.681]
Epoch 3/4:  86%|████████▌ | 1074/1250 [19:26<02:59,  1.02s/it, loss=0.706]
Epoch 3/4:  86%|████████▌ | 1075/1250 [19:26<03:03,  1.05s/it, loss=0.706]
Epoch 3/4:  86%|████████▌ | 1075/1250 [19:27<03:03,  1.05s/it, loss=0.641]
Epoch 3/4:  86%|████████▌ | 1076/1250 [19:27<03:10,  1.10s/it, loss=0.641]
Epoch 3/4:  86%|████████▌ | 1076/1250 [19:28<03:10,  1.10s/it, loss=0.491]
Epoch 3/4:  86%|████████▌ | 1077/1250 [19:28<03:08,  1.09s/it, loss=0.491]
Epoch 3/4:  86%|████████▌ | 1077/1250 [19:29<03:08,  1.09s/it, loss=0.591]
Epoch 3/4:  86%|████████▌ | 1078/1250 [19:29<03:00,  1.05s/it, loss=0.591]
Epoch 3/4:  86%|████████▌ | 1078/1250 [19:30<03:00,  1.05s/it, loss=0.397]
Epoch 3/4:  86%|████████▋ | 1079/1250 [19:30<02:52,  1.01s/it, loss=0.397]
Epoch 3/4:  86%|████████▋ | 1079/1250 [19:31<02:52,  1.01s/it, loss=0.826]
Epoch 3/4:  86%|████████▋ | 1080/1250 [19:31<02:51,  1.01s/it, loss=0.826]
Epoch 3/4:  86%|████████▋ | 1080/1250 [19:32<02:51,  1.01s/it, loss=0.607]
Epoch 3/4:  86%|████████▋ | 1081/1250 [19:32<02:58,  1.06s/it, loss=0.607]
Epoch 3/4:  86%|████████▋ | 1081/1250 [19:33<02:58,  1.06s/it, loss=0.495]
Epoch 3/4:  87%|████████▋ | 1082/1250 [19:33<03:01,  1.08s/it, loss=0.495]
Epoch 3/4:  87%|████████▋ | 1082/1250 [19:34<03:01,  1.08s/it, loss=0.559]
Epoch 3/4:  87%|████████▋ | 1083/1250 [19:34<02:55,  1.05s/it, loss=0.559]
Epoch 3/4:  87%|████████▋ | 1083/1250 [19:35<02:55,  1.05s/it, loss=0.443]
Epoch 3/4:  87%|████████▋ | 1084/1250 [19:35<02:52,  1.04s/it, loss=0.443]
Epoch 3/4:  87%|████████▋ | 1084/1250 [19:36<02:52,  1.04s/it, loss=0.452]
Epoch 3/4:  87%|████████▋ | 1085/1250 [19:36<02:53,  1.05s/it, loss=0.452]
Epoch 3/4:  87%|████████▋ | 1085/1250 [19:37<02:53,  1.05s/it, loss=0.384]
Epoch 3/4:  87%|████████▋ | 1086/1250 [19:37<02:54,  1.06s/it, loss=0.384]
Epoch 3/4:  87%|████████▋ | 1086/1250 [19:38<02:54,  1.06s/it, loss=0.658]
Epoch 3/4:  87%|████████▋ | 1087/1250 [19:38<02:51,  1.05s/it, loss=0.658]
Epoch 3/4:  87%|████████▋ | 1087/1250 [19:39<02:51,  1.05s/it, loss=0.663]
Epoch 3/4:  87%|████████▋ | 1088/1250 [19:39<02:47,  1.03s/it, loss=0.663]
Epoch 3/4:  87%|████████▋ | 1088/1250 [19:40<02:47,  1.03s/it, loss=0.34] 
Epoch 3/4:  87%|████████▋ | 1089/1250 [19:40<02:44,  1.02s/it, loss=0.34]
Epoch 3/4:  87%|████████▋ | 1089/1250 [19:41<02:44,  1.02s/it, loss=0.624]
Epoch 3/4:  87%|████████▋ | 1090/1250 [19:41<02:45,  1.04s/it, loss=0.624]
Epoch 3/4:  87%|████████▋ | 1090/1250 [19:42<02:45,  1.04s/it, loss=0.682]
Epoch 3/4:  87%|████████▋ | 1091/1250 [19:42<02:47,  1.05s/it, loss=0.682]
Epoch 3/4:  87%|████████▋ | 1091/1250 [19:43<02:47,  1.05s/it, loss=0.604]
Epoch 3/4:  87%|████████▋ | 1092/1250 [19:43<02:45,  1.05s/it, loss=0.604]
Epoch 3/4:  87%|████████▋ | 1092/1250 [19:44<02:45,  1.05s/it, loss=0.475]
Epoch 3/4:  87%|████████▋ | 1093/1250 [19:44<02:43,  1.04s/it, loss=0.475]
Epoch 3/4:  87%|████████▋ | 1093/1250 [19:46<02:43,  1.04s/it, loss=0.551]
Epoch 3/4:  88%|████████▊ | 1094/1250 [19:46<02:43,  1.05s/it, loss=0.551]
Epoch 3/4:  88%|████████▊ | 1094/1250 [19:47<02:43,  1.05s/it, loss=0.367]
Epoch 3/4:  88%|████████▊ | 1095/1250 [19:47<02:46,  1.07s/it, loss=0.367]
Epoch 3/4:  88%|████████▊ | 1095/1250 [19:48<02:46,  1.07s/it, loss=0.516]
Epoch 3/4:  88%|████████▊ | 1096/1250 [19:48<02:47,  1.09s/it, loss=0.516]
Epoch 3/4:  88%|████████▊ | 1096/1250 [19:49<02:47,  1.09s/it, loss=0.822]
Epoch 3/4:  88%|████████▊ | 1097/1250 [19:49<02:42,  1.06s/it, loss=0.822]
Epoch 3/4:  88%|████████▊ | 1097/1250 [19:50<02:42,  1.06s/it, loss=0.778]
Epoch 3/4:  88%|████████▊ | 1098/1250 [19:50<02:35,  1.03s/it, loss=0.778]
Epoch 3/4:  88%|████████▊ | 1098/1250 [19:51<02:35,  1.03s/it, loss=0.815]
Epoch 3/4:  88%|████████▊ | 1099/1250 [19:51<02:32,  1.01s/it, loss=0.815]
Epoch 3/4:  88%|████████▊ | 1099/1250 [19:52<02:32,  1.01s/it, loss=0.656]
Epoch 3/4:  88%|████████▊ | 1100/1250 [19:52<02:35,  1.04s/it, loss=0.656]
Epoch 3/4:  88%|████████▊ | 1100/1250 [19:53<02:35,  1.04s/it, loss=0.714]
Epoch 3/4:  88%|████████▊ | 1101/1250 [19:53<02:41,  1.08s/it, loss=0.714]
Epoch 3/4:  88%|████████▊ | 1101/1250 [19:54<02:41,  1.08s/it, loss=0.832]
Epoch 3/4:  88%|████████▊ | 1102/1250 [19:54<02:37,  1.07s/it, loss=0.832]
Epoch 3/4:  88%|████████▊ | 1102/1250 [19:55<02:37,  1.07s/it, loss=0.559]
Epoch 3/4:  88%|████████▊ | 1103/1250 [19:55<02:32,  1.04s/it, loss=0.559]
Epoch 3/4:  88%|████████▊ | 1103/1250 [19:56<02:32,  1.04s/it, loss=0.612]
Epoch 3/4:  88%|████████▊ | 1104/1250 [19:56<02:30,  1.03s/it, loss=0.612]
Epoch 3/4:  88%|████████▊ | 1104/1250 [19:57<02:30,  1.03s/it, loss=0.645]
Epoch 3/4:  88%|████████▊ | 1105/1250 [19:57<02:32,  1.05s/it, loss=0.645]
Epoch 3/4:  88%|████████▊ | 1105/1250 [19:58<02:32,  1.05s/it, loss=0.7]  
Epoch 3/4:  88%|████████▊ | 1106/1250 [19:58<02:33,  1.07s/it, loss=0.7]
Epoch 3/4:  88%|████████▊ | 1106/1250 [19:59<02:33,  1.07s/it, loss=0.575]
Epoch 3/4:  89%|████████▊ | 1107/1250 [19:59<02:30,  1.05s/it, loss=0.575]
Epoch 3/4:  89%|████████▊ | 1107/1250 [20:00<02:30,  1.05s/it, loss=0.449]
Epoch 3/4:  89%|████████▊ | 1108/1250 [20:00<02:28,  1.04s/it, loss=0.449]
Epoch 3/4:  89%|████████▊ | 1108/1250 [20:01<02:28,  1.04s/it, loss=0.659]
Epoch 3/4:  89%|████████▊ | 1109/1250 [20:01<02:27,  1.05s/it, loss=0.659]
Epoch 3/4:  89%|████████▊ | 1109/1250 [20:02<02:27,  1.05s/it, loss=0.656]
Epoch 3/4:  89%|████████▉ | 1110/1250 [20:02<02:28,  1.06s/it, loss=0.656]
Epoch 3/4:  89%|████████▉ | 1110/1250 [20:03<02:28,  1.06s/it, loss=0.705]
Epoch 3/4:  89%|████████▉ | 1111/1250 [20:03<02:25,  1.05s/it, loss=0.705]
Epoch 3/4:  89%|████████▉ | 1111/1250 [20:04<02:25,  1.05s/it, loss=0.554]
Epoch 3/4:  89%|████████▉ | 1112/1250 [20:04<02:20,  1.02s/it, loss=0.554]
Epoch 3/4:  89%|████████▉ | 1112/1250 [20:05<02:20,  1.02s/it, loss=0.535]
Epoch 3/4:  89%|████████▉ | 1113/1250 [20:05<02:17,  1.00s/it, loss=0.535]
Epoch 3/4:  89%|████████▉ | 1113/1250 [20:06<02:17,  1.00s/it, loss=0.767]
Epoch 3/4:  89%|████████▉ | 1114/1250 [20:06<02:19,  1.03s/it, loss=0.767]
Epoch 3/4:  89%|████████▉ | 1114/1250 [20:08<02:19,  1.03s/it, loss=0.672]
Epoch 3/4:  89%|████████▉ | 1115/1250 [20:08<02:23,  1.06s/it, loss=0.672]
Epoch 3/4:  89%|████████▉ | 1115/1250 [20:09<02:23,  1.06s/it, loss=0.92] 
Epoch 3/4:  89%|████████▉ | 1116/1250 [20:09<02:23,  1.07s/it, loss=0.92]
Epoch 3/4:  89%|████████▉ | 1116/1250 [20:10<02:23,  1.07s/it, loss=0.488]
Epoch 3/4:  89%|████████▉ | 1117/1250 [20:10<02:20,  1.06s/it, loss=0.488]
Epoch 3/4:  89%|████████▉ | 1117/1250 [20:11<02:20,  1.06s/it, loss=0.521]
Epoch 3/4:  89%|████████▉ | 1118/1250 [20:11<02:19,  1.05s/it, loss=0.521]
Epoch 3/4:  89%|████████▉ | 1118/1250 [20:12<02:19,  1.05s/it, loss=0.53] 
Epoch 3/4:  90%|████████▉ | 1119/1250 [20:12<02:21,  1.08s/it, loss=0.53]
Epoch 3/4:  90%|████████▉ | 1119/1250 [20:13<02:21,  1.08s/it, loss=0.479]
Epoch 3/4:  90%|████████▉ | 1120/1250 [20:13<02:22,  1.10s/it, loss=0.479]
Epoch 3/4:  90%|████████▉ | 1120/1250 [20:14<02:22,  1.10s/it, loss=0.729]
Epoch 3/4:  90%|████████▉ | 1121/1250 [20:14<02:18,  1.07s/it, loss=0.729]
Epoch 3/4:  90%|████████▉ | 1121/1250 [20:15<02:18,  1.07s/it, loss=0.689]
Epoch 3/4:  90%|████████▉ | 1122/1250 [20:15<02:14,  1.05s/it, loss=0.689]
Epoch 3/4:  90%|████████▉ | 1122/1250 [20:16<02:14,  1.05s/it, loss=0.497]
Epoch 3/4:  90%|████████▉ | 1123/1250 [20:16<02:12,  1.04s/it, loss=0.497]
Epoch 3/4:  90%|████████▉ | 1123/1250 [20:17<02:12,  1.04s/it, loss=0.576]
Epoch 3/4:  90%|████████▉ | 1124/1250 [20:17<02:12,  1.05s/it, loss=0.576]
Epoch 3/4:  90%|████████▉ | 1124/1250 [20:18<02:12,  1.05s/it, loss=0.53] 
Epoch 3/4:  90%|█████████ | 1125/1250 [20:18<02:12,  1.06s/it, loss=0.53]
Epoch 3/4:  90%|█████████ | 1125/1250 [20:19<02:12,  1.06s/it, loss=0.801]
Epoch 3/4:  90%|█████████ | 1126/1250 [20:19<02:10,  1.05s/it, loss=0.801]
Epoch 3/4:  90%|█████████ | 1126/1250 [20:20<02:10,  1.05s/it, loss=0.551]
Epoch 3/4:  90%|█████████ | 1127/1250 [20:20<02:06,  1.03s/it, loss=0.551]
Epoch 3/4:  90%|█████████ | 1127/1250 [20:21<02:06,  1.03s/it, loss=0.432]
Epoch 3/4:  90%|█████████ | 1128/1250 [20:21<02:05,  1.03s/it, loss=0.432]
Epoch 3/4:  90%|█████████ | 1128/1250 [20:22<02:05,  1.03s/it, loss=0.5]  
Epoch 3/4:  90%|█████████ | 1129/1250 [20:22<02:07,  1.05s/it, loss=0.5]
Epoch 3/4:  90%|█████████ | 1129/1250 [20:23<02:07,  1.05s/it, loss=0.588]
Epoch 3/4:  90%|█████████ | 1130/1250 [20:23<02:07,  1.06s/it, loss=0.588]
Epoch 3/4:  90%|█████████ | 1130/1250 [20:24<02:07,  1.06s/it, loss=0.538]
Epoch 3/4:  90%|█████████ | 1131/1250 [20:24<02:03,  1.04s/it, loss=0.538]
Epoch 3/4:  90%|█████████ | 1131/1250 [20:25<02:03,  1.04s/it, loss=0.548]
Epoch 3/4:  91%|█████████ | 1132/1250 [20:25<02:01,  1.03s/it, loss=0.548]
Epoch 3/4:  91%|█████████ | 1132/1250 [20:26<02:01,  1.03s/it, loss=0.7]  
Epoch 3/4:  91%|█████████ | 1133/1250 [20:26<02:01,  1.03s/it, loss=0.7]
Epoch 3/4:  91%|█████████ | 1133/1250 [20:28<02:01,  1.03s/it, loss=0.526]
Epoch 3/4:  91%|█████████ | 1134/1250 [20:28<02:01,  1.05s/it, loss=0.526]
Epoch 3/4:  91%|█████████ | 1134/1250 [20:29<02:01,  1.05s/it, loss=0.593]
Epoch 3/4:  91%|█████████ | 1135/1250 [20:29<02:00,  1.05s/it, loss=0.593]
Epoch 3/4:  91%|█████████ | 1135/1250 [20:30<02:00,  1.05s/it, loss=0.606]
Epoch 3/4:  91%|█████████ | 1136/1250 [20:30<01:57,  1.03s/it, loss=0.606]
Epoch 3/4:  91%|█████████ | 1136/1250 [20:31<01:57,  1.03s/it, loss=0.439]
Epoch 3/4:  91%|█████████ | 1137/1250 [20:31<01:54,  1.02s/it, loss=0.439]
Epoch 3/4:  91%|█████████ | 1137/1250 [20:32<01:54,  1.02s/it, loss=0.55] 
Epoch 3/4:  91%|█████████ | 1138/1250 [20:32<01:56,  1.04s/it, loss=0.55]
Epoch 3/4:  91%|█████████ | 1138/1250 [20:33<01:56,  1.04s/it, loss=0.719]
Epoch 3/4:  91%|█████████ | 1139/1250 [20:33<01:59,  1.08s/it, loss=0.719]
Epoch 3/4:  91%|█████████ | 1139/1250 [20:34<01:59,  1.08s/it, loss=0.71] 
Epoch 3/4:  91%|█████████ | 1140/1250 [20:34<01:56,  1.06s/it, loss=0.71]
Epoch 3/4:  91%|█████████ | 1140/1250 [20:35<01:56,  1.06s/it, loss=0.629]
Epoch 3/4:  91%|█████████▏| 1141/1250 [20:35<01:53,  1.04s/it, loss=0.629]
Epoch 3/4:  91%|█████████▏| 1141/1250 [20:36<01:53,  1.04s/it, loss=0.61] 
Epoch 3/4:  91%|█████████▏| 1142/1250 [20:36<01:52,  1.04s/it, loss=0.61]
Epoch 3/4:  91%|█████████▏| 1142/1250 [20:37<01:52,  1.04s/it, loss=0.567]
Epoch 3/4:  91%|█████████▏| 1143/1250 [20:37<01:52,  1.05s/it, loss=0.567]
Epoch 3/4:  91%|█████████▏| 1143/1250 [20:38<01:52,  1.05s/it, loss=0.539]
Epoch 3/4:  92%|█████████▏| 1144/1250 [20:38<01:51,  1.05s/it, loss=0.539]
Epoch 3/4:  92%|█████████▏| 1144/1250 [20:39<01:51,  1.05s/it, loss=0.625]
Epoch 3/4:  92%|█████████▏| 1145/1250 [20:39<01:49,  1.04s/it, loss=0.625]
Epoch 3/4:  92%|█████████▏| 1145/1250 [20:40<01:49,  1.04s/it, loss=0.756]
Epoch 3/4:  92%|█████████▏| 1146/1250 [20:40<01:45,  1.02s/it, loss=0.756]
Epoch 3/4:  92%|█████████▏| 1146/1250 [20:41<01:45,  1.02s/it, loss=0.725]
Epoch 3/4:  92%|█████████▏| 1147/1250 [20:41<01:43,  1.01s/it, loss=0.725]
Epoch 3/4:  92%|█████████▏| 1147/1250 [20:42<01:43,  1.01s/it, loss=0.765]
Epoch 3/4:  92%|█████████▏| 1148/1250 [20:42<01:45,  1.04s/it, loss=0.765]
Epoch 3/4:  92%|█████████▏| 1148/1250 [20:43<01:45,  1.04s/it, loss=0.498]
Epoch 3/4:  92%|█████████▏| 1149/1250 [20:43<01:47,  1.06s/it, loss=0.498]
Epoch 3/4:  92%|█████████▏| 1149/1250 [20:44<01:47,  1.06s/it, loss=0.595]
Epoch 3/4:  92%|█████████▏| 1150/1250 [20:44<01:45,  1.05s/it, loss=0.595]
Epoch 3/4:  92%|█████████▏| 1150/1250 [20:45<01:45,  1.05s/it, loss=0.61] 
Epoch 3/4:  92%|█████████▏| 1151/1250 [20:45<01:44,  1.05s/it, loss=0.61]
Epoch 3/4:  92%|█████████▏| 1151/1250 [20:46<01:44,  1.05s/it, loss=0.48]
Epoch 3/4:  92%|█████████▏| 1152/1250 [20:46<01:44,  1.07s/it, loss=0.48]
Epoch 3/4:  92%|█████████▏| 1152/1250 [20:47<01:44,  1.07s/it, loss=0.679]
Epoch 3/4:  92%|█████████▏| 1153/1250 [20:47<01:44,  1.08s/it, loss=0.679]
Epoch 3/4:  92%|█████████▏| 1153/1250 [20:49<01:44,  1.08s/it, loss=0.436]
Epoch 3/4:  92%|█████████▏| 1154/1250 [20:49<01:42,  1.07s/it, loss=0.436]
Epoch 3/4:  92%|█████████▏| 1154/1250 [20:50<01:42,  1.07s/it, loss=0.529]
Epoch 3/4:  92%|█████████▏| 1155/1250 [20:50<01:39,  1.05s/it, loss=0.529]
Epoch 3/4:  92%|█████████▏| 1155/1250 [20:50<01:39,  1.05s/it, loss=0.545]
Epoch 3/4:  92%|█████████▏| 1156/1250 [20:50<01:36,  1.03s/it, loss=0.545]
Epoch 3/4:  92%|█████████▏| 1156/1250 [20:52<01:36,  1.03s/it, loss=0.49] 
Epoch 3/4:  93%|█████████▎| 1157/1250 [20:52<01:35,  1.03s/it, loss=0.49]
Epoch 3/4:  93%|█████████▎| 1157/1250 [20:53<01:35,  1.03s/it, loss=0.627]
Epoch 3/4:  93%|█████████▎| 1158/1250 [20:53<01:38,  1.08s/it, loss=0.627]
Epoch 3/4:  93%|█████████▎| 1158/1250 [20:54<01:38,  1.08s/it, loss=0.569]
Epoch 3/4:  93%|█████████▎| 1159/1250 [20:54<01:40,  1.10s/it, loss=0.569]
Epoch 3/4:  93%|█████████▎| 1159/1250 [20:55<01:40,  1.10s/it, loss=0.531]
Epoch 3/4:  93%|█████████▎| 1160/1250 [20:55<01:36,  1.07s/it, loss=0.531]
Epoch 3/4:  93%|█████████▎| 1160/1250 [20:56<01:36,  1.07s/it, loss=0.659]
Epoch 3/4:  93%|█████████▎| 1161/1250 [20:56<01:32,  1.04s/it, loss=0.659]
Epoch 3/4:  93%|█████████▎| 1161/1250 [20:57<01:32,  1.04s/it, loss=0.444]
Epoch 3/4:  93%|█████████▎| 1162/1250 [20:57<01:30,  1.03s/it, loss=0.444]
Epoch 3/4:  93%|█████████▎| 1162/1250 [20:58<01:30,  1.03s/it, loss=0.496]
Epoch 3/4:  93%|█████████▎| 1163/1250 [20:58<01:30,  1.04s/it, loss=0.496]
Epoch 3/4:  93%|█████████▎| 1163/1250 [20:59<01:30,  1.04s/it, loss=0.451]
Epoch 3/4:  93%|█████████▎| 1164/1250 [20:59<01:29,  1.04s/it, loss=0.451]
Epoch 3/4:  93%|█████████▎| 1164/1250 [21:00<01:29,  1.04s/it, loss=0.615]
Epoch 3/4:  93%|█████████▎| 1165/1250 [21:00<01:27,  1.03s/it, loss=0.615]
Epoch 3/4:  93%|█████████▎| 1165/1250 [21:01<01:27,  1.03s/it, loss=0.547]
Epoch 3/4:  93%|█████████▎| 1166/1250 [21:01<01:25,  1.02s/it, loss=0.547]
Epoch 3/4:  93%|█████████▎| 1166/1250 [21:02<01:25,  1.02s/it, loss=0.507]
Epoch 3/4:  93%|█████████▎| 1167/1250 [21:02<01:25,  1.03s/it, loss=0.507]
Epoch 3/4:  93%|█████████▎| 1167/1250 [21:03<01:25,  1.03s/it, loss=0.96] 
Epoch 3/4:  93%|█████████▎| 1168/1250 [21:03<01:26,  1.05s/it, loss=0.96]
Epoch 3/4:  93%|█████████▎| 1168/1250 [21:04<01:26,  1.05s/it, loss=0.473]
Epoch 3/4:  94%|█████████▎| 1169/1250 [21:04<01:24,  1.05s/it, loss=0.473]
Epoch 3/4:  94%|█████████▎| 1169/1250 [21:05<01:24,  1.05s/it, loss=0.859]
Epoch 3/4:  94%|█████████▎| 1170/1250 [21:05<01:22,  1.03s/it, loss=0.859]
Epoch 3/4:  94%|█████████▎| 1170/1250 [21:06<01:22,  1.03s/it, loss=0.526]
Epoch 3/4:  94%|█████████▎| 1171/1250 [21:06<01:21,  1.03s/it, loss=0.526]
Epoch 3/4:  94%|█████████▎| 1171/1250 [21:07<01:21,  1.03s/it, loss=0.823]
Epoch 3/4:  94%|█████████▍| 1172/1250 [21:07<01:21,  1.05s/it, loss=0.823]
Epoch 3/4:  94%|█████████▍| 1172/1250 [21:08<01:21,  1.05s/it, loss=0.373]
Epoch 3/4:  94%|█████████▍| 1173/1250 [21:08<01:21,  1.06s/it, loss=0.373]
Epoch 3/4:  94%|█████████▍| 1173/1250 [21:09<01:21,  1.06s/it, loss=0.568]
Epoch 3/4:  94%|█████████▍| 1174/1250 [21:09<01:19,  1.05s/it, loss=0.568]
Epoch 3/4:  94%|█████████▍| 1174/1250 [21:10<01:19,  1.05s/it, loss=0.399]
Epoch 3/4:  94%|█████████▍| 1175/1250 [21:10<01:17,  1.03s/it, loss=0.399]
Epoch 3/4:  94%|█████████▍| 1175/1250 [21:11<01:17,  1.03s/it, loss=0.575]
Epoch 3/4:  94%|█████████▍| 1176/1250 [21:11<01:16,  1.03s/it, loss=0.575]
Epoch 3/4:  94%|█████████▍| 1176/1250 [21:12<01:16,  1.03s/it, loss=0.624]
Epoch 3/4:  94%|█████████▍| 1177/1250 [21:12<01:16,  1.05s/it, loss=0.624]
Epoch 3/4:  94%|█████████▍| 1177/1250 [21:14<01:16,  1.05s/it, loss=0.635]
Epoch 3/4:  94%|█████████▍| 1178/1250 [21:14<01:16,  1.06s/it, loss=0.635]
Epoch 3/4:  94%|█████████▍| 1178/1250 [21:15<01:16,  1.06s/it, loss=0.444]
Epoch 3/4:  94%|█████████▍| 1179/1250 [21:15<01:14,  1.05s/it, loss=0.444]
Epoch 3/4:  94%|█████████▍| 1179/1250 [21:16<01:14,  1.05s/it, loss=0.566]
Epoch 3/4:  94%|█████████▍| 1180/1250 [21:16<01:13,  1.04s/it, loss=0.566]
Epoch 3/4:  94%|█████████▍| 1180/1250 [21:17<01:13,  1.04s/it, loss=0.639]
Epoch 3/4:  94%|█████████▍| 1181/1250 [21:17<01:11,  1.04s/it, loss=0.639]
Epoch 3/4:  94%|█████████▍| 1181/1250 [21:18<01:11,  1.04s/it, loss=0.642]
Epoch 3/4:  95%|█████████▍| 1182/1250 [21:18<01:11,  1.05s/it, loss=0.642]
Epoch 3/4:  95%|█████████▍| 1182/1250 [21:19<01:11,  1.05s/it, loss=0.546]
Epoch 3/4:  95%|█████████▍| 1183/1250 [21:19<01:11,  1.06s/it, loss=0.546]
Epoch 3/4:  95%|█████████▍| 1183/1250 [21:20<01:11,  1.06s/it, loss=0.507]
Epoch 3/4:  95%|█████████▍| 1184/1250 [21:20<01:09,  1.05s/it, loss=0.507]
Epoch 3/4:  95%|█████████▍| 1184/1250 [21:21<01:09,  1.05s/it, loss=0.606]
Epoch 3/4:  95%|█████████▍| 1185/1250 [21:21<01:06,  1.02s/it, loss=0.606]
Epoch 3/4:  95%|█████████▍| 1185/1250 [21:22<01:06,  1.02s/it, loss=0.624]
Epoch 3/4:  95%|█████████▍| 1186/1250 [21:22<01:05,  1.03s/it, loss=0.624]
Epoch 3/4:  95%|█████████▍| 1186/1250 [21:23<01:05,  1.03s/it, loss=0.752]
Epoch 3/4:  95%|█████████▍| 1187/1250 [21:23<01:06,  1.05s/it, loss=0.752]
Epoch 3/4:  95%|█████████▍| 1187/1250 [21:24<01:06,  1.05s/it, loss=0.62] 
Epoch 3/4:  95%|█████████▌| 1188/1250 [21:24<01:05,  1.06s/it, loss=0.62]
Epoch 3/4:  95%|█████████▌| 1188/1250 [21:25<01:05,  1.06s/it, loss=0.55]
Epoch 3/4:  95%|█████████▌| 1189/1250 [21:25<01:03,  1.04s/it, loss=0.55]
Epoch 3/4:  95%|█████████▌| 1189/1250 [21:26<01:03,  1.04s/it, loss=0.52]
Epoch 3/4:  95%|█████████▌| 1190/1250 [21:26<01:01,  1.02s/it, loss=0.52]
Epoch 3/4:  95%|█████████▌| 1190/1250 [21:27<01:01,  1.02s/it, loss=0.531]
Epoch 3/4:  95%|█████████▌| 1191/1250 [21:27<01:00,  1.02s/it, loss=0.531]
Epoch 3/4:  95%|█████████▌| 1191/1250 [21:28<01:00,  1.02s/it, loss=0.623]
Epoch 3/4:  95%|█████████▌| 1192/1250 [21:28<01:00,  1.04s/it, loss=0.623]
Epoch 3/4:  95%|█████████▌| 1192/1250 [21:29<01:00,  1.04s/it, loss=0.654]
Epoch 3/4:  95%|█████████▌| 1193/1250 [21:29<01:00,  1.06s/it, loss=0.654]
Epoch 3/4:  95%|█████████▌| 1193/1250 [21:30<01:00,  1.06s/it, loss=0.516]
Epoch 3/4:  96%|█████████▌| 1194/1250 [21:30<00:58,  1.04s/it, loss=0.516]
Epoch 3/4:  96%|█████████▌| 1194/1250 [21:31<00:58,  1.04s/it, loss=0.625]
Epoch 3/4:  96%|█████████▌| 1195/1250 [21:31<00:55,  1.01s/it, loss=0.625]
Epoch 3/4:  96%|█████████▌| 1195/1250 [21:32<00:55,  1.01s/it, loss=0.455]
Epoch 3/4:  96%|█████████▌| 1196/1250 [21:32<00:54,  1.01s/it, loss=0.455]
Epoch 3/4:  96%|█████████▌| 1196/1250 [21:33<00:54,  1.01s/it, loss=0.832]
Epoch 3/4:  96%|█████████▌| 1197/1250 [21:33<00:54,  1.04s/it, loss=0.832]
Epoch 3/4:  96%|█████████▌| 1197/1250 [21:34<00:54,  1.04s/it, loss=0.501]
Epoch 3/4:  96%|█████████▌| 1198/1250 [21:34<00:54,  1.05s/it, loss=0.501]
Epoch 3/4:  96%|█████████▌| 1198/1250 [21:35<00:54,  1.05s/it, loss=0.432]
Epoch 3/4:  96%|█████████▌| 1199/1250 [21:35<00:53,  1.05s/it, loss=0.432]
Epoch 3/4:  96%|█████████▌| 1199/1250 [21:36<00:53,  1.05s/it, loss=0.423]
Epoch 3/4:  96%|█████████▌| 1200/1250 [21:36<00:52,  1.05s/it, loss=0.423]
Epoch 3/4:  96%|█████████▌| 1200/1250 [21:38<00:52,  1.05s/it, loss=0.75] 
Epoch 3/4:  96%|█████████▌| 1201/1250 [21:38<00:52,  1.07s/it, loss=0.75]
Epoch 3/4:  96%|█████████▌| 1201/1250 [21:39<00:52,  1.07s/it, loss=0.609]
Epoch 3/4:  96%|█████████▌| 1202/1250 [21:39<00:52,  1.08s/it, loss=0.609]
Epoch 3/4:  96%|█████████▌| 1202/1250 [21:40<00:52,  1.08s/it, loss=0.7]  
Epoch 3/4:  96%|█████████▌| 1203/1250 [21:40<00:50,  1.07s/it, loss=0.7]
Epoch 3/4:  96%|█████████▌| 1203/1250 [21:41<00:50,  1.07s/it, loss=0.463]
Epoch 3/4:  96%|█████████▋| 1204/1250 [21:41<00:47,  1.03s/it, loss=0.463]
Epoch 3/4:  96%|█████████▋| 1204/1250 [21:42<00:47,  1.03s/it, loss=0.596]
Epoch 3/4:  96%|█████████▋| 1205/1250 [21:42<00:46,  1.03s/it, loss=0.596]
Epoch 3/4:  96%|█████████▋| 1205/1250 [21:43<00:46,  1.03s/it, loss=0.485]
Epoch 3/4:  96%|█████████▋| 1206/1250 [21:43<00:46,  1.05s/it, loss=0.485]
Epoch 3/4:  96%|█████████▋| 1206/1250 [21:44<00:46,  1.05s/it, loss=0.495]
Epoch 3/4:  97%|█████████▋| 1207/1250 [21:44<00:46,  1.08s/it, loss=0.495]
Epoch 3/4:  97%|█████████▋| 1207/1250 [21:45<00:46,  1.08s/it, loss=0.827]
Epoch 3/4:  97%|█████████▋| 1208/1250 [21:45<00:44,  1.07s/it, loss=0.827]
Epoch 3/4:  97%|█████████▋| 1208/1250 [21:46<00:44,  1.07s/it, loss=0.695]
Epoch 3/4:  97%|█████████▋| 1209/1250 [21:46<00:42,  1.05s/it, loss=0.695]
Epoch 3/4:  97%|█████████▋| 1209/1250 [21:47<00:42,  1.05s/it, loss=0.662]
Epoch 3/4:  97%|█████████▋| 1210/1250 [21:47<00:41,  1.04s/it, loss=0.662]
Epoch 3/4:  97%|█████████▋| 1210/1250 [21:48<00:41,  1.04s/it, loss=0.618]
Epoch 3/4:  97%|█████████▋| 1211/1250 [21:48<00:40,  1.04s/it, loss=0.618]
Epoch 3/4:  97%|█████████▋| 1211/1250 [21:49<00:40,  1.04s/it, loss=0.563]
Epoch 3/4:  97%|█████████▋| 1212/1250 [21:49<00:39,  1.03s/it, loss=0.563]
Epoch 3/4:  97%|█████████▋| 1212/1250 [21:50<00:39,  1.03s/it, loss=0.633]
Epoch 3/4:  97%|█████████▋| 1213/1250 [21:50<00:37,  1.02s/it, loss=0.633]
Epoch 3/4:  97%|█████████▋| 1213/1250 [21:51<00:37,  1.02s/it, loss=0.436]
Epoch 3/4:  97%|█████████▋| 1214/1250 [21:51<00:36,  1.00s/it, loss=0.436]
Epoch 3/4:  97%|█████████▋| 1214/1250 [21:52<00:36,  1.00s/it, loss=0.796]
Epoch 3/4:  97%|█████████▋| 1215/1250 [21:52<00:35,  1.02s/it, loss=0.796]
Epoch 3/4:  97%|█████████▋| 1215/1250 [21:53<00:35,  1.02s/it, loss=0.493]
Epoch 3/4:  97%|█████████▋| 1216/1250 [21:53<00:36,  1.07s/it, loss=0.493]
Epoch 3/4:  97%|█████████▋| 1216/1250 [21:54<00:36,  1.07s/it, loss=0.507]
Epoch 3/4:  97%|█████████▋| 1217/1250 [21:54<00:35,  1.09s/it, loss=0.507]
Epoch 3/4:  97%|█████████▋| 1217/1250 [21:55<00:35,  1.09s/it, loss=0.541]
Epoch 3/4:  97%|█████████▋| 1218/1250 [21:55<00:33,  1.06s/it, loss=0.541]
Epoch 3/4:  97%|█████████▋| 1218/1250 [21:56<00:33,  1.06s/it, loss=0.636]
Epoch 3/4:  98%|█████████▊| 1219/1250 [21:56<00:31,  1.03s/it, loss=0.636]
Epoch 3/4:  98%|█████████▊| 1219/1250 [21:57<00:31,  1.03s/it, loss=0.544]
Epoch 3/4:  98%|█████████▊| 1220/1250 [21:57<00:30,  1.03s/it, loss=0.544]
Epoch 3/4:  98%|█████████▊| 1220/1250 [21:58<00:30,  1.03s/it, loss=0.615]
Epoch 3/4:  98%|█████████▊| 1221/1250 [21:58<00:30,  1.06s/it, loss=0.615]
Epoch 3/4:  98%|█████████▊| 1221/1250 [22:00<00:30,  1.06s/it, loss=0.567]
Epoch 3/4:  98%|█████████▊| 1222/1250 [22:00<00:30,  1.07s/it, loss=0.567]
Epoch 3/4:  98%|█████████▊| 1222/1250 [22:01<00:30,  1.07s/it, loss=0.43] 
Epoch 3/4:  98%|█████████▊| 1223/1250 [22:01<00:28,  1.05s/it, loss=0.43]
Epoch 3/4:  98%|█████████▊| 1223/1250 [22:02<00:28,  1.05s/it, loss=0.515]
Epoch 3/4:  98%|█████████▊| 1224/1250 [22:02<00:27,  1.04s/it, loss=0.515]
Epoch 3/4:  98%|█████████▊| 1224/1250 [22:03<00:27,  1.04s/it, loss=0.594]
Epoch 3/4:  98%|█████████▊| 1225/1250 [22:03<00:26,  1.05s/it, loss=0.594]
Epoch 3/4:  98%|█████████▊| 1225/1250 [22:04<00:26,  1.05s/it, loss=0.554]
Epoch 3/4:  98%|█████████▊| 1226/1250 [22:04<00:25,  1.06s/it, loss=0.554]
Epoch 3/4:  98%|█████████▊| 1226/1250 [22:05<00:25,  1.06s/it, loss=0.532]
Epoch 3/4:  98%|█████████▊| 1227/1250 [22:05<00:24,  1.04s/it, loss=0.532]
Epoch 3/4:  98%|█████████▊| 1227/1250 [22:06<00:24,  1.04s/it, loss=0.794]
Epoch 3/4:  98%|█████████▊| 1228/1250 [22:06<00:22,  1.02s/it, loss=0.794]
Epoch 3/4:  98%|█████████▊| 1228/1250 [22:07<00:22,  1.02s/it, loss=0.792]
Epoch 3/4:  98%|█████████▊| 1229/1250 [22:07<00:21,  1.00s/it, loss=0.792]
Epoch 3/4:  98%|█████████▊| 1229/1250 [22:08<00:21,  1.00s/it, loss=0.457]
Epoch 3/4:  98%|█████████▊| 1230/1250 [22:08<00:20,  1.02s/it, loss=0.457]
Epoch 3/4:  98%|█████████▊| 1230/1250 [22:09<00:20,  1.02s/it, loss=0.403]
Epoch 3/4:  98%|█████████▊| 1231/1250 [22:09<00:19,  1.05s/it, loss=0.403]
Epoch 3/4:  98%|█████████▊| 1231/1250 [22:10<00:19,  1.05s/it, loss=0.67] 
Epoch 3/4:  99%|█████████▊| 1232/1250 [22:10<00:18,  1.05s/it, loss=0.67]
Epoch 3/4:  99%|█████████▊| 1232/1250 [22:11<00:18,  1.05s/it, loss=0.451]
Epoch 3/4:  99%|█████████▊| 1233/1250 [22:11<00:17,  1.05s/it, loss=0.451]
Epoch 3/4:  99%|█████████▊| 1233/1250 [22:12<00:17,  1.05s/it, loss=0.447]
Epoch 3/4:  99%|█████████▊| 1234/1250 [22:12<00:16,  1.06s/it, loss=0.447]
Epoch 3/4:  99%|█████████▊| 1234/1250 [22:13<00:16,  1.06s/it, loss=0.648]
Epoch 3/4:  99%|█████████▉| 1235/1250 [22:13<00:16,  1.07s/it, loss=0.648]
Epoch 3/4:  99%|█████████▉| 1235/1250 [22:14<00:16,  1.07s/it, loss=0.445]
Epoch 3/4:  99%|█████████▉| 1236/1250 [22:14<00:15,  1.07s/it, loss=0.445]
Epoch 3/4:  99%|█████████▉| 1236/1250 [22:15<00:15,  1.07s/it, loss=0.559]
Epoch 3/4:  99%|█████████▉| 1237/1250 [22:15<00:13,  1.06s/it, loss=0.559]
Epoch 3/4:  99%|█████████▉| 1237/1250 [22:16<00:13,  1.06s/it, loss=0.658]
Epoch 3/4:  99%|█████████▉| 1238/1250 [22:16<00:12,  1.03s/it, loss=0.658]
Epoch 3/4:  99%|█████████▉| 1238/1250 [22:17<00:12,  1.03s/it, loss=0.686]
Epoch 3/4:  99%|█████████▉| 1239/1250 [22:17<00:11,  1.03s/it, loss=0.686]
Epoch 3/4:  99%|█████████▉| 1239/1250 [22:18<00:11,  1.03s/it, loss=0.677]
Epoch 3/4:  99%|█████████▉| 1240/1250 [22:18<00:10,  1.04s/it, loss=0.677]
Epoch 3/4:  99%|█████████▉| 1240/1250 [22:19<00:10,  1.04s/it, loss=0.571]
Epoch 3/4:  99%|█████████▉| 1241/1250 [22:19<00:09,  1.06s/it, loss=0.571]
Epoch 3/4:  99%|█████████▉| 1241/1250 [22:20<00:09,  1.06s/it, loss=0.483]
Epoch 3/4:  99%|█████████▉| 1242/1250 [22:20<00:08,  1.05s/it, loss=0.483]
Epoch 3/4:  99%|█████████▉| 1242/1250 [22:21<00:08,  1.05s/it, loss=0.442]
Epoch 3/4:  99%|█████████▉| 1243/1250 [22:21<00:07,  1.03s/it, loss=0.442]
Epoch 3/4:  99%|█████████▉| 1243/1250 [22:22<00:07,  1.03s/it, loss=0.651]
Epoch 3/4: 100%|█████████▉| 1244/1250 [22:22<00:06,  1.05s/it, loss=0.651]
Epoch 3/4: 100%|█████████▉| 1244/1250 [22:24<00:06,  1.05s/it, loss=0.718]
Epoch 3/4: 100%|█████████▉| 1245/1250 [22:24<00:05,  1.06s/it, loss=0.718]
Epoch 3/4: 100%|█████████▉| 1245/1250 [22:25<00:05,  1.06s/it, loss=0.664]
Epoch 3/4: 100%|█████████▉| 1246/1250 [22:25<00:04,  1.07s/it, loss=0.664]
Epoch 3/4: 100%|█████████▉| 1246/1250 [22:26<00:04,  1.07s/it, loss=0.407]
Epoch 3/4: 100%|█████████▉| 1247/1250 [22:26<00:03,  1.06s/it, loss=0.407]
Epoch 3/4: 100%|█████████▉| 1247/1250 [22:27<00:03,  1.06s/it, loss=0.493]
Epoch 3/4: 100%|█████████▉| 1248/1250 [22:27<00:02,  1.03s/it, loss=0.493]
Epoch 3/4: 100%|█████████▉| 1248/1250 [22:28<00:02,  1.03s/it, loss=0.499]
Epoch 3/4: 100%|█████████▉| 1249/1250 [22:28<00:01,  1.02s/it, loss=0.499]
Epoch 3/4: 100%|█████████▉| 1249/1250 [22:29<00:01,  1.02s/it, loss=0.671]
Epoch 3/4: 100%|██████████| 1250/1250 [22:29<00:00,  1.03s/it, loss=0.671]
Epoch 3/4: 100%|██████████| 1250/1250 [22:29<00:00,  1.08s/it, loss=0.671]
Epoch 3 Avg Loss: 0.5675

Epoch 4/4:   0%|          | 0/1250 [00:00<?, ?it/s]
Epoch 4/4:   0%|          | 0/1250 [00:01<?, ?it/s, loss=0.371]
Epoch 4/4:   0%|          | 1/1250 [00:01<21:29,  1.03s/it, loss=0.371]
Epoch 4/4:   0%|          | 1/1250 [00:02<21:29,  1.03s/it, loss=0.463]
Epoch 4/4:   0%|          | 2/1250 [00:02<20:57,  1.01s/it, loss=0.463]
Epoch 4/4:   0%|          | 2/1250 [00:03<20:57,  1.01s/it, loss=0.413]
Epoch 4/4:   0%|          | 3/1250 [00:03<20:59,  1.01s/it, loss=0.413]
Epoch 4/4:   0%|          | 3/1250 [00:04<20:59,  1.01s/it, loss=0.343]
Epoch 4/4:   0%|          | 4/1250 [00:04<21:31,  1.04s/it, loss=0.343]
Epoch 4/4:   0%|          | 4/1250 [00:05<21:31,  1.04s/it, loss=0.508]
Epoch 4/4:   0%|          | 5/1250 [00:05<21:58,  1.06s/it, loss=0.508]
Epoch 4/4:   0%|          | 5/1250 [00:06<21:58,  1.06s/it, loss=0.43] 
Epoch 4/4:   0%|          | 6/1250 [00:06<22:01,  1.06s/it, loss=0.43]
Epoch 4/4:   0%|          | 6/1250 [00:07<22:01,  1.06s/it, loss=0.391]
Epoch 4/4:   1%|          | 7/1250 [00:07<21:36,  1.04s/it, loss=0.391]
Epoch 4/4:   1%|          | 7/1250 [00:08<21:36,  1.04s/it, loss=0.44] 
Epoch 4/4:   1%|          | 8/1250 [00:08<21:08,  1.02s/it, loss=0.44]
Epoch 4/4:   1%|          | 8/1250 [00:09<21:08,  1.02s/it, loss=0.36]
Epoch 4/4:   1%|          | 9/1250 [00:09<21:22,  1.03s/it, loss=0.36]
Epoch 4/4:   1%|          | 9/1250 [00:10<21:22,  1.03s/it, loss=0.406]
Epoch 4/4:   1%|          | 10/1250 [00:10<21:47,  1.05s/it, loss=0.406]
Epoch 4/4:   1%|          | 10/1250 [00:11<21:47,  1.05s/it, loss=0.565]
Epoch 4/4:   1%|          | 11/1250 [00:11<21:38,  1.05s/it, loss=0.565]
Epoch 4/4:   1%|          | 11/1250 [00:12<21:38,  1.05s/it, loss=0.415]
Epoch 4/4:   1%|          | 12/1250 [00:12<21:29,  1.04s/it, loss=0.415]
Epoch 4/4:   1%|          | 12/1250 [00:13<21:29,  1.04s/it, loss=0.491]
Epoch 4/4:   1%|          | 13/1250 [00:13<21:37,  1.05s/it, loss=0.491]
Epoch 4/4:   1%|          | 13/1250 [00:14<21:37,  1.05s/it, loss=0.44] 
Epoch 4/4:   1%|          | 14/1250 [00:14<21:54,  1.06s/it, loss=0.44]
Epoch 4/4:   1%|          | 14/1250 [00:15<21:54,  1.06s/it, loss=0.42]
Epoch 4/4:   1%|          | 15/1250 [00:15<22:06,  1.07s/it, loss=0.42]
Epoch 4/4:   1%|          | 15/1250 [00:16<22:06,  1.07s/it, loss=0.461]
Epoch 4/4:   1%|▏         | 16/1250 [00:16<21:57,  1.07s/it, loss=0.461]
Epoch 4/4:   1%|▏         | 16/1250 [00:17<21:57,  1.07s/it, loss=0.455]
Epoch 4/4:   1%|▏         | 17/1250 [00:17<21:23,  1.04s/it, loss=0.455]
Epoch 4/4:   1%|▏         | 17/1250 [00:18<21:23,  1.04s/it, loss=0.473]
Epoch 4/4:   1%|▏         | 18/1250 [00:18<21:05,  1.03s/it, loss=0.473]
Epoch 4/4:   1%|▏         | 18/1250 [00:19<21:05,  1.03s/it, loss=0.521]
Epoch 4/4:   2%|▏         | 19/1250 [00:19<21:37,  1.05s/it, loss=0.521]
Epoch 4/4:   2%|▏         | 19/1250 [00:20<21:37,  1.05s/it, loss=0.516]
Epoch 4/4:   2%|▏         | 20/1250 [00:20<21:57,  1.07s/it, loss=0.516]
Epoch 4/4:   2%|▏         | 20/1250 [00:21<21:57,  1.07s/it, loss=0.393]
Epoch 4/4:   2%|▏         | 21/1250 [00:21<21:28,  1.05s/it, loss=0.393]
Epoch 4/4:   2%|▏         | 21/1250 [00:22<21:28,  1.05s/it, loss=0.495]
Epoch 4/4:   2%|▏         | 22/1250 [00:22<20:55,  1.02s/it, loss=0.495]
Epoch 4/4:   2%|▏         | 22/1250 [00:23<20:55,  1.02s/it, loss=0.314]
Epoch 4/4:   2%|▏         | 23/1250 [00:23<20:55,  1.02s/it, loss=0.314]
Epoch 4/4:   2%|▏         | 23/1250 [00:25<20:55,  1.02s/it, loss=0.367]
Epoch 4/4:   2%|▏         | 24/1250 [00:25<21:17,  1.04s/it, loss=0.367]
Epoch 4/4:   2%|▏         | 24/1250 [00:26<21:17,  1.04s/it, loss=0.429]
Epoch 4/4:   2%|▏         | 25/1250 [00:26<21:26,  1.05s/it, loss=0.429]
Epoch 4/4:   2%|▏         | 25/1250 [00:27<21:26,  1.05s/it, loss=0.59] 
Epoch 4/4:   2%|▏         | 26/1250 [00:27<21:03,  1.03s/it, loss=0.59]
Epoch 4/4:   2%|▏         | 26/1250 [00:28<21:03,  1.03s/it, loss=0.338]
Epoch 4/4:   2%|▏         | 27/1250 [00:28<20:47,  1.02s/it, loss=0.338]
Epoch 4/4:   2%|▏         | 27/1250 [00:29<20:47,  1.02s/it, loss=0.367]
Epoch 4/4:   2%|▏         | 28/1250 [00:29<21:12,  1.04s/it, loss=0.367]
Epoch 4/4:   2%|▏         | 28/1250 [00:30<21:12,  1.04s/it, loss=0.437]
Epoch 4/4:   2%|▏         | 29/1250 [00:30<21:36,  1.06s/it, loss=0.437]
Epoch 4/4:   2%|▏         | 29/1250 [00:31<21:36,  1.06s/it, loss=0.595]
Epoch 4/4:   2%|▏         | 30/1250 [00:31<21:30,  1.06s/it, loss=0.595]
Epoch 4/4:   2%|▏         | 30/1250 [00:32<21:30,  1.06s/it, loss=0.453]
Epoch 4/4:   2%|▏         | 31/1250 [00:32<20:57,  1.03s/it, loss=0.453]
Epoch 4/4:   2%|▏         | 31/1250 [00:33<20:57,  1.03s/it, loss=0.372]
Epoch 4/4:   3%|▎         | 32/1250 [00:33<20:30,  1.01s/it, loss=0.372]
Epoch 4/4:   3%|▎         | 32/1250 [00:34<20:30,  1.01s/it, loss=0.426]
Epoch 4/4:   3%|▎         | 33/1250 [00:34<20:56,  1.03s/it, loss=0.426]
Epoch 4/4:   3%|▎         | 33/1250 [00:35<20:56,  1.03s/it, loss=0.295]
Epoch 4/4:   3%|▎         | 34/1250 [00:35<21:48,  1.08s/it, loss=0.295]
Epoch 4/4:   3%|▎         | 34/1250 [00:36<21:48,  1.08s/it, loss=0.481]
Epoch 4/4:   3%|▎         | 35/1250 [00:36<22:07,  1.09s/it, loss=0.481]
Epoch 4/4:   3%|▎         | 35/1250 [00:37<22:07,  1.09s/it, loss=0.442]
Epoch 4/4:   3%|▎         | 36/1250 [00:37<21:43,  1.07s/it, loss=0.442]
Epoch 4/4:   3%|▎         | 36/1250 [00:38<21:43,  1.07s/it, loss=0.393]
Epoch 4/4:   3%|▎         | 37/1250 [00:38<21:04,  1.04s/it, loss=0.393]
Epoch 4/4:   3%|▎         | 37/1250 [00:39<21:04,  1.04s/it, loss=0.618]
Epoch 4/4:   3%|▎         | 38/1250 [00:39<20:54,  1.04s/it, loss=0.618]
Epoch 4/4:   3%|▎         | 38/1250 [00:40<20:54,  1.04s/it, loss=0.33] 
Epoch 4/4:   3%|▎         | 39/1250 [00:40<21:05,  1.04s/it, loss=0.33]
Epoch 4/4:   3%|▎         | 39/1250 [00:41<21:05,  1.04s/it, loss=0.379]
Epoch 4/4:   3%|▎         | 40/1250 [00:41<21:03,  1.04s/it, loss=0.379]
Epoch 4/4:   3%|▎         | 40/1250 [00:42<21:03,  1.04s/it, loss=0.378]
Epoch 4/4:   3%|▎         | 41/1250 [00:42<20:48,  1.03s/it, loss=0.378]
Epoch 4/4:   3%|▎         | 41/1250 [00:43<20:48,  1.03s/it, loss=0.232]
Epoch 4/4:   3%|▎         | 42/1250 [00:43<20:38,  1.03s/it, loss=0.232]
Epoch 4/4:   3%|▎         | 42/1250 [00:44<20:38,  1.03s/it, loss=0.335]
Epoch 4/4:   3%|▎         | 43/1250 [00:44<21:03,  1.05s/it, loss=0.335]
Epoch 4/4:   3%|▎         | 43/1250 [00:46<21:03,  1.05s/it, loss=0.415]
Epoch 4/4:   4%|▎         | 44/1250 [00:46<21:38,  1.08s/it, loss=0.415]
Epoch 4/4:   4%|▎         | 44/1250 [00:47<21:38,  1.08s/it, loss=0.336]
Epoch 4/4:   4%|▎         | 45/1250 [00:47<21:44,  1.08s/it, loss=0.336]
Epoch 4/4:   4%|▎         | 45/1250 [00:48<21:44,  1.08s/it, loss=0.512]
Epoch 4/4:   4%|▎         | 46/1250 [00:48<21:11,  1.06s/it, loss=0.512]
Epoch 4/4:   4%|▎         | 46/1250 [00:49<21:11,  1.06s/it, loss=0.505]
Epoch 4/4:   4%|▍         | 47/1250 [00:49<20:48,  1.04s/it, loss=0.505]
Epoch 4/4:   4%|▍         | 47/1250 [00:50<20:48,  1.04s/it, loss=0.366]
Epoch 4/4:   4%|▍         | 48/1250 [00:50<20:41,  1.03s/it, loss=0.366]
Epoch 4/4:   4%|▍         | 48/1250 [00:51<20:41,  1.03s/it, loss=0.456]
Epoch 4/4:   4%|▍         | 49/1250 [00:51<20:49,  1.04s/it, loss=0.456]
Epoch 4/4:   4%|▍         | 49/1250 [00:52<20:49,  1.04s/it, loss=0.448]
Epoch 4/4:   4%|▍         | 50/1250 [00:52<20:28,  1.02s/it, loss=0.448]
Epoch 4/4:   4%|▍         | 50/1250 [00:53<20:28,  1.02s/it, loss=0.321]
Epoch 4/4:   4%|▍         | 51/1250 [00:53<20:08,  1.01s/it, loss=0.321]
Epoch 4/4:   4%|▍         | 51/1250 [00:54<20:08,  1.01s/it, loss=0.325]
Epoch 4/4:   4%|▍         | 52/1250 [00:54<20:13,  1.01s/it, loss=0.325]
Epoch 4/4:   4%|▍         | 52/1250 [00:55<20:13,  1.01s/it, loss=0.317]
Epoch 4/4:   4%|▍         | 53/1250 [00:55<20:49,  1.04s/it, loss=0.317]
Epoch 4/4:   4%|▍         | 53/1250 [00:56<20:49,  1.04s/it, loss=0.446]
Epoch 4/4:   4%|▍         | 54/1250 [00:56<21:16,  1.07s/it, loss=0.446]
Epoch 4/4:   4%|▍         | 54/1250 [00:57<21:16,  1.07s/it, loss=0.358]
Epoch 4/4:   4%|▍         | 55/1250 [00:57<21:07,  1.06s/it, loss=0.358]
Epoch 4/4:   4%|▍         | 55/1250 [00:58<21:07,  1.06s/it, loss=0.321]
Epoch 4/4:   4%|▍         | 56/1250 [00:58<20:53,  1.05s/it, loss=0.321]
Epoch 4/4:   4%|▍         | 56/1250 [00:59<20:53,  1.05s/it, loss=0.379]
Epoch 4/4:   5%|▍         | 57/1250 [00:59<20:50,  1.05s/it, loss=0.379]
Epoch 4/4:   5%|▍         | 57/1250 [01:00<20:50,  1.05s/it, loss=0.361]
Epoch 4/4:   5%|▍         | 58/1250 [01:00<20:57,  1.05s/it, loss=0.361]
Epoch 4/4:   5%|▍         | 58/1250 [01:01<20:57,  1.05s/it, loss=0.314]
Epoch 4/4:   5%|▍         | 59/1250 [01:01<20:48,  1.05s/it, loss=0.314]
Epoch 4/4:   5%|▍         | 59/1250 [01:02<20:48,  1.05s/it, loss=0.437]
Epoch 4/4:   5%|▍         | 60/1250 [01:02<20:18,  1.02s/it, loss=0.437]
Epoch 4/4:   5%|▍         | 60/1250 [01:03<20:18,  1.02s/it, loss=0.429]
Epoch 4/4:   5%|▍         | 61/1250 [01:03<19:53,  1.00s/it, loss=0.429]
Epoch 4/4:   5%|▍         | 61/1250 [01:04<19:53,  1.00s/it, loss=0.355]
Epoch 4/4:   5%|▍         | 62/1250 [01:04<19:59,  1.01s/it, loss=0.355]
Epoch 4/4:   5%|▍         | 62/1250 [01:05<19:59,  1.01s/it, loss=0.391]
Epoch 4/4:   5%|▌         | 63/1250 [01:05<20:41,  1.05s/it, loss=0.391]
Epoch 4/4:   5%|▌         | 63/1250 [01:06<20:41,  1.05s/it, loss=0.381]
Epoch 4/4:   5%|▌         | 64/1250 [01:06<21:06,  1.07s/it, loss=0.381]
Epoch 4/4:   5%|▌         | 64/1250 [01:07<21:06,  1.07s/it, loss=0.351]
Epoch 4/4:   5%|▌         | 65/1250 [01:07<20:54,  1.06s/it, loss=0.351]
Epoch 4/4:   5%|▌         | 65/1250 [01:08<20:54,  1.06s/it, loss=0.382]
Epoch 4/4:   5%|▌         | 66/1250 [01:08<20:46,  1.05s/it, loss=0.382]
Epoch 4/4:   5%|▌         | 66/1250 [01:10<20:46,  1.05s/it, loss=0.553]
Epoch 4/4:   5%|▌         | 67/1250 [01:10<21:00,  1.07s/it, loss=0.553]
Epoch 4/4:   5%|▌         | 67/1250 [01:11<21:00,  1.07s/it, loss=0.502]
Epoch 4/4:   5%|▌         | 68/1250 [01:11<21:09,  1.07s/it, loss=0.502]
Epoch 4/4:   5%|▌         | 68/1250 [01:12<21:09,  1.07s/it, loss=0.552]
Epoch 4/4:   6%|▌         | 69/1250 [01:12<20:49,  1.06s/it, loss=0.552]
Epoch 4/4:   6%|▌         | 69/1250 [01:13<20:49,  1.06s/it, loss=0.51] 
Epoch 4/4:   6%|▌         | 70/1250 [01:13<20:16,  1.03s/it, loss=0.51]
Epoch 4/4:   6%|▌         | 70/1250 [01:14<20:16,  1.03s/it, loss=0.349]
Epoch 4/4:   6%|▌         | 71/1250 [01:14<20:03,  1.02s/it, loss=0.349]
Epoch 4/4:   6%|▌         | 71/1250 [01:15<20:03,  1.02s/it, loss=0.456]
Epoch 4/4:   6%|▌         | 72/1250 [01:15<20:17,  1.03s/it, loss=0.456]
Epoch 4/4:   6%|▌         | 72/1250 [01:16<20:17,  1.03s/it, loss=0.431]
Epoch 4/4:   6%|▌         | 73/1250 [01:16<21:23,  1.09s/it, loss=0.431]
Epoch 4/4:   6%|▌         | 73/1250 [01:17<21:23,  1.09s/it, loss=0.382]
Epoch 4/4:   6%|▌         | 74/1250 [01:17<21:37,  1.10s/it, loss=0.382]
Epoch 4/4:   6%|▌         | 74/1250 [01:18<21:37,  1.10s/it, loss=0.394]
Epoch 4/4:   6%|▌         | 75/1250 [01:18<21:03,  1.08s/it, loss=0.394]
Epoch 4/4:   6%|▌         | 75/1250 [01:19<21:03,  1.08s/it, loss=0.418]
Epoch 4/4:   6%|▌         | 76/1250 [01:19<20:35,  1.05s/it, loss=0.418]
Epoch 4/4:   6%|▌         | 76/1250 [01:20<20:35,  1.05s/it, loss=0.259]
Epoch 4/4:   6%|▌         | 77/1250 [01:20<20:27,  1.05s/it, loss=0.259]
Epoch 4/4:   6%|▌         | 77/1250 [01:21<20:27,  1.05s/it, loss=0.426]
Epoch 4/4:   6%|▌         | 78/1250 [01:21<20:26,  1.05s/it, loss=0.426]
Epoch 4/4:   6%|▌         | 78/1250 [01:22<20:26,  1.05s/it, loss=1.18] 
Epoch 4/4:   6%|▋         | 79/1250 [01:22<20:03,  1.03s/it, loss=1.18]
Epoch 4/4:   6%|▋         | 79/1250 [01:23<20:03,  1.03s/it, loss=0.411]
Epoch 4/4:   6%|▋         | 80/1250 [01:23<19:45,  1.01s/it, loss=0.411]
Epoch 4/4:   6%|▋         | 80/1250 [01:24<19:45,  1.01s/it, loss=0.416]
Epoch 4/4:   6%|▋         | 81/1250 [01:24<19:54,  1.02s/it, loss=0.416]
Epoch 4/4:   6%|▋         | 81/1250 [01:25<19:54,  1.02s/it, loss=0.494]
Epoch 4/4:   7%|▋         | 82/1250 [01:25<20:20,  1.04s/it, loss=0.494]
Epoch 4/4:   7%|▋         | 82/1250 [01:26<20:20,  1.04s/it, loss=0.468]
Epoch 4/4:   7%|▋         | 83/1250 [01:26<20:29,  1.05s/it, loss=0.468]
Epoch 4/4:   7%|▋         | 83/1250 [01:27<20:29,  1.05s/it, loss=0.552]
Epoch 4/4:   7%|▋         | 84/1250 [01:27<20:10,  1.04s/it, loss=0.552]
Epoch 4/4:   7%|▋         | 84/1250 [01:28<20:10,  1.04s/it, loss=0.469]
Epoch 4/4:   7%|▋         | 85/1250 [01:28<20:05,  1.03s/it, loss=0.469]
Epoch 4/4:   7%|▋         | 85/1250 [01:29<20:05,  1.03s/it, loss=0.373]
Epoch 4/4:   7%|▋         | 86/1250 [01:29<20:15,  1.04s/it, loss=0.373]
Epoch 4/4:   7%|▋         | 86/1250 [01:31<20:15,  1.04s/it, loss=0.389]
Epoch 4/4:   7%|▋         | 87/1250 [01:31<20:34,  1.06s/it, loss=0.389]
Epoch 4/4:   7%|▋         | 87/1250 [01:32<20:34,  1.06s/it, loss=0.483]
Epoch 4/4:   7%|▋         | 88/1250 [01:32<20:26,  1.06s/it, loss=0.483]
Epoch 4/4:   7%|▋         | 88/1250 [01:33<20:26,  1.06s/it, loss=0.405]
Epoch 4/4:   7%|▋         | 89/1250 [01:33<20:00,  1.03s/it, loss=0.405]
Epoch 4/4:   7%|▋         | 89/1250 [01:33<20:00,  1.03s/it, loss=0.365]
Epoch 4/4:   7%|▋         | 90/1250 [01:33<19:36,  1.01s/it, loss=0.365]
Epoch 4/4:   7%|▋         | 90/1250 [01:35<19:36,  1.01s/it, loss=0.448]
Epoch 4/4:   7%|▋         | 91/1250 [01:35<19:45,  1.02s/it, loss=0.448]
Epoch 4/4:   7%|▋         | 91/1250 [01:36<19:45,  1.02s/it, loss=0.292]
Epoch 4/4:   7%|▋         | 92/1250 [01:36<20:22,  1.06s/it, loss=0.292]
Epoch 4/4:   7%|▋         | 92/1250 [01:37<20:22,  1.06s/it, loss=0.395]
Epoch 4/4:   7%|▋         | 93/1250 [01:37<20:35,  1.07s/it, loss=0.395]
Epoch 4/4:   7%|▋         | 93/1250 [01:38<20:35,  1.07s/it, loss=0.337]
Epoch 4/4:   8%|▊         | 94/1250 [01:38<20:10,  1.05s/it, loss=0.337]
Epoch 4/4:   8%|▊         | 94/1250 [01:39<20:10,  1.05s/it, loss=0.387]
Epoch 4/4:   8%|▊         | 95/1250 [01:39<19:50,  1.03s/it, loss=0.387]
Epoch 4/4:   8%|▊         | 95/1250 [01:40<19:50,  1.03s/it, loss=0.462]
Epoch 4/4:   8%|▊         | 96/1250 [01:40<20:14,  1.05s/it, loss=0.462]
Epoch 4/4:   8%|▊         | 96/1250 [01:41<20:14,  1.05s/it, loss=0.438]
Epoch 4/4:   8%|▊         | 97/1250 [01:41<20:55,  1.09s/it, loss=0.438]
Epoch 4/4:   8%|▊         | 97/1250 [01:42<20:55,  1.09s/it, loss=0.445]
Epoch 4/4:   8%|▊         | 98/1250 [01:42<20:54,  1.09s/it, loss=0.445]
Epoch 4/4:   8%|▊         | 98/1250 [01:43<20:54,  1.09s/it, loss=0.349]
Epoch 4/4:   8%|▊         | 99/1250 [01:43<20:15,  1.06s/it, loss=0.349]
Epoch 4/4:   8%|▊         | 99/1250 [01:44<20:15,  1.06s/it, loss=0.336]
Epoch 4/4:   8%|▊         | 100/1250 [01:44<19:44,  1.03s/it, loss=0.336]
Epoch 4/4:   8%|▊         | 100/1250 [01:45<19:44,  1.03s/it, loss=0.384]
Epoch 4/4:   8%|▊         | 101/1250 [01:45<19:54,  1.04s/it, loss=0.384]
Epoch 4/4:   8%|▊         | 101/1250 [01:46<19:54,  1.04s/it, loss=0.378]
Epoch 4/4:   8%|▊         | 102/1250 [01:46<20:39,  1.08s/it, loss=0.378]
Epoch 4/4:   8%|▊         | 102/1250 [01:47<20:39,  1.08s/it, loss=0.383]
Epoch 4/4:   8%|▊         | 103/1250 [01:47<20:29,  1.07s/it, loss=0.383]
Epoch 4/4:   8%|▊         | 103/1250 [01:48<20:29,  1.07s/it, loss=0.326]
Epoch 4/4:   8%|▊         | 104/1250 [01:48<19:51,  1.04s/it, loss=0.326]
Epoch 4/4:   8%|▊         | 104/1250 [01:49<19:51,  1.04s/it, loss=0.389]
Epoch 4/4:   8%|▊         | 105/1250 [01:49<19:28,  1.02s/it, loss=0.389]
Epoch 4/4:   8%|▊         | 105/1250 [01:50<19:28,  1.02s/it, loss=0.35] 
Epoch 4/4:   8%|▊         | 106/1250 [01:50<19:31,  1.02s/it, loss=0.35]
Epoch 4/4:   8%|▊         | 106/1250 [01:51<19:31,  1.02s/it, loss=0.497]
Epoch 4/4:   9%|▊         | 107/1250 [01:51<19:45,  1.04s/it, loss=0.497]
Epoch 4/4:   9%|▊         | 107/1250 [01:52<19:45,  1.04s/it, loss=0.418]
Epoch 4/4:   9%|▊         | 108/1250 [01:52<19:35,  1.03s/it, loss=0.418]
Epoch 4/4:   9%|▊         | 108/1250 [01:53<19:35,  1.03s/it, loss=0.389]
Epoch 4/4:   9%|▊         | 109/1250 [01:53<19:32,  1.03s/it, loss=0.389]
Epoch 4/4:   9%|▊         | 109/1250 [01:56<19:32,  1.03s/it, loss=0.443]
Epoch 4/4:   9%|▉         | 110/1250 [01:56<25:54,  1.36s/it, loss=0.443]
Epoch 4/4:   9%|▉         | 110/1250 [01:57<25:54,  1.36s/it, loss=0.598]
Epoch 4/4:   9%|▉         | 111/1250 [01:57<26:11,  1.38s/it, loss=0.598]
Epoch 4/4:   9%|▉         | 111/1250 [01:59<26:11,  1.38s/it, loss=0.472]
Epoch 4/4:   9%|▉         | 112/1250 [01:59<28:01,  1.48s/it, loss=0.472]
Epoch 4/4:   9%|▉         | 112/1250 [02:01<28:01,  1.48s/it, loss=0.343]
Epoch 4/4:   9%|▉         | 113/1250 [02:01<30:48,  1.63s/it, loss=0.343]
Epoch 4/4:   9%|▉         | 113/1250 [02:03<30:48,  1.63s/it, loss=0.473]
Epoch 4/4:   9%|▉         | 114/1250 [02:03<32:13,  1.70s/it, loss=0.473]
Epoch 4/4:   9%|▉         | 114/1250 [02:04<32:13,  1.70s/it, loss=0.296]
Epoch 4/4:   9%|▉         | 115/1250 [02:04<31:29,  1.66s/it, loss=0.296]
Epoch 4/4:   9%|▉         | 115/1250 [02:06<31:29,  1.66s/it, loss=0.372]
Epoch 4/4:   9%|▉         | 116/1250 [02:06<32:28,  1.72s/it, loss=0.372]
Epoch 4/4:   9%|▉         | 116/1250 [02:08<32:28,  1.72s/it, loss=0.38] 
Epoch 4/4:   9%|▉         | 117/1250 [02:08<32:03,  1.70s/it, loss=0.38]
Epoch 4/4:   9%|▉         | 117/1250 [02:09<32:03,  1.70s/it, loss=0.432]
Epoch 4/4:   9%|▉         | 118/1250 [02:09<31:37,  1.68s/it, loss=0.432]
Epoch 4/4:   9%|▉         | 118/1250 [02:11<31:37,  1.68s/it, loss=0.399]
Epoch 4/4:  10%|▉         | 119/1250 [02:11<33:05,  1.76s/it, loss=0.399]
Epoch 4/4:  10%|▉         | 119/1250 [02:14<33:05,  1.76s/it, loss=0.631]
Epoch 4/4:  10%|▉         | 120/1250 [02:14<39:41,  2.11s/it, loss=0.631]
Epoch 4/4:  10%|▉         | 120/1250 [02:18<39:41,  2.11s/it, loss=0.501]
Epoch 4/4:  10%|▉         | 121/1250 [02:18<46:51,  2.49s/it, loss=0.501]
Epoch 4/4:  10%|▉         | 121/1250 [02:19<46:51,  2.49s/it, loss=0.345]
Epoch 4/4:  10%|▉         | 122/1250 [02:19<43:48,  2.33s/it, loss=0.345]
Epoch 4/4:  10%|▉         | 122/1250 [02:21<43:48,  2.33s/it, loss=0.437]
Epoch 4/4:  10%|▉         | 123/1250 [02:21<38:37,  2.06s/it, loss=0.437]
Epoch 4/4:  10%|▉         | 123/1250 [02:24<38:37,  2.06s/it, loss=0.308]
Epoch 4/4:  10%|▉         | 124/1250 [02:24<41:47,  2.23s/it, loss=0.308]
Epoch 4/4:  10%|▉         | 124/1250 [02:25<41:47,  2.23s/it, loss=0.454]
Epoch 4/4:  10%|█         | 125/1250 [02:25<38:40,  2.06s/it, loss=0.454]
Epoch 4/4:  10%|█         | 125/1250 [02:27<38:40,  2.06s/it, loss=0.342]
Epoch 4/4:  10%|█         | 126/1250 [02:27<38:32,  2.06s/it, loss=0.342]
Epoch 4/4:  10%|█         | 126/1250 [02:29<38:32,  2.06s/it, loss=0.585]
Epoch 4/4:  10%|█         | 127/1250 [02:29<38:56,  2.08s/it, loss=0.585]
Epoch 4/4:  10%|█         | 127/1250 [02:32<38:56,  2.08s/it, loss=0.423]
Epoch 4/4:  10%|█         | 128/1250 [02:32<41:09,  2.20s/it, loss=0.423]
Epoch 4/4:  10%|█         | 128/1250 [02:34<41:09,  2.20s/it, loss=0.616]
Epoch 4/4:  10%|█         | 129/1250 [02:34<38:59,  2.09s/it, loss=0.616]
Epoch 4/4:  10%|█         | 129/1250 [02:36<38:59,  2.09s/it, loss=0.512]
Epoch 4/4:  10%|█         | 130/1250 [02:36<38:52,  2.08s/it, loss=0.512]
Epoch 4/4:  10%|█         | 130/1250 [02:37<38:52,  2.08s/it, loss=0.486]
Epoch 4/4:  10%|█         | 131/1250 [02:37<34:55,  1.87s/it, loss=0.486]
Epoch 4/4:  10%|█         | 131/1250 [02:39<34:55,  1.87s/it, loss=0.471]
Epoch 4/4:  11%|█         | 132/1250 [02:39<34:59,  1.88s/it, loss=0.471]
Epoch 4/4:  11%|█         | 132/1250 [02:41<34:59,  1.88s/it, loss=0.441]
Epoch 4/4:  11%|█         | 133/1250 [02:41<33:51,  1.82s/it, loss=0.441]
Epoch 4/4:  11%|█         | 133/1250 [02:43<33:51,  1.82s/it, loss=0.519]
Epoch 4/4:  11%|█         | 134/1250 [02:43<34:23,  1.85s/it, loss=0.519]
Epoch 4/4:  11%|█         | 134/1250 [02:44<34:23,  1.85s/it, loss=0.394]
Epoch 4/4:  11%|█         | 135/1250 [02:44<33:02,  1.78s/it, loss=0.394]
Epoch 4/4:  11%|█         | 135/1250 [02:46<33:02,  1.78s/it, loss=0.475]
Epoch 4/4:  11%|█         | 136/1250 [02:46<33:13,  1.79s/it, loss=0.475]
Epoch 4/4:  11%|█         | 136/1250 [02:48<33:13,  1.79s/it, loss=0.4]  
Epoch 4/4:  11%|█         | 137/1250 [02:48<33:49,  1.82s/it, loss=0.4]
Epoch 4/4:  11%|█         | 137/1250 [02:50<33:49,  1.82s/it, loss=0.517]
Epoch 4/4:  11%|█         | 138/1250 [02:50<32:36,  1.76s/it, loss=0.517]
Epoch 4/4:  11%|█         | 138/1250 [02:51<32:36,  1.76s/it, loss=0.326]
Epoch 4/4:  11%|█         | 139/1250 [02:51<31:42,  1.71s/it, loss=0.326]
Epoch 4/4:  11%|█         | 139/1250 [02:53<31:42,  1.71s/it, loss=0.459]
Epoch 4/4:  11%|█         | 140/1250 [02:53<32:36,  1.76s/it, loss=0.459]
Epoch 4/4:  11%|█         | 140/1250 [02:56<32:36,  1.76s/it, loss=0.359]
Epoch 4/4:  11%|█▏        | 141/1250 [02:56<37:29,  2.03s/it, loss=0.359]
Epoch 4/4:  11%|█▏        | 141/1250 [02:58<37:29,  2.03s/it, loss=0.465]
Epoch 4/4:  11%|█▏        | 142/1250 [02:58<38:18,  2.07s/it, loss=0.465]
Epoch 4/4:  11%|█▏        | 142/1250 [03:00<38:18,  2.07s/it, loss=0.509]
Epoch 4/4:  11%|█▏        | 143/1250 [03:00<35:58,  1.95s/it, loss=0.509]
Epoch 4/4:  11%|█▏        | 143/1250 [03:01<35:58,  1.95s/it, loss=0.445]
Epoch 4/4:  12%|█▏        | 144/1250 [03:02<36:03,  1.96s/it, loss=0.445]
Epoch 4/4:  12%|█▏        | 144/1250 [03:04<36:03,  1.96s/it, loss=0.379]
Epoch 4/4:  12%|█▏        | 145/1250 [03:04<39:16,  2.13s/it, loss=0.379]
Epoch 4/4:  12%|█▏        | 145/1250 [03:06<39:16,  2.13s/it, loss=0.548]
Epoch 4/4:  12%|█▏        | 146/1250 [03:06<35:37,  1.94s/it, loss=0.548]
Epoch 4/4:  12%|█▏        | 146/1250 [03:07<35:37,  1.94s/it, loss=0.337]
Epoch 4/4:  12%|█▏        | 147/1250 [03:07<34:40,  1.89s/it, loss=0.337]
Epoch 4/4:  12%|█▏        | 147/1250 [03:09<34:40,  1.89s/it, loss=0.4]  
Epoch 4/4:  12%|█▏        | 148/1250 [03:09<33:05,  1.80s/it, loss=0.4]
Epoch 4/4:  12%|█▏        | 148/1250 [03:11<33:05,  1.80s/it, loss=0.4]
Epoch 4/4:  12%|█▏        | 149/1250 [03:11<33:20,  1.82s/it, loss=0.4]
Epoch 4/4:  12%|█▏        | 149/1250 [03:12<33:20,  1.82s/it, loss=0.418]
Epoch 4/4:  12%|█▏        | 150/1250 [03:12<31:49,  1.74s/it, loss=0.418]
Epoch 4/4:  12%|█▏        | 150/1250 [03:15<31:49,  1.74s/it, loss=0.564]
Epoch 4/4:  12%|█▏        | 151/1250 [03:15<34:54,  1.91s/it, loss=0.564]
Epoch 4/4:  12%|█▏        | 151/1250 [03:18<34:54,  1.91s/it, loss=0.286]
Epoch 4/4:  12%|█▏        | 152/1250 [03:18<40:41,  2.22s/it, loss=0.286]
Epoch 4/4:  12%|█▏        | 152/1250 [03:21<40:41,  2.22s/it, loss=0.447]
Epoch 4/4:  12%|█▏        | 153/1250 [03:21<47:21,  2.59s/it, loss=0.447]
Epoch 4/4:  12%|█▏        | 153/1250 [03:23<47:21,  2.59s/it, loss=0.486]
Epoch 4/4:  12%|█▏        | 154/1250 [03:23<46:00,  2.52s/it, loss=0.486]
Epoch 4/4:  12%|█▏        | 154/1250 [03:25<46:00,  2.52s/it, loss=0.33] 
Epoch 4/4:  12%|█▏        | 155/1250 [03:25<41:10,  2.26s/it, loss=0.33]
Epoch 4/4:  12%|█▏        | 155/1250 [03:26<41:10,  2.26s/it, loss=0.452]
Epoch 4/4:  12%|█▏        | 156/1250 [03:26<36:56,  2.03s/it, loss=0.452]
Epoch 4/4:  12%|█▏        | 156/1250 [03:28<36:56,  2.03s/it, loss=0.385]
Epoch 4/4:  13%|█▎        | 157/1250 [03:28<33:47,  1.85s/it, loss=0.385]
Epoch 4/4:  13%|█▎        | 157/1250 [03:30<33:47,  1.85s/it, loss=0.349]
Epoch 4/4:  13%|█▎        | 158/1250 [03:30<35:13,  1.94s/it, loss=0.349]
Epoch 4/4:  13%|█▎        | 158/1250 [03:32<35:13,  1.94s/it, loss=0.494]
Epoch 4/4:  13%|█▎        | 159/1250 [03:32<33:19,  1.83s/it, loss=0.494]
Epoch 4/4:  13%|█▎        | 159/1250 [03:34<33:19,  1.83s/it, loss=0.374]
Epoch 4/4:  13%|█▎        | 160/1250 [03:34<35:20,  1.95s/it, loss=0.374]
Epoch 4/4:  13%|█▎        | 160/1250 [03:36<35:20,  1.95s/it, loss=0.311]
Epoch 4/4:  13%|█▎        | 161/1250 [03:36<36:23,  2.00s/it, loss=0.311]
Epoch 4/4:  13%|█▎        | 161/1250 [03:38<36:23,  2.00s/it, loss=0.277]
Epoch 4/4:  13%|█▎        | 162/1250 [03:38<34:59,  1.93s/it, loss=0.277]
Epoch 4/4:  13%|█▎        | 162/1250 [03:39<34:59,  1.93s/it, loss=0.319]
Epoch 4/4:  13%|█▎        | 163/1250 [03:39<33:41,  1.86s/it, loss=0.319]
Epoch 4/4:  13%|█▎        | 163/1250 [03:41<33:41,  1.86s/it, loss=0.307]
Epoch 4/4:  13%|█▎        | 164/1250 [03:41<34:34,  1.91s/it, loss=0.307]
Epoch 4/4:  13%|█▎        | 164/1250 [03:43<34:34,  1.91s/it, loss=0.39] 
Epoch 4/4:  13%|█▎        | 165/1250 [03:43<31:52,  1.76s/it, loss=0.39]
Epoch 4/4:  13%|█▎        | 165/1250 [03:46<31:52,  1.76s/it, loss=0.508]
Epoch 4/4:  13%|█▎        | 166/1250 [03:46<40:19,  2.23s/it, loss=0.508]
Epoch 4/4:  13%|█▎        | 166/1250 [03:49<40:19,  2.23s/it, loss=0.477]
Epoch 4/4:  13%|█▎        | 167/1250 [03:49<40:37,  2.25s/it, loss=0.477]
Epoch 4/4:  13%|█▎        | 167/1250 [03:51<40:37,  2.25s/it, loss=0.381]
Epoch 4/4:  13%|█▎        | 168/1250 [03:51<39:51,  2.21s/it, loss=0.381]
Epoch 4/4:  13%|█▎        | 168/1250 [03:53<39:51,  2.21s/it, loss=0.526]
Epoch 4/4:  14%|█▎        | 169/1250 [03:53<39:03,  2.17s/it, loss=0.526]
Epoch 4/4:  14%|█▎        | 169/1250 [03:54<39:03,  2.17s/it, loss=0.423]
Epoch 4/4:  14%|█▎        | 170/1250 [03:54<35:02,  1.95s/it, loss=0.423]
Epoch 4/4:  14%|█▎        | 170/1250 [03:57<35:02,  1.95s/it, loss=0.433]
Epoch 4/4:  14%|█▎        | 171/1250 [03:57<40:33,  2.25s/it, loss=0.433]
Epoch 4/4:  14%|█▎        | 171/1250 [04:01<40:33,  2.25s/it, loss=0.407]
Epoch 4/4:  14%|█▍        | 172/1250 [04:01<50:47,  2.83s/it, loss=0.407]
Epoch 4/4:  14%|█▍        | 172/1250 [04:03<50:47,  2.83s/it, loss=0.569]
Epoch 4/4:  14%|█▍        | 173/1250 [04:03<45:10,  2.52s/it, loss=0.569]
Epoch 4/4:  14%|█▍        | 173/1250 [04:10<45:10,  2.52s/it, loss=0.279]
Epoch 4/4:  14%|█▍        | 174/1250 [04:10<1:11:07,  3.97s/it, loss=0.279]
Epoch 4/4:  14%|█▍        | 174/1250 [04:16<1:11:07,  3.97s/it, loss=0.407]
Epoch 4/4:  14%|█▍        | 175/1250 [04:16<1:21:04,  4.52s/it, loss=0.407]
Epoch 4/4:  14%|█▍        | 175/1250 [04:18<1:21:04,  4.52s/it, loss=0.339]
Epoch 4/4:  14%|█▍        | 176/1250 [04:18<1:08:17,  3.81s/it, loss=0.339]
Epoch 4/4:  14%|█▍        | 176/1250 [04:21<1:08:17,  3.81s/it, loss=0.338]
Epoch 4/4:  14%|█▍        | 177/1250 [04:21<1:01:06,  3.42s/it, loss=0.338]
Epoch 4/4:  14%|█▍        | 177/1250 [04:25<1:01:06,  3.42s/it, loss=0.528]
Epoch 4/4:  14%|█▍        | 178/1250 [04:25<1:02:08,  3.48s/it, loss=0.528]
Epoch 4/4:  14%|█▍        | 178/1250 [04:27<1:02:08,  3.48s/it, loss=0.404]
Epoch 4/4:  14%|█▍        | 179/1250 [04:27<58:06,  3.26s/it, loss=0.404]  
Epoch 4/4:  14%|█▍        | 179/1250 [04:29<58:06,  3.26s/it, loss=0.326]
Epoch 4/4:  14%|█▍        | 180/1250 [04:29<52:03,  2.92s/it, loss=0.326]
Epoch 4/4:  14%|█▍        | 180/1250 [04:31<52:03,  2.92s/it, loss=0.375]
Epoch 4/4:  14%|█▍        | 181/1250 [04:31<45:11,  2.54s/it, loss=0.375]
Epoch 4/4:  14%|█▍        | 181/1250 [04:33<45:11,  2.54s/it, loss=0.527]
Epoch 4/4:  15%|█▍        | 182/1250 [04:33<43:46,  2.46s/it, loss=0.527]
Epoch 4/4:  15%|█▍        | 182/1250 [04:36<43:46,  2.46s/it, loss=0.318]
Epoch 4/4:  15%|█▍        | 183/1250 [04:36<42:40,  2.40s/it, loss=0.318]
Epoch 4/4:  15%|█▍        | 183/1250 [04:38<42:40,  2.40s/it, loss=0.532]
Epoch 4/4:  15%|█▍        | 184/1250 [04:38<43:13,  2.43s/it, loss=0.532]
Epoch 4/4:  15%|█▍        | 184/1250 [04:40<43:13,  2.43s/it, loss=0.45] 
Epoch 4/4:  15%|█▍        | 185/1250 [04:41<43:04,  2.43s/it, loss=0.45]
Epoch 4/4:  15%|█▍        | 185/1250 [04:45<43:04,  2.43s/it, loss=0.427]
Epoch 4/4:  15%|█▍        | 186/1250 [04:45<53:01,  2.99s/it, loss=0.427]
Epoch 4/4:  15%|█▍        | 186/1250 [04:46<53:01,  2.99s/it, loss=0.378]
Epoch 4/4:  15%|█▍        | 187/1250 [04:46<45:08,  2.55s/it, loss=0.378]
Epoch 4/4:  15%|█▍        | 187/1250 [04:49<45:08,  2.55s/it, loss=0.346]
Epoch 4/4:  15%|█▌        | 188/1250 [04:49<43:10,  2.44s/it, loss=0.346]
Epoch 4/4:  15%|█▌        | 188/1250 [04:51<43:10,  2.44s/it, loss=0.413]
Epoch 4/4:  15%|█▌        | 189/1250 [04:51<43:29,  2.46s/it, loss=0.413]
Epoch 4/4:  15%|█▌        | 189/1250 [04:53<43:29,  2.46s/it, loss=0.51] 
Epoch 4/4:  15%|█▌        | 190/1250 [04:53<42:45,  2.42s/it, loss=0.51]
Epoch 4/4:  15%|█▌        | 190/1250 [04:55<42:45,  2.42s/it, loss=0.406]
Epoch 4/4:  15%|█▌        | 191/1250 [04:55<39:48,  2.26s/it, loss=0.406]
Epoch 4/4:  15%|█▌        | 191/1250 [04:57<39:48,  2.26s/it, loss=0.411]
Epoch 4/4:  15%|█▌        | 192/1250 [04:57<38:37,  2.19s/it, loss=0.411]
Epoch 4/4:  15%|█▌        | 192/1250 [05:00<38:37,  2.19s/it, loss=0.375]
Epoch 4/4:  15%|█▌        | 193/1250 [05:00<40:07,  2.28s/it, loss=0.375]
Epoch 4/4:  15%|█▌        | 193/1250 [05:02<40:07,  2.28s/it, loss=0.24] 
Epoch 4/4:  16%|█▌        | 194/1250 [05:02<40:15,  2.29s/it, loss=0.24]
Epoch 4/4:  16%|█▌        | 194/1250 [05:04<40:15,  2.29s/it, loss=0.333]
Epoch 4/4:  16%|█▌        | 195/1250 [05:04<39:35,  2.25s/it, loss=0.333]
Epoch 4/4:  16%|█▌        | 195/1250 [05:06<39:35,  2.25s/it, loss=0.506]
Epoch 4/4:  16%|█▌        | 196/1250 [05:06<38:54,  2.21s/it, loss=0.506]
Epoch 4/4:  16%|█▌        | 196/1250 [05:09<38:54,  2.21s/it, loss=0.347]
Epoch 4/4:  16%|█▌        | 197/1250 [05:09<42:09,  2.40s/it, loss=0.347]
Epoch 4/4:  16%|█▌        | 197/1250 [05:13<42:09,  2.40s/it, loss=0.389]
Epoch 4/4:  16%|█▌        | 198/1250 [05:13<51:07,  2.92s/it, loss=0.389]
Epoch 4/4:  16%|█▌        | 198/1250 [05:18<51:07,  2.92s/it, loss=0.389]
Epoch 4/4:  16%|█▌        | 199/1250 [05:18<1:01:09,  3.49s/it, loss=0.389]
Epoch 4/4:  16%|█▌        | 199/1250 [05:21<1:01:09,  3.49s/it, loss=0.449]
Epoch 4/4:  16%|█▌        | 200/1250 [05:21<56:42,  3.24s/it, loss=0.449]  
Epoch 4/4:  16%|█▌        | 200/1250 [05:23<56:42,  3.24s/it, loss=0.359]
Epoch 4/4:  16%|█▌        | 201/1250 [05:23<50:49,  2.91s/it, loss=0.359]
Epoch 4/4:  16%|█▌        | 201/1250 [05:24<50:49,  2.91s/it, loss=0.279]
Epoch 4/4:  16%|█▌        | 202/1250 [05:24<43:07,  2.47s/it, loss=0.279]
Epoch 4/4:  16%|█▌        | 202/1250 [05:27<43:07,  2.47s/it, loss=0.35] 
Epoch 4/4:  16%|█▌        | 203/1250 [05:27<42:14,  2.42s/it, loss=0.35]
Epoch 4/4:  16%|█▌        | 203/1250 [05:28<42:14,  2.42s/it, loss=0.392]
Epoch 4/4:  16%|█▋        | 204/1250 [05:28<37:39,  2.16s/it, loss=0.392]
Epoch 4/4:  16%|█▋        | 204/1250 [05:30<37:39,  2.16s/it, loss=0.443]
Epoch 4/4:  16%|█▋        | 205/1250 [05:30<34:16,  1.97s/it, loss=0.443]
Epoch 4/4:  16%|█▋        | 205/1250 [05:31<34:16,  1.97s/it, loss=0.467]
Epoch 4/4:  16%|█▋        | 206/1250 [05:31<33:08,  1.90s/it, loss=0.467]
Epoch 4/4:  16%|█▋        | 206/1250 [05:33<33:08,  1.90s/it, loss=0.271]
Epoch 4/4:  17%|█▋        | 207/1250 [05:33<30:55,  1.78s/it, loss=0.271]
Epoch 4/4:  17%|█▋        | 207/1250 [05:34<30:55,  1.78s/it, loss=0.312]
Epoch 4/4:  17%|█▋        | 208/1250 [05:34<29:25,  1.69s/it, loss=0.312]
Epoch 4/4:  17%|█▋        | 208/1250 [05:36<29:25,  1.69s/it, loss=0.38] 
Epoch 4/4:  17%|█▋        | 209/1250 [05:36<29:22,  1.69s/it, loss=0.38]
Epoch 4/4:  17%|█▋        | 209/1250 [05:38<29:22,  1.69s/it, loss=0.412]
Epoch 4/4:  17%|█▋        | 210/1250 [05:38<28:12,  1.63s/it, loss=0.412]
Epoch 4/4:  17%|█▋        | 210/1250 [05:39<28:12,  1.63s/it, loss=0.453]
Epoch 4/4:  17%|█▋        | 211/1250 [05:39<27:32,  1.59s/it, loss=0.453]
Epoch 4/4:  17%|█▋        | 211/1250 [05:41<27:32,  1.59s/it, loss=0.381]
Epoch 4/4:  17%|█▋        | 212/1250 [05:41<29:41,  1.72s/it, loss=0.381]
Epoch 4/4:  17%|█▋        | 212/1250 [05:43<29:41,  1.72s/it, loss=0.364]
Epoch 4/4:  17%|█▋        | 213/1250 [05:43<30:25,  1.76s/it, loss=0.364]
Epoch 4/4:  17%|█▋        | 213/1250 [05:45<30:25,  1.76s/it, loss=0.34] 
Epoch 4/4:  17%|█▋        | 214/1250 [05:45<29:57,  1.73s/it, loss=0.34]
Epoch 4/4:  17%|█▋        | 214/1250 [05:47<29:57,  1.73s/it, loss=0.498]
Epoch 4/4:  17%|█▋        | 215/1250 [05:47<31:57,  1.85s/it, loss=0.498]
Epoch 4/4:  17%|█▋        | 215/1250 [05:49<31:57,  1.85s/it, loss=0.381]
Epoch 4/4:  17%|█▋        | 216/1250 [05:49<32:26,  1.88s/it, loss=0.381]
Epoch 4/4:  17%|█▋        | 216/1250 [05:51<32:26,  1.88s/it, loss=0.812]
Epoch 4/4:  17%|█▋        | 217/1250 [05:51<33:36,  1.95s/it, loss=0.812]
Epoch 4/4:  17%|█▋        | 217/1250 [05:53<33:36,  1.95s/it, loss=0.363]
Epoch 4/4:  17%|█▋        | 218/1250 [05:53<35:43,  2.08s/it, loss=0.363]
Epoch 4/4:  17%|█▋        | 218/1250 [05:55<35:43,  2.08s/it, loss=0.39] 
Epoch 4/4:  18%|█▊        | 219/1250 [05:55<33:04,  1.93s/it, loss=0.39]
Epoch 4/4:  18%|█▊        | 219/1250 [05:56<33:04,  1.93s/it, loss=0.412]
Epoch 4/4:  18%|█▊        | 220/1250 [05:56<31:13,  1.82s/it, loss=0.412]
Epoch 4/4:  18%|█▊        | 220/1250 [05:58<31:13,  1.82s/it, loss=0.346]
Epoch 4/4:  18%|█▊        | 221/1250 [05:58<29:44,  1.73s/it, loss=0.346]
Epoch 4/4:  18%|█▊        | 221/1250 [05:59<29:44,  1.73s/it, loss=0.473]
Epoch 4/4:  18%|█▊        | 222/1250 [05:59<28:39,  1.67s/it, loss=0.473]
Epoch 4/4:  18%|█▊        | 222/1250 [06:02<28:39,  1.67s/it, loss=0.505]
Epoch 4/4:  18%|█▊        | 223/1250 [06:02<33:41,  1.97s/it, loss=0.505]
Epoch 4/4:  18%|█▊        | 223/1250 [06:04<33:41,  1.97s/it, loss=0.605]
Epoch 4/4:  18%|█▊        | 224/1250 [06:04<31:52,  1.86s/it, loss=0.605]
Epoch 4/4:  18%|█▊        | 224/1250 [06:05<31:52,  1.86s/it, loss=0.31] 
Epoch 4/4:  18%|█▊        | 225/1250 [06:05<30:07,  1.76s/it, loss=0.31]
Epoch 4/4:  18%|█▊        | 225/1250 [06:07<30:07,  1.76s/it, loss=0.5] 
Epoch 4/4:  18%|█▊        | 226/1250 [06:07<30:23,  1.78s/it, loss=0.5]
Epoch 4/4:  18%|█▊        | 226/1250 [06:09<30:23,  1.78s/it, loss=0.351]
Epoch 4/4:  18%|█▊        | 227/1250 [06:09<32:48,  1.92s/it, loss=0.351]
Epoch 4/4:  18%|█▊        | 227/1250 [06:12<32:48,  1.92s/it, loss=0.322]
Epoch 4/4:  18%|█▊        | 228/1250 [06:12<34:25,  2.02s/it, loss=0.322]
Epoch 4/4:  18%|█▊        | 228/1250 [06:14<34:25,  2.02s/it, loss=0.457]
Epoch 4/4:  18%|█▊        | 229/1250 [06:14<34:32,  2.03s/it, loss=0.457]
Epoch 4/4:  18%|█▊        | 229/1250 [06:16<34:32,  2.03s/it, loss=0.465]
Epoch 4/4:  18%|█▊        | 230/1250 [06:16<34:27,  2.03s/it, loss=0.465]
Epoch 4/4:  18%|█▊        | 230/1250 [06:18<34:27,  2.03s/it, loss=0.321]
Epoch 4/4:  18%|█▊        | 231/1250 [06:18<34:31,  2.03s/it, loss=0.321]
Epoch 4/4:  18%|█▊        | 231/1250 [06:20<34:31,  2.03s/it, loss=0.409]
Epoch 4/4:  19%|█▊        | 232/1250 [06:20<34:44,  2.05s/it, loss=0.409]
Epoch 4/4:  19%|█▊        | 232/1250 [06:22<34:44,  2.05s/it, loss=0.438]
Epoch 4/4:  19%|█▊        | 233/1250 [06:22<34:50,  2.06s/it, loss=0.438]
Epoch 4/4:  19%|█▊        | 233/1250 [06:24<34:50,  2.06s/it, loss=0.38] 
Epoch 4/4:  19%|█▊        | 234/1250 [06:24<35:32,  2.10s/it, loss=0.38]
Epoch 4/4:  19%|█▊        | 234/1250 [06:27<35:32,  2.10s/it, loss=0.448]
Epoch 4/4:  19%|█▉        | 235/1250 [06:27<38:40,  2.29s/it, loss=0.448]
Epoch 4/4:  19%|█▉        | 235/1250 [06:31<38:40,  2.29s/it, loss=0.559]
Epoch 4/4:  19%|█▉        | 236/1250 [06:31<47:46,  2.83s/it, loss=0.559]
Epoch 4/4:  19%|█▉        | 236/1250 [06:40<47:46,  2.83s/it, loss=0.461]
Epoch 4/4:  19%|█▉        | 237/1250 [06:40<1:18:32,  4.65s/it, loss=0.461]
Epoch 4/4:  19%|█▉        | 237/1250 [06:42<1:18:32,  4.65s/it, loss=0.442]
Epoch 4/4:  19%|█▉        | 238/1250 [06:42<1:04:16,  3.81s/it, loss=0.442]
Epoch 4/4:  19%|█▉        | 238/1250 [06:43<1:04:16,  3.81s/it, loss=0.501]
Epoch 4/4:  19%|█▉        | 239/1250 [06:43<53:45,  3.19s/it, loss=0.501]  
Epoch 4/4:  19%|█▉        | 239/1250 [06:45<53:45,  3.19s/it, loss=0.436]
Epoch 4/4:  19%|█▉        | 240/1250 [06:45<45:09,  2.68s/it, loss=0.436]
Epoch 4/4:  19%|█▉        | 240/1250 [06:47<45:09,  2.68s/it, loss=0.456]
Epoch 4/4:  19%|█▉        | 241/1250 [06:47<39:43,  2.36s/it, loss=0.456]
Epoch 4/4:  19%|█▉        | 241/1250 [06:49<39:43,  2.36s/it, loss=0.367]
Epoch 4/4:  19%|█▉        | 242/1250 [06:49<42:15,  2.52s/it, loss=0.367]
Epoch 4/4:  19%|█▉        | 242/1250 [06:51<42:15,  2.52s/it, loss=0.315]
Epoch 4/4:  19%|█▉        | 243/1250 [06:51<37:30,  2.23s/it, loss=0.315]
Epoch 4/4:  19%|█▉        | 243/1250 [06:52<37:30,  2.23s/it, loss=0.451]
Epoch 4/4:  20%|█▉        | 244/1250 [06:52<33:37,  2.01s/it, loss=0.451]
Epoch 4/4:  20%|█▉        | 244/1250 [06:54<33:37,  2.01s/it, loss=0.465]
Epoch 4/4:  20%|█▉        | 245/1250 [06:54<31:09,  1.86s/it, loss=0.465]
Epoch 4/4:  20%|█▉        | 245/1250 [06:56<31:09,  1.86s/it, loss=0.411]
Epoch 4/4:  20%|█▉        | 246/1250 [06:56<29:36,  1.77s/it, loss=0.411]
Epoch 4/4:  20%|█▉        | 246/1250 [06:57<29:36,  1.77s/it, loss=0.339]
Epoch 4/4:  20%|█▉        | 247/1250 [06:57<28:24,  1.70s/it, loss=0.339]
Epoch 4/4:  20%|█▉        | 247/1250 [06:59<28:24,  1.70s/it, loss=0.285]
Epoch 4/4:  20%|█▉        | 248/1250 [06:59<29:10,  1.75s/it, loss=0.285]
Epoch 4/4:  20%|█▉        | 248/1250 [07:00<29:10,  1.75s/it, loss=0.476]
Epoch 4/4:  20%|█▉        | 249/1250 [07:00<28:25,  1.70s/it, loss=0.476]
Epoch 4/4:  20%|█▉        | 249/1250 [07:02<28:25,  1.70s/it, loss=0.345]
Epoch 4/4:  20%|██        | 250/1250 [07:02<27:30,  1.65s/it, loss=0.345]
Epoch 4/4:  20%|██        | 250/1250 [07:04<27:30,  1.65s/it, loss=0.449]
Epoch 4/4:  20%|██        | 251/1250 [07:04<27:17,  1.64s/it, loss=0.449]
Epoch 4/4:  20%|██        | 251/1250 [07:05<27:17,  1.64s/it, loss=0.436]
Epoch 4/4:  20%|██        | 252/1250 [07:05<26:49,  1.61s/it, loss=0.436]
Epoch 4/4:  20%|██        | 252/1250 [07:07<26:49,  1.61s/it, loss=0.376]
Epoch 4/4:  20%|██        | 253/1250 [07:07<29:58,  1.80s/it, loss=0.376]
Epoch 4/4:  20%|██        | 253/1250 [07:10<29:58,  1.80s/it, loss=0.372]
Epoch 4/4:  20%|██        | 254/1250 [07:10<32:26,  1.95s/it, loss=0.372]
Epoch 4/4:  20%|██        | 254/1250 [07:12<32:26,  1.95s/it, loss=0.796]
Epoch 4/4:  20%|██        | 255/1250 [07:12<33:28,  2.02s/it, loss=0.796]
Epoch 4/4:  20%|██        | 255/1250 [07:15<33:28,  2.02s/it, loss=0.464]
Epoch 4/4:  20%|██        | 256/1250 [07:15<36:38,  2.21s/it, loss=0.464]
Epoch 4/4:  20%|██        | 256/1250 [07:17<36:38,  2.21s/it, loss=0.445]
Epoch 4/4:  21%|██        | 257/1250 [07:17<36:41,  2.22s/it, loss=0.445]
Epoch 4/4:  21%|██        | 257/1250 [07:19<36:41,  2.22s/it, loss=0.52] 
Epoch 4/4:  21%|██        | 258/1250 [07:19<36:02,  2.18s/it, loss=0.52]
Epoch 4/4:  21%|██        | 258/1250 [07:21<36:02,  2.18s/it, loss=0.349]
Epoch 4/4:  21%|██        | 259/1250 [07:21<35:48,  2.17s/it, loss=0.349]
Epoch 4/4:  21%|██        | 259/1250 [07:23<35:48,  2.17s/it, loss=0.317]
Epoch 4/4:  21%|██        | 260/1250 [07:23<35:09,  2.13s/it, loss=0.317]
Epoch 4/4:  21%|██        | 260/1250 [07:25<35:09,  2.13s/it, loss=0.357]
Epoch 4/4:  21%|██        | 261/1250 [07:25<34:12,  2.08s/it, loss=0.357]
Epoch 4/4:  21%|██        | 261/1250 [07:27<34:12,  2.08s/it, loss=0.456]
Epoch 4/4:  21%|██        | 262/1250 [07:27<33:10,  2.01s/it, loss=0.456]
Epoch 4/4:  21%|██        | 262/1250 [07:29<33:10,  2.01s/it, loss=0.275]
Epoch 4/4:  21%|██        | 263/1250 [07:29<32:38,  1.98s/it, loss=0.275]
Epoch 4/4:  21%|██        | 263/1250 [07:31<32:38,  1.98s/it, loss=0.419]
Epoch 4/4:  21%|██        | 264/1250 [07:31<31:33,  1.92s/it, loss=0.419]
Epoch 4/4:  21%|██        | 264/1250 [07:32<31:33,  1.92s/it, loss=0.333]
Epoch 4/4:  21%|██        | 265/1250 [07:32<31:23,  1.91s/it, loss=0.333]
Epoch 4/4:  21%|██        | 265/1250 [07:35<31:23,  1.91s/it, loss=0.26] 
Epoch 4/4:  21%|██▏       | 266/1250 [07:35<36:20,  2.22s/it, loss=0.26]
Epoch 4/4:  21%|██▏       | 266/1250 [07:37<36:20,  2.22s/it, loss=0.358]
Epoch 4/4:  21%|██▏       | 267/1250 [07:37<34:12,  2.09s/it, loss=0.358]
Epoch 4/4:  21%|██▏       | 267/1250 [07:39<34:12,  2.09s/it, loss=0.408]
Epoch 4/4:  21%|██▏       | 268/1250 [07:39<30:30,  1.86s/it, loss=0.408]
Epoch 4/4:  21%|██▏       | 268/1250 [07:40<30:30,  1.86s/it, loss=0.422]
Epoch 4/4:  22%|██▏       | 269/1250 [07:40<29:03,  1.78s/it, loss=0.422]
Epoch 4/4:  22%|██▏       | 269/1250 [07:42<29:03,  1.78s/it, loss=0.518]
Epoch 4/4:  22%|██▏       | 270/1250 [07:42<31:40,  1.94s/it, loss=0.518]
Epoch 4/4:  22%|██▏       | 270/1250 [07:45<31:40,  1.94s/it, loss=0.38] 
Epoch 4/4:  22%|██▏       | 271/1250 [07:45<34:41,  2.13s/it, loss=0.38]
Epoch 4/4:  22%|██▏       | 271/1250 [07:48<34:41,  2.13s/it, loss=0.352]
Epoch 4/4:  22%|██▏       | 272/1250 [07:48<38:38,  2.37s/it, loss=0.352]
Epoch 4/4:  22%|██▏       | 272/1250 [07:50<38:38,  2.37s/it, loss=0.365]
Epoch 4/4:  22%|██▏       | 273/1250 [07:50<38:55,  2.39s/it, loss=0.365]
Epoch 4/4:  22%|██▏       | 273/1250 [07:53<38:55,  2.39s/it, loss=0.262]
Epoch 4/4:  22%|██▏       | 274/1250 [07:53<39:44,  2.44s/it, loss=0.262]
Epoch 4/4:  22%|██▏       | 274/1250 [07:55<39:44,  2.44s/it, loss=0.422]
Epoch 4/4:  22%|██▏       | 275/1250 [07:55<38:33,  2.37s/it, loss=0.422]
Epoch 4/4:  22%|██▏       | 275/1250 [07:58<38:33,  2.37s/it, loss=0.513]
Epoch 4/4:  22%|██▏       | 276/1250 [07:58<42:32,  2.62s/it, loss=0.513]
Epoch 4/4:  22%|██▏       | 276/1250 [08:01<42:32,  2.62s/it, loss=0.432]
Epoch 4/4:  22%|██▏       | 277/1250 [08:01<43:10,  2.66s/it, loss=0.432]
Epoch 4/4:  22%|██▏       | 277/1250 [08:07<43:10,  2.66s/it, loss=0.515]
Epoch 4/4:  22%|██▏       | 278/1250 [08:07<1:00:51,  3.76s/it, loss=0.515]
Epoch 4/4:  22%|██▏       | 278/1250 [08:10<1:00:51,  3.76s/it, loss=0.466]
Epoch 4/4:  22%|██▏       | 279/1250 [08:10<53:57,  3.33s/it, loss=0.466]  
Epoch 4/4:  22%|██▏       | 279/1250 [08:11<53:57,  3.33s/it, loss=0.42] 
Epoch 4/4:  22%|██▏       | 280/1250 [08:11<44:40,  2.76s/it, loss=0.42]
Epoch 4/4:  22%|██▏       | 280/1250 [08:13<44:40,  2.76s/it, loss=0.458]
Epoch 4/4:  22%|██▏       | 281/1250 [08:13<38:53,  2.41s/it, loss=0.458]
Epoch 4/4:  22%|██▏       | 281/1250 [08:14<38:53,  2.41s/it, loss=0.454]
Epoch 4/4:  23%|██▎       | 282/1250 [08:14<34:55,  2.17s/it, loss=0.454]
Epoch 4/4:  23%|██▎       | 282/1250 [08:17<34:55,  2.17s/it, loss=0.465]
Epoch 4/4:  23%|██▎       | 283/1250 [08:17<34:51,  2.16s/it, loss=0.465]
Epoch 4/4:  23%|██▎       | 283/1250 [08:18<34:51,  2.16s/it, loss=0.321]
Epoch 4/4:  23%|██▎       | 284/1250 [08:18<33:29,  2.08s/it, loss=0.321]
Epoch 4/4:  23%|██▎       | 284/1250 [08:20<33:29,  2.08s/it, loss=0.441]
Epoch 4/4:  23%|██▎       | 285/1250 [08:20<32:47,  2.04s/it, loss=0.441]
Epoch 4/4:  23%|██▎       | 285/1250 [08:22<32:47,  2.04s/it, loss=0.586]
Epoch 4/4:  23%|██▎       | 286/1250 [08:22<32:57,  2.05s/it, loss=0.586]
Epoch 4/4:  23%|██▎       | 286/1250 [08:24<32:57,  2.05s/it, loss=0.381]
Epoch 4/4:  23%|██▎       | 287/1250 [08:24<30:10,  1.88s/it, loss=0.381]
Epoch 4/4:  23%|██▎       | 287/1250 [08:26<30:10,  1.88s/it, loss=0.44] 
Epoch 4/4:  23%|██▎       | 288/1250 [08:26<29:03,  1.81s/it, loss=0.44]
Epoch 4/4:  23%|██▎       | 288/1250 [08:27<29:03,  1.81s/it, loss=0.285]
Epoch 4/4:  23%|██▎       | 289/1250 [08:27<29:21,  1.83s/it, loss=0.285]
Epoch 4/4:  23%|██▎       | 289/1250 [08:29<29:21,  1.83s/it, loss=0.409]
Epoch 4/4:  23%|██▎       | 290/1250 [08:29<30:13,  1.89s/it, loss=0.409]
Epoch 4/4:  23%|██▎       | 290/1250 [08:31<30:13,  1.89s/it, loss=0.275]
Epoch 4/4:  23%|██▎       | 291/1250 [08:31<30:50,  1.93s/it, loss=0.275]
Epoch 4/4:  23%|██▎       | 291/1250 [08:33<30:50,  1.93s/it, loss=0.42] 
Epoch 4/4:  23%|██▎       | 292/1250 [08:33<29:36,  1.85s/it, loss=0.42]
Epoch 4/4:  23%|██▎       | 292/1250 [08:35<29:36,  1.85s/it, loss=0.27]
Epoch 4/4:  23%|██▎       | 293/1250 [08:35<29:36,  1.86s/it, loss=0.27]
Epoch 4/4:  23%|██▎       | 293/1250 [08:37<29:36,  1.86s/it, loss=0.425]
Epoch 4/4:  24%|██▎       | 294/1250 [08:37<29:52,  1.87s/it, loss=0.425]
Epoch 4/4:  24%|██▎       | 294/1250 [08:39<29:52,  1.87s/it, loss=0.332]
Epoch 4/4:  24%|██▎       | 295/1250 [08:39<30:03,  1.89s/it, loss=0.332]
Epoch 4/4:  24%|██▎       | 295/1250 [08:41<30:03,  1.89s/it, loss=0.43] 
Epoch 4/4:  24%|██▎       | 296/1250 [08:41<30:08,  1.90s/it, loss=0.43]
Epoch 4/4:  24%|██▎       | 296/1250 [08:42<30:08,  1.90s/it, loss=0.428]
Epoch 4/4:  24%|██▍       | 297/1250 [08:42<28:53,  1.82s/it, loss=0.428]
Epoch 4/4:  24%|██▍       | 297/1250 [08:44<28:53,  1.82s/it, loss=0.547]
Epoch 4/4:  24%|██▍       | 298/1250 [08:44<29:46,  1.88s/it, loss=0.547]
Epoch 4/4:  24%|██▍       | 298/1250 [08:46<29:46,  1.88s/it, loss=0.431]
Epoch 4/4:  24%|██▍       | 299/1250 [08:46<29:14,  1.85s/it, loss=0.431]
Epoch 4/4:  24%|██▍       | 299/1250 [08:48<29:14,  1.85s/it, loss=0.392]
Epoch 4/4:  24%|██▍       | 300/1250 [08:48<27:41,  1.75s/it, loss=0.392]
Epoch 4/4:  24%|██▍       | 300/1250 [08:50<27:41,  1.75s/it, loss=0.564]
Epoch 4/4:  24%|██▍       | 301/1250 [08:50<27:57,  1.77s/it, loss=0.564]
Epoch 4/4:  24%|██▍       | 301/1250 [08:51<27:57,  1.77s/it, loss=0.535]
Epoch 4/4:  24%|██▍       | 302/1250 [08:51<28:35,  1.81s/it, loss=0.535]
Epoch 4/4:  24%|██▍       | 302/1250 [08:53<28:35,  1.81s/it, loss=0.381]
Epoch 4/4:  24%|██▍       | 303/1250 [08:53<28:27,  1.80s/it, loss=0.381]
Epoch 4/4:  24%|██▍       | 303/1250 [08:55<28:27,  1.80s/it, loss=0.503]
Epoch 4/4:  24%|██▍       | 304/1250 [08:55<27:18,  1.73s/it, loss=0.503]
Epoch 4/4:  24%|██▍       | 304/1250 [08:56<27:18,  1.73s/it, loss=0.487]
Epoch 4/4:  24%|██▍       | 305/1250 [08:56<26:39,  1.69s/it, loss=0.487]
Epoch 4/4:  24%|██▍       | 305/1250 [08:58<26:39,  1.69s/it, loss=0.473]
Epoch 4/4:  24%|██▍       | 306/1250 [08:58<27:51,  1.77s/it, loss=0.473]
Epoch 4/4:  24%|██▍       | 306/1250 [09:00<27:51,  1.77s/it, loss=0.468]
Epoch 4/4:  25%|██▍       | 307/1250 [09:00<28:05,  1.79s/it, loss=0.468]
Epoch 4/4:  25%|██▍       | 307/1250 [09:02<28:05,  1.79s/it, loss=0.426]
Epoch 4/4:  25%|██▍       | 308/1250 [09:02<27:20,  1.74s/it, loss=0.426]
Epoch 4/4:  25%|██▍       | 308/1250 [09:04<27:20,  1.74s/it, loss=0.492]
Epoch 4/4:  25%|██▍       | 309/1250 [09:04<27:54,  1.78s/it, loss=0.492]
Epoch 4/4:  25%|██▍       | 309/1250 [09:06<27:54,  1.78s/it, loss=0.455]
Epoch 4/4:  25%|██▍       | 310/1250 [09:06<28:17,  1.81s/it, loss=0.455]
Epoch 4/4:  25%|██▍       | 310/1250 [09:07<28:17,  1.81s/it, loss=0.364]
Epoch 4/4:  25%|██▍       | 311/1250 [09:07<27:31,  1.76s/it, loss=0.364]
Epoch 4/4:  25%|██▍       | 311/1250 [09:09<27:31,  1.76s/it, loss=0.435]
Epoch 4/4:  25%|██▍       | 312/1250 [09:09<29:05,  1.86s/it, loss=0.435]
Epoch 4/4:  25%|██▍       | 312/1250 [09:11<29:05,  1.86s/it, loss=0.337]
Epoch 4/4:  25%|██▌       | 313/1250 [09:11<29:28,  1.89s/it, loss=0.337]
Epoch 4/4:  25%|██▌       | 313/1250 [09:13<29:28,  1.89s/it, loss=0.445]
Epoch 4/4:  25%|██▌       | 314/1250 [09:13<28:24,  1.82s/it, loss=0.445]
Epoch 4/4:  25%|██▌       | 314/1250 [09:14<28:24,  1.82s/it, loss=0.263]
Epoch 4/4:  25%|██▌       | 315/1250 [09:14<27:04,  1.74s/it, loss=0.263]
Epoch 4/4:  25%|██▌       | 315/1250 [09:16<27:04,  1.74s/it, loss=0.337]
Epoch 4/4:  25%|██▌       | 316/1250 [09:16<27:42,  1.78s/it, loss=0.337]
Epoch 4/4:  25%|██▌       | 316/1250 [09:19<27:42,  1.78s/it, loss=0.374]
Epoch 4/4:  25%|██▌       | 317/1250 [09:19<29:40,  1.91s/it, loss=0.374]
Epoch 4/4:  25%|██▌       | 317/1250 [09:22<29:40,  1.91s/it, loss=0.44] 
Epoch 4/4:  25%|██▌       | 318/1250 [09:22<34:36,  2.23s/it, loss=0.44]
Epoch 4/4:  25%|██▌       | 318/1250 [09:24<34:36,  2.23s/it, loss=0.437]
Epoch 4/4:  26%|██▌       | 319/1250 [09:24<36:35,  2.36s/it, loss=0.437]
Epoch 4/4:  26%|██▌       | 319/1250 [09:28<36:35,  2.36s/it, loss=0.494]
Epoch 4/4:  26%|██▌       | 320/1250 [09:28<42:47,  2.76s/it, loss=0.494]
Epoch 4/4:  26%|██▌       | 320/1250 [09:36<42:47,  2.76s/it, loss=0.473]
Epoch 4/4:  26%|██▌       | 321/1250 [09:36<1:06:43,  4.31s/it, loss=0.473]
Epoch 4/4:  26%|██▌       | 321/1250 [09:38<1:06:43,  4.31s/it, loss=0.318]
Epoch 4/4:  26%|██▌       | 322/1250 [09:38<57:58,  3.75s/it, loss=0.318]  
Epoch 4/4:  26%|██▌       | 322/1250 [09:40<57:58,  3.75s/it, loss=0.474]
Epoch 4/4:  26%|██▌       | 323/1250 [09:40<49:26,  3.20s/it, loss=0.474]
Epoch 4/4:  26%|██▌       | 323/1250 [09:42<49:26,  3.20s/it, loss=0.447]
Epoch 4/4:  26%|██▌       | 324/1250 [09:42<45:15,  2.93s/it, loss=0.447]
Epoch 4/4:  26%|██▌       | 324/1250 [09:45<45:15,  2.93s/it, loss=0.447]
Epoch 4/4:  26%|██▌       | 325/1250 [09:45<45:37,  2.96s/it, loss=0.447]
Epoch 4/4:  26%|██▌       | 325/1250 [09:49<45:37,  2.96s/it, loss=0.455]
Epoch 4/4:  26%|██▌       | 326/1250 [09:49<49:12,  3.19s/it, loss=0.455]
Epoch 4/4:  26%|██▌       | 326/1250 [09:52<49:12,  3.19s/it, loss=0.407]
Epoch 4/4:  26%|██▌       | 327/1250 [09:52<45:01,  2.93s/it, loss=0.407]
Epoch 4/4:  26%|██▌       | 327/1250 [09:54<45:01,  2.93s/it, loss=0.423]
Epoch 4/4:  26%|██▌       | 328/1250 [09:54<41:57,  2.73s/it, loss=0.423]
Epoch 4/4:  26%|██▌       | 328/1250 [09:56<41:57,  2.73s/it, loss=0.408]
Epoch 4/4:  26%|██▋       | 329/1250 [09:56<40:47,  2.66s/it, loss=0.408]
Epoch 4/4:  26%|██▋       | 329/1250 [10:07<40:47,  2.66s/it, loss=0.279]
Epoch 4/4:  26%|██▋       | 330/1250 [10:07<1:19:24,  5.18s/it, loss=0.279]
Epoch 4/4:  26%|██▋       | 330/1250 [10:10<1:19:24,  5.18s/it, loss=0.346]
Epoch 4/4:  26%|██▋       | 331/1250 [10:10<1:06:14,  4.32s/it, loss=0.346]
Epoch 4/4:  26%|██▋       | 331/1250 [10:11<1:06:14,  4.32s/it, loss=0.345]
Epoch 4/4:  27%|██▋       | 332/1250 [10:11<54:22,  3.55s/it, loss=0.345]  
Epoch 4/4:  27%|██▋       | 332/1250 [10:13<54:22,  3.55s/it, loss=0.54] 
Epoch 4/4:  27%|██▋       | 333/1250 [10:13<45:07,  2.95s/it, loss=0.54]
Epoch 4/4:  27%|██▋       | 333/1250 [10:15<45:07,  2.95s/it, loss=0.454]
Epoch 4/4:  27%|██▋       | 334/1250 [10:15<39:07,  2.56s/it, loss=0.454]
Epoch 4/4:  27%|██▋       | 334/1250 [10:17<39:07,  2.56s/it, loss=0.367]
Epoch 4/4:  27%|██▋       | 335/1250 [10:17<35:59,  2.36s/it, loss=0.367]
Epoch 4/4:  27%|██▋       | 335/1250 [10:18<35:59,  2.36s/it, loss=0.451]
Epoch 4/4:  27%|██▋       | 336/1250 [10:18<33:06,  2.17s/it, loss=0.451]
Epoch 4/4:  27%|██▋       | 336/1250 [10:20<33:06,  2.17s/it, loss=0.348]
Epoch 4/4:  27%|██▋       | 337/1250 [10:20<31:26,  2.07s/it, loss=0.348]
Epoch 4/4:  27%|██▋       | 337/1250 [10:22<31:26,  2.07s/it, loss=0.428]
Epoch 4/4:  27%|██▋       | 338/1250 [10:22<31:10,  2.05s/it, loss=0.428]
Epoch 4/4:  27%|██▋       | 338/1250 [10:24<31:10,  2.05s/it, loss=0.48] 
Epoch 4/4:  27%|██▋       | 339/1250 [10:24<29:51,  1.97s/it, loss=0.48]
Epoch 4/4:  27%|██▋       | 339/1250 [10:26<29:51,  1.97s/it, loss=0.311]
Epoch 4/4:  27%|██▋       | 340/1250 [10:26<29:12,  1.93s/it, loss=0.311]
Epoch 4/4:  27%|██▋       | 340/1250 [10:27<29:12,  1.93s/it, loss=0.443]
Epoch 4/4:  27%|██▋       | 341/1250 [10:27<27:27,  1.81s/it, loss=0.443]
Epoch 4/4:  27%|██▋       | 341/1250 [10:29<27:27,  1.81s/it, loss=0.373]
Epoch 4/4:  27%|██▋       | 342/1250 [10:29<28:03,  1.85s/it, loss=0.373]
Epoch 4/4:  27%|██▋       | 342/1250 [10:31<28:03,  1.85s/it, loss=0.353]
Epoch 4/4:  27%|██▋       | 343/1250 [10:31<28:13,  1.87s/it, loss=0.353]
Epoch 4/4:  27%|██▋       | 343/1250 [10:33<28:13,  1.87s/it, loss=0.349]
Epoch 4/4:  28%|██▊       | 344/1250 [10:33<27:43,  1.84s/it, loss=0.349]
Epoch 4/4:  28%|██▊       | 344/1250 [10:35<27:43,  1.84s/it, loss=0.563]
Epoch 4/4:  28%|██▊       | 345/1250 [10:35<27:56,  1.85s/it, loss=0.563]
Epoch 4/4:  28%|██▊       | 345/1250 [10:36<27:56,  1.85s/it, loss=0.326]
Epoch 4/4:  28%|██▊       | 346/1250 [10:36<26:13,  1.74s/it, loss=0.326]
Epoch 4/4:  28%|██▊       | 346/1250 [10:38<26:13,  1.74s/it, loss=0.488]
Epoch 4/4:  28%|██▊       | 347/1250 [10:38<26:56,  1.79s/it, loss=0.488]
Epoch 4/4:  28%|██▊       | 347/1250 [10:40<26:56,  1.79s/it, loss=0.452]
Epoch 4/4:  28%|██▊       | 348/1250 [10:40<28:09,  1.87s/it, loss=0.452]
Epoch 4/4:  28%|██▊       | 348/1250 [10:42<28:09,  1.87s/it, loss=0.266]
Epoch 4/4:  28%|██▊       | 349/1250 [10:42<29:12,  1.95s/it, loss=0.266]
Epoch 4/4:  28%|██▊       | 349/1250 [10:44<29:12,  1.95s/it, loss=0.473]
Epoch 4/4:  28%|██▊       | 350/1250 [10:44<27:10,  1.81s/it, loss=0.473]
Epoch 4/4:  28%|██▊       | 350/1250 [10:45<27:10,  1.81s/it, loss=0.486]
Epoch 4/4:  28%|██▊       | 351/1250 [10:45<25:55,  1.73s/it, loss=0.486]
Epoch 4/4:  28%|██▊       | 351/1250 [10:47<25:55,  1.73s/it, loss=0.453]
Epoch 4/4:  28%|██▊       | 352/1250 [10:47<26:33,  1.77s/it, loss=0.453]
Epoch 4/4:  28%|██▊       | 352/1250 [10:49<26:33,  1.77s/it, loss=0.4]  
Epoch 4/4:  28%|██▊       | 353/1250 [10:49<26:49,  1.79s/it, loss=0.4]
Epoch 4/4:  28%|██▊       | 353/1250 [10:51<26:49,  1.79s/it, loss=0.45]
Epoch 4/4:  28%|██▊       | 354/1250 [10:51<25:55,  1.74s/it, loss=0.45]
Epoch 4/4:  28%|██▊       | 354/1250 [10:53<25:55,  1.74s/it, loss=0.392]
Epoch 4/4:  28%|██▊       | 355/1250 [10:53<26:26,  1.77s/it, loss=0.392]
Epoch 4/4:  28%|██▊       | 355/1250 [10:54<26:26,  1.77s/it, loss=0.505]
Epoch 4/4:  28%|██▊       | 356/1250 [10:54<26:30,  1.78s/it, loss=0.505]
Epoch 4/4:  28%|██▊       | 356/1250 [10:56<26:30,  1.78s/it, loss=0.646]
Epoch 4/4:  29%|██▊       | 357/1250 [10:56<26:17,  1.77s/it, loss=0.646]
Epoch 4/4:  29%|██▊       | 357/1250 [10:58<26:17,  1.77s/it, loss=0.425]
Epoch 4/4:  29%|██▊       | 358/1250 [10:58<26:40,  1.79s/it, loss=0.425]
Epoch 4/4:  29%|██▊       | 358/1250 [10:59<26:40,  1.79s/it, loss=0.422]
Epoch 4/4:  29%|██▊       | 359/1250 [10:59<25:22,  1.71s/it, loss=0.422]
Epoch 4/4:  29%|██▊       | 359/1250 [11:02<25:22,  1.71s/it, loss=0.493]
Epoch 4/4:  29%|██▉       | 360/1250 [11:02<28:57,  1.95s/it, loss=0.493]
Epoch 4/4:  29%|██▉       | 360/1250 [11:05<28:57,  1.95s/it, loss=0.538]
Epoch 4/4:  29%|██▉       | 361/1250 [11:05<35:04,  2.37s/it, loss=0.538]
Epoch 4/4:  29%|██▉       | 361/1250 [11:07<35:04,  2.37s/it, loss=0.315]
Epoch 4/4:  29%|██▉       | 362/1250 [11:07<33:52,  2.29s/it, loss=0.315]
Epoch 4/4:  29%|██▉       | 362/1250 [11:10<33:52,  2.29s/it, loss=0.562]
Epoch 4/4:  29%|██▉       | 363/1250 [11:10<33:19,  2.25s/it, loss=0.562]
Epoch 4/4:  29%|██▉       | 363/1250 [11:12<33:19,  2.25s/it, loss=0.421]
Epoch 4/4:  29%|██▉       | 364/1250 [11:12<31:49,  2.16s/it, loss=0.421]
Epoch 4/4:  29%|██▉       | 364/1250 [11:14<31:49,  2.16s/it, loss=0.416]
Epoch 4/4:  29%|██▉       | 365/1250 [11:14<31:41,  2.15s/it, loss=0.416]
Epoch 4/4:  29%|██▉       | 365/1250 [11:16<31:41,  2.15s/it, loss=0.457]
Epoch 4/4:  29%|██▉       | 366/1250 [11:16<30:29,  2.07s/it, loss=0.457]
Epoch 4/4:  29%|██▉       | 366/1250 [11:17<30:29,  2.07s/it, loss=0.41] 
Epoch 4/4:  29%|██▉       | 367/1250 [11:17<29:03,  1.97s/it, loss=0.41]
Epoch 4/4:  29%|██▉       | 367/1250 [11:20<29:03,  1.97s/it, loss=0.564]
Epoch 4/4:  29%|██▉       | 368/1250 [11:20<30:14,  2.06s/it, loss=0.564]
Epoch 4/4:  29%|██▉       | 368/1250 [11:21<30:14,  2.06s/it, loss=0.665]
Epoch 4/4:  30%|██▉       | 369/1250 [11:21<29:04,  1.98s/it, loss=0.665]
Epoch 4/4:  30%|██▉       | 369/1250 [11:23<29:04,  1.98s/it, loss=0.403]
Epoch 4/4:  30%|██▉       | 370/1250 [11:23<29:27,  2.01s/it, loss=0.403]
Epoch 4/4:  30%|██▉       | 370/1250 [11:25<29:27,  2.01s/it, loss=0.476]
Epoch 4/4:  30%|██▉       | 371/1250 [11:25<29:28,  2.01s/it, loss=0.476]
Epoch 4/4:  30%|██▉       | 371/1250 [11:27<29:28,  2.01s/it, loss=0.447]
Epoch 4/4:  30%|██▉       | 372/1250 [11:27<28:56,  1.98s/it, loss=0.447]
Epoch 4/4:  30%|██▉       | 372/1250 [11:29<28:56,  1.98s/it, loss=0.437]
Epoch 4/4:  30%|██▉       | 373/1250 [11:29<29:05,  1.99s/it, loss=0.437]
Epoch 4/4:  30%|██▉       | 373/1250 [11:31<29:05,  1.99s/it, loss=0.435]
Epoch 4/4:  30%|██▉       | 374/1250 [11:31<28:48,  1.97s/it, loss=0.435]
Epoch 4/4:  30%|██▉       | 374/1250 [11:33<28:48,  1.97s/it, loss=0.433]
Epoch 4/4:  30%|███       | 375/1250 [11:33<29:00,  1.99s/it, loss=0.433]
Epoch 4/4:  30%|███       | 375/1250 [11:35<29:00,  1.99s/it, loss=0.483]
Epoch 4/4:  30%|███       | 376/1250 [11:35<29:49,  2.05s/it, loss=0.483]
Epoch 4/4:  30%|███       | 376/1250 [11:37<29:49,  2.05s/it, loss=0.456]
Epoch 4/4:  30%|███       | 377/1250 [11:37<29:27,  2.02s/it, loss=0.456]
Epoch 4/4:  30%|███       | 377/1250 [11:40<29:27,  2.02s/it, loss=0.307]
Epoch 4/4:  30%|███       | 378/1250 [11:40<30:43,  2.11s/it, loss=0.307]
Epoch 4/4:  30%|███       | 378/1250 [11:42<30:43,  2.11s/it, loss=0.428]
Epoch 4/4:  30%|███       | 379/1250 [11:42<30:42,  2.12s/it, loss=0.428]
Epoch 4/4:  30%|███       | 379/1250 [11:43<30:42,  2.12s/it, loss=0.371]
Epoch 4/4:  30%|███       | 380/1250 [11:43<28:02,  1.93s/it, loss=0.371]
Epoch 4/4:  30%|███       | 380/1250 [11:45<28:02,  1.93s/it, loss=0.471]
Epoch 4/4:  30%|███       | 381/1250 [11:45<28:40,  1.98s/it, loss=0.471]
Epoch 4/4:  30%|███       | 381/1250 [11:47<28:40,  1.98s/it, loss=0.437]
Epoch 4/4:  31%|███       | 382/1250 [11:47<27:18,  1.89s/it, loss=0.437]
Epoch 4/4:  31%|███       | 382/1250 [11:49<27:18,  1.89s/it, loss=0.297]
Epoch 4/4:  31%|███       | 383/1250 [11:49<27:29,  1.90s/it, loss=0.297]
Epoch 4/4:  31%|███       | 383/1250 [11:51<27:29,  1.90s/it, loss=0.465]
Epoch 4/4:  31%|███       | 384/1250 [11:51<27:59,  1.94s/it, loss=0.465]
Epoch 4/4:  31%|███       | 384/1250 [11:53<27:59,  1.94s/it, loss=0.412]
Epoch 4/4:  31%|███       | 385/1250 [11:53<26:12,  1.82s/it, loss=0.412]
Epoch 4/4:  31%|███       | 385/1250 [11:55<26:12,  1.82s/it, loss=0.397]
Epoch 4/4:  31%|███       | 386/1250 [11:55<27:24,  1.90s/it, loss=0.397]
Epoch 4/4:  31%|███       | 386/1250 [11:56<27:24,  1.90s/it, loss=0.55] 
Epoch 4/4:  31%|███       | 387/1250 [11:56<25:56,  1.80s/it, loss=0.55]
Epoch 4/4:  31%|███       | 387/1250 [11:58<25:56,  1.80s/it, loss=0.52]
Epoch 4/4:  31%|███       | 388/1250 [11:58<26:46,  1.86s/it, loss=0.52]
Epoch 4/4:  31%|███       | 388/1250 [12:00<26:46,  1.86s/it, loss=0.465]
Epoch 4/4:  31%|███       | 389/1250 [12:00<26:34,  1.85s/it, loss=0.465]
Epoch 4/4:  31%|███       | 389/1250 [12:02<26:34,  1.85s/it, loss=0.347]
Epoch 4/4:  31%|███       | 390/1250 [12:02<25:27,  1.78s/it, loss=0.347]
Epoch 4/4:  31%|███       | 390/1250 [12:04<25:27,  1.78s/it, loss=0.328]
Epoch 4/4:  31%|███▏      | 391/1250 [12:04<26:08,  1.83s/it, loss=0.328]
Epoch 4/4:  31%|███▏      | 391/1250 [12:05<26:08,  1.83s/it, loss=0.397]
Epoch 4/4:  31%|███▏      | 392/1250 [12:05<25:00,  1.75s/it, loss=0.397]
Epoch 4/4:  31%|███▏      | 392/1250 [12:07<25:00,  1.75s/it, loss=0.489]
Epoch 4/4:  31%|███▏      | 393/1250 [12:07<23:52,  1.67s/it, loss=0.489]
Epoch 4/4:  31%|███▏      | 393/1250 [12:08<23:52,  1.67s/it, loss=0.382]
Epoch 4/4:  32%|███▏      | 394/1250 [12:08<23:05,  1.62s/it, loss=0.382]
Epoch 4/4:  32%|███▏      | 394/1250 [12:10<23:05,  1.62s/it, loss=0.34] 
Epoch 4/4:  32%|███▏      | 395/1250 [12:10<22:28,  1.58s/it, loss=0.34]
Epoch 4/4:  32%|███▏      | 395/1250 [12:11<22:28,  1.58s/it, loss=0.463]
Epoch 4/4:  32%|███▏      | 396/1250 [12:11<21:56,  1.54s/it, loss=0.463]
Epoch 4/4:  32%|███▏      | 396/1250 [12:13<21:56,  1.54s/it, loss=0.413]
Epoch 4/4:  32%|███▏      | 397/1250 [12:13<21:44,  1.53s/it, loss=0.413]
Epoch 4/4:  32%|███▏      | 397/1250 [12:14<21:44,  1.53s/it, loss=0.345]
Epoch 4/4:  32%|███▏      | 398/1250 [12:14<21:33,  1.52s/it, loss=0.345]
Epoch 4/4:  32%|███▏      | 398/1250 [12:16<21:33,  1.52s/it, loss=0.4]  
Epoch 4/4:  32%|███▏      | 399/1250 [12:16<21:25,  1.51s/it, loss=0.4]
Epoch 4/4:  32%|███▏      | 399/1250 [12:17<21:25,  1.51s/it, loss=0.387]
Epoch 4/4:  32%|███▏      | 400/1250 [12:17<21:02,  1.48s/it, loss=0.387]
Epoch 4/4:  32%|███▏      | 400/1250 [12:19<21:02,  1.48s/it, loss=0.569]
Epoch 4/4:  32%|███▏      | 401/1250 [12:19<21:14,  1.50s/it, loss=0.569]
Epoch 4/4:  32%|███▏      | 401/1250 [12:20<21:14,  1.50s/it, loss=0.349]
Epoch 4/4:  32%|███▏      | 402/1250 [12:20<21:18,  1.51s/it, loss=0.349]
Epoch 4/4:  32%|███▏      | 402/1250 [12:22<21:18,  1.51s/it, loss=0.404]
Epoch 4/4:  32%|███▏      | 403/1250 [12:22<21:10,  1.50s/it, loss=0.404]
Epoch 4/4:  32%|███▏      | 403/1250 [12:23<21:10,  1.50s/it, loss=0.34] 
Epoch 4/4:  32%|███▏      | 404/1250 [12:23<22:28,  1.59s/it, loss=0.34]
Epoch 4/4:  32%|███▏      | 404/1250 [12:26<22:28,  1.59s/it, loss=0.388]
Epoch 4/4:  32%|███▏      | 405/1250 [12:26<25:48,  1.83s/it, loss=0.388]
Epoch 4/4:  32%|███▏      | 405/1250 [12:27<25:48,  1.83s/it, loss=0.436]
Epoch 4/4:  32%|███▏      | 406/1250 [12:27<22:18,  1.59s/it, loss=0.436]
Epoch 4/4:  32%|███▏      | 406/1250 [12:28<22:18,  1.59s/it, loss=0.377]
Epoch 4/4:  33%|███▎      | 407/1250 [12:28<22:01,  1.57s/it, loss=0.377]
Epoch 4/4:  33%|███▎      | 407/1250 [12:30<22:01,  1.57s/it, loss=0.517]
Epoch 4/4:  33%|███▎      | 408/1250 [12:30<21:57,  1.57s/it, loss=0.517]
Epoch 4/4:  33%|███▎      | 408/1250 [12:31<21:57,  1.57s/it, loss=0.329]
Epoch 4/4:  33%|███▎      | 409/1250 [12:31<21:31,  1.54s/it, loss=0.329]
Epoch 4/4:  33%|███▎      | 409/1250 [12:33<21:31,  1.54s/it, loss=0.468]
Epoch 4/4:  33%|███▎      | 410/1250 [12:33<22:41,  1.62s/it, loss=0.468]
Epoch 4/4:  33%|███▎      | 410/1250 [12:35<22:41,  1.62s/it, loss=0.455]
Epoch 4/4:  33%|███▎      | 411/1250 [12:35<24:13,  1.73s/it, loss=0.455]
Epoch 4/4:  33%|███▎      | 411/1250 [12:37<24:13,  1.73s/it, loss=0.342]
Epoch 4/4:  33%|███▎      | 412/1250 [12:37<22:35,  1.62s/it, loss=0.342]
Epoch 4/4:  33%|███▎      | 412/1250 [12:39<22:35,  1.62s/it, loss=0.33] 
Epoch 4/4:  33%|███▎      | 413/1250 [12:39<23:52,  1.71s/it, loss=0.33]
Epoch 4/4:  33%|███▎      | 413/1250 [12:40<23:52,  1.71s/it, loss=0.49]
Epoch 4/4:  33%|███▎      | 414/1250 [12:40<23:05,  1.66s/it, loss=0.49]
Epoch 4/4:  33%|███▎      | 414/1250 [12:41<23:05,  1.66s/it, loss=0.602]
Epoch 4/4:  33%|███▎      | 415/1250 [12:41<22:10,  1.59s/it, loss=0.602]
Epoch 4/4:  33%|███▎      | 415/1250 [12:43<22:10,  1.59s/it, loss=0.367]
Epoch 4/4:  33%|███▎      | 416/1250 [12:43<21:17,  1.53s/it, loss=0.367]
Epoch 4/4:  33%|███▎      | 416/1250 [12:44<21:17,  1.53s/it, loss=0.544]
Epoch 4/4:  33%|███▎      | 417/1250 [12:44<21:27,  1.55s/it, loss=0.544]
Epoch 4/4:  33%|███▎      | 417/1250 [12:46<21:27,  1.55s/it, loss=0.342]
Epoch 4/4:  33%|███▎      | 418/1250 [12:46<21:42,  1.57s/it, loss=0.342]
Epoch 4/4:  33%|███▎      | 418/1250 [12:48<21:42,  1.57s/it, loss=0.579]
Epoch 4/4:  34%|███▎      | 419/1250 [12:48<21:27,  1.55s/it, loss=0.579]
Epoch 4/4:  34%|███▎      | 419/1250 [12:49<21:27,  1.55s/it, loss=0.362]
Epoch 4/4:  34%|███▎      | 420/1250 [12:49<22:42,  1.64s/it, loss=0.362]
Epoch 4/4:  34%|███▎      | 420/1250 [12:51<22:42,  1.64s/it, loss=0.401]
Epoch 4/4:  34%|███▎      | 421/1250 [12:51<22:21,  1.62s/it, loss=0.401]
Epoch 4/4:  34%|███▎      | 421/1250 [12:52<22:21,  1.62s/it, loss=0.369]
Epoch 4/4:  34%|███▍      | 422/1250 [12:52<21:44,  1.58s/it, loss=0.369]
Epoch 4/4:  34%|███▍      | 422/1250 [12:54<21:44,  1.58s/it, loss=0.395]
Epoch 4/4:  34%|███▍      | 423/1250 [12:54<21:41,  1.57s/it, loss=0.395]
Epoch 4/4:  34%|███▍      | 423/1250 [12:56<21:41,  1.57s/it, loss=0.459]
Epoch 4/4:  34%|███▍      | 424/1250 [12:56<21:43,  1.58s/it, loss=0.459]
Epoch 4/4:  34%|███▍      | 424/1250 [12:58<21:43,  1.58s/it, loss=0.563]
Epoch 4/4:  34%|███▍      | 425/1250 [12:58<22:59,  1.67s/it, loss=0.563]
Epoch 4/4:  34%|███▍      | 425/1250 [12:59<22:59,  1.67s/it, loss=0.471]
Epoch 4/4:  34%|███▍      | 426/1250 [12:59<22:22,  1.63s/it, loss=0.471]
Epoch 4/4:  34%|███▍      | 426/1250 [13:01<22:22,  1.63s/it, loss=0.386]
Epoch 4/4:  34%|███▍      | 427/1250 [13:01<22:29,  1.64s/it, loss=0.386]
Epoch 4/4:  34%|███▍      | 427/1250 [13:03<22:29,  1.64s/it, loss=0.568]
Epoch 4/4:  34%|███▍      | 428/1250 [13:03<23:17,  1.70s/it, loss=0.568]
Epoch 4/4:  34%|███▍      | 428/1250 [13:04<23:17,  1.70s/it, loss=0.267]
Epoch 4/4:  34%|███▍      | 429/1250 [13:04<22:07,  1.62s/it, loss=0.267]
Epoch 4/4:  34%|███▍      | 429/1250 [13:06<22:07,  1.62s/it, loss=0.429]
Epoch 4/4:  34%|███▍      | 430/1250 [13:06<21:53,  1.60s/it, loss=0.429]
Epoch 4/4:  34%|███▍      | 430/1250 [13:07<21:53,  1.60s/it, loss=0.437]
Epoch 4/4:  34%|███▍      | 431/1250 [13:07<21:25,  1.57s/it, loss=0.437]
Epoch 4/4:  34%|███▍      | 431/1250 [13:08<21:25,  1.57s/it, loss=0.407]
Epoch 4/4:  35%|███▍      | 432/1250 [13:08<20:43,  1.52s/it, loss=0.407]
Epoch 4/4:  35%|███▍      | 432/1250 [13:10<20:43,  1.52s/it, loss=0.525]
Epoch 4/4:  35%|███▍      | 433/1250 [13:10<22:13,  1.63s/it, loss=0.525]
Epoch 4/4:  35%|███▍      | 433/1250 [13:12<22:13,  1.63s/it, loss=0.437]
Epoch 4/4:  35%|███▍      | 434/1250 [13:12<23:13,  1.71s/it, loss=0.437]
Epoch 4/4:  35%|███▍      | 434/1250 [13:14<23:13,  1.71s/it, loss=0.57] 
Epoch 4/4:  35%|███▍      | 435/1250 [13:14<22:08,  1.63s/it, loss=0.57]
Epoch 4/4:  35%|███▍      | 435/1250 [13:16<22:08,  1.63s/it, loss=0.524]
Epoch 4/4:  35%|███▍      | 436/1250 [13:16<23:21,  1.72s/it, loss=0.524]
Epoch 4/4:  35%|███▍      | 436/1250 [13:18<23:21,  1.72s/it, loss=0.468]
Epoch 4/4:  35%|███▍      | 437/1250 [13:18<24:45,  1.83s/it, loss=0.468]
Epoch 4/4:  35%|███▍      | 437/1250 [13:21<24:45,  1.83s/it, loss=0.751]
Epoch 4/4:  35%|███▌      | 438/1250 [13:21<32:47,  2.42s/it, loss=0.751]
Epoch 4/4:  35%|███▌      | 438/1250 [13:25<32:47,  2.42s/it, loss=0.274]
Epoch 4/4:  35%|███▌      | 439/1250 [13:25<37:07,  2.75s/it, loss=0.274]
Epoch 4/4:  35%|███▌      | 439/1250 [13:26<37:07,  2.75s/it, loss=0.571]
Epoch 4/4:  35%|███▌      | 440/1250 [13:26<31:50,  2.36s/it, loss=0.571]
Epoch 4/4:  35%|███▌      | 440/1250 [13:28<31:50,  2.36s/it, loss=0.613]
Epoch 4/4:  35%|███▌      | 441/1250 [13:28<28:35,  2.12s/it, loss=0.613]
Epoch 4/4:  35%|███▌      | 441/1250 [13:30<28:35,  2.12s/it, loss=0.407]
Epoch 4/4:  35%|███▌      | 442/1250 [13:30<26:23,  1.96s/it, loss=0.407]
Epoch 4/4:  35%|███▌      | 442/1250 [13:32<26:23,  1.96s/it, loss=0.451]
Epoch 4/4:  35%|███▌      | 443/1250 [13:32<26:16,  1.95s/it, loss=0.451]
Epoch 4/4:  35%|███▌      | 443/1250 [13:33<26:16,  1.95s/it, loss=0.415]
Epoch 4/4:  36%|███▌      | 444/1250 [13:33<24:30,  1.82s/it, loss=0.415]
Epoch 4/4:  36%|███▌      | 444/1250 [13:35<24:30,  1.82s/it, loss=0.417]
Epoch 4/4:  36%|███▌      | 445/1250 [13:35<23:48,  1.77s/it, loss=0.417]
Epoch 4/4:  36%|███▌      | 445/1250 [13:36<23:48,  1.77s/it, loss=0.49] 
Epoch 4/4:  36%|███▌      | 446/1250 [13:36<22:52,  1.71s/it, loss=0.49]
Epoch 4/4:  36%|███▌      | 446/1250 [13:38<22:52,  1.71s/it, loss=0.477]
Epoch 4/4:  36%|███▌      | 447/1250 [13:38<21:56,  1.64s/it, loss=0.477]
Epoch 4/4:  36%|███▌      | 447/1250 [13:39<21:56,  1.64s/it, loss=0.369]
Epoch 4/4:  36%|███▌      | 448/1250 [13:39<21:33,  1.61s/it, loss=0.369]
Epoch 4/4:  36%|███▌      | 448/1250 [13:41<21:33,  1.61s/it, loss=0.504]
Epoch 4/4:  36%|███▌      | 449/1250 [13:41<22:24,  1.68s/it, loss=0.504]
Epoch 4/4:  36%|███▌      | 449/1250 [13:43<22:24,  1.68s/it, loss=0.421]
Epoch 4/4:  36%|███▌      | 450/1250 [13:43<21:56,  1.65s/it, loss=0.421]
Epoch 4/4:  36%|███▌      | 450/1250 [13:44<21:56,  1.65s/it, loss=0.421]
Epoch 4/4:  36%|███▌      | 451/1250 [13:44<21:14,  1.59s/it, loss=0.421]
Epoch 4/4:  36%|███▌      | 451/1250 [13:46<21:14,  1.59s/it, loss=0.377]
Epoch 4/4:  36%|███▌      | 452/1250 [13:46<21:31,  1.62s/it, loss=0.377]
Epoch 4/4:  36%|███▌      | 452/1250 [13:47<21:31,  1.62s/it, loss=0.365]
Epoch 4/4:  36%|███▌      | 453/1250 [13:47<21:11,  1.60s/it, loss=0.365]
Epoch 4/4:  36%|███▌      | 453/1250 [13:49<21:11,  1.60s/it, loss=0.474]
Epoch 4/4:  36%|███▋      | 454/1250 [13:49<21:16,  1.60s/it, loss=0.474]
Epoch 4/4:  36%|███▋      | 454/1250 [13:51<21:16,  1.60s/it, loss=0.266]
Epoch 4/4:  36%|███▋      | 455/1250 [13:51<22:19,  1.68s/it, loss=0.266]
Epoch 4/4:  36%|███▋      | 455/1250 [13:53<22:19,  1.68s/it, loss=0.477]
Epoch 4/4:  36%|███▋      | 456/1250 [13:53<22:57,  1.74s/it, loss=0.477]
Epoch 4/4:  36%|███▋      | 456/1250 [13:54<22:57,  1.74s/it, loss=0.39] 
Epoch 4/4:  37%|███▋      | 457/1250 [13:54<22:26,  1.70s/it, loss=0.39]
Epoch 4/4:  37%|███▋      | 457/1250 [13:56<22:26,  1.70s/it, loss=0.587]
Epoch 4/4:  37%|███▋      | 458/1250 [13:56<23:51,  1.81s/it, loss=0.587]
Epoch 4/4:  37%|███▋      | 458/1250 [13:58<23:51,  1.81s/it, loss=0.496]
Epoch 4/4:  37%|███▋      | 459/1250 [13:58<24:01,  1.82s/it, loss=0.496]
Epoch 4/4:  37%|███▋      | 459/1250 [14:00<24:01,  1.82s/it, loss=0.418]
Epoch 4/4:  37%|███▋      | 460/1250 [14:00<22:40,  1.72s/it, loss=0.418]
Epoch 4/4:  37%|███▋      | 460/1250 [14:01<22:40,  1.72s/it, loss=0.509]
Epoch 4/4:  37%|███▋      | 461/1250 [14:01<22:02,  1.68s/it, loss=0.509]
Epoch 4/4:  37%|███▋      | 461/1250 [14:03<22:02,  1.68s/it, loss=0.601]
Epoch 4/4:  37%|███▋      | 462/1250 [14:03<21:39,  1.65s/it, loss=0.601]
Epoch 4/4:  37%|███▋      | 462/1250 [14:04<21:39,  1.65s/it, loss=0.357]
Epoch 4/4:  37%|███▋      | 463/1250 [14:04<21:04,  1.61s/it, loss=0.357]
Epoch 4/4:  37%|███▋      | 463/1250 [14:06<21:04,  1.61s/it, loss=0.503]
Epoch 4/4:  37%|███▋      | 464/1250 [14:06<20:53,  1.59s/it, loss=0.503]
Epoch 4/4:  37%|███▋      | 464/1250 [14:08<20:53,  1.59s/it, loss=0.318]
Epoch 4/4:  37%|███▋      | 465/1250 [14:08<20:43,  1.58s/it, loss=0.318]
Epoch 4/4:  37%|███▋      | 465/1250 [14:09<20:43,  1.58s/it, loss=0.411]
Epoch 4/4:  37%|███▋      | 466/1250 [14:09<20:37,  1.58s/it, loss=0.411]
Epoch 4/4:  37%|███▋      | 466/1250 [14:11<20:37,  1.58s/it, loss=0.497]
Epoch 4/4:  37%|███▋      | 467/1250 [14:11<20:28,  1.57s/it, loss=0.497]
Epoch 4/4:  37%|███▋      | 467/1250 [14:13<20:28,  1.57s/it, loss=0.341]
Epoch 4/4:  37%|███▋      | 468/1250 [14:13<21:33,  1.65s/it, loss=0.341]
Epoch 4/4:  37%|███▋      | 468/1250 [14:14<21:33,  1.65s/it, loss=0.453]
Epoch 4/4:  38%|███▊      | 469/1250 [14:14<21:51,  1.68s/it, loss=0.453]
Epoch 4/4:  38%|███▊      | 469/1250 [14:16<21:51,  1.68s/it, loss=0.436]
Epoch 4/4:  38%|███▊      | 470/1250 [14:16<22:25,  1.73s/it, loss=0.436]
Epoch 4/4:  38%|███▊      | 470/1250 [14:18<22:25,  1.73s/it, loss=0.471]
Epoch 4/4:  38%|███▊      | 471/1250 [14:18<22:49,  1.76s/it, loss=0.471]
Epoch 4/4:  38%|███▊      | 471/1250 [14:19<22:49,  1.76s/it, loss=0.634]
Epoch 4/4:  38%|███▊      | 472/1250 [14:19<21:27,  1.66s/it, loss=0.634]
Epoch 4/4:  38%|███▊      | 472/1250 [14:21<21:27,  1.66s/it, loss=0.399]
Epoch 4/4:  38%|███▊      | 473/1250 [14:21<22:13,  1.72s/it, loss=0.399]
Epoch 4/4:  38%|███▊      | 473/1250 [14:23<22:13,  1.72s/it, loss=0.44] 
Epoch 4/4:  38%|███▊      | 474/1250 [14:23<22:09,  1.71s/it, loss=0.44]
Epoch 4/4:  38%|███▊      | 474/1250 [14:25<22:09,  1.71s/it, loss=0.339]
Epoch 4/4:  38%|███▊      | 475/1250 [14:25<22:23,  1.73s/it, loss=0.339]
Epoch 4/4:  38%|███▊      | 475/1250 [14:26<22:23,  1.73s/it, loss=0.454]
Epoch 4/4:  38%|███▊      | 476/1250 [14:26<21:49,  1.69s/it, loss=0.454]
Epoch 4/4:  38%|███▊      | 476/1250 [14:28<21:49,  1.69s/it, loss=0.539]
Epoch 4/4:  38%|███▊      | 477/1250 [14:28<22:20,  1.73s/it, loss=0.539]
Epoch 4/4:  38%|███▊      | 477/1250 [14:30<22:20,  1.73s/it, loss=0.594]
Epoch 4/4:  38%|███▊      | 478/1250 [14:30<21:25,  1.67s/it, loss=0.594]
Epoch 4/4:  38%|███▊      | 478/1250 [14:31<21:25,  1.67s/it, loss=0.438]
Epoch 4/4:  38%|███▊      | 479/1250 [14:31<20:48,  1.62s/it, loss=0.438]
Epoch 4/4:  38%|███▊      | 479/1250 [14:33<20:48,  1.62s/it, loss=0.534]
Epoch 4/4:  38%|███▊      | 480/1250 [14:33<20:22,  1.59s/it, loss=0.534]
Epoch 4/4:  38%|███▊      | 480/1250 [14:34<20:22,  1.59s/it, loss=0.322]
Epoch 4/4:  38%|███▊      | 481/1250 [14:34<20:19,  1.59s/it, loss=0.322]
Epoch 4/4:  38%|███▊      | 481/1250 [14:36<20:19,  1.59s/it, loss=0.37] 
Epoch 4/4:  39%|███▊      | 482/1250 [14:36<19:52,  1.55s/it, loss=0.37]
Epoch 4/4:  39%|███▊      | 482/1250 [14:37<19:52,  1.55s/it, loss=0.431]
Epoch 4/4:  39%|███▊      | 483/1250 [14:37<19:56,  1.56s/it, loss=0.431]
Epoch 4/4:  39%|███▊      | 483/1250 [14:39<19:56,  1.56s/it, loss=0.603]
Epoch 4/4:  39%|███▊      | 484/1250 [14:39<20:01,  1.57s/it, loss=0.603]
Epoch 4/4:  39%|███▊      | 484/1250 [14:40<20:01,  1.57s/it, loss=0.413]
Epoch 4/4:  39%|███▉      | 485/1250 [14:40<20:11,  1.58s/it, loss=0.413]
Epoch 4/4:  39%|███▉      | 485/1250 [14:42<20:11,  1.58s/it, loss=0.358]
Epoch 4/4:  39%|███▉      | 486/1250 [14:42<21:38,  1.70s/it, loss=0.358]
Epoch 4/4:  39%|███▉      | 486/1250 [14:44<21:38,  1.70s/it, loss=0.489]
Epoch 4/4:  39%|███▉      | 487/1250 [14:44<22:04,  1.74s/it, loss=0.489]
Epoch 4/4:  39%|███▉      | 487/1250 [14:46<22:04,  1.74s/it, loss=0.432]
Epoch 4/4:  39%|███▉      | 488/1250 [14:46<22:03,  1.74s/it, loss=0.432]
Epoch 4/4:  39%|███▉      | 488/1250 [14:48<22:03,  1.74s/it, loss=0.396]
Epoch 4/4:  39%|███▉      | 489/1250 [14:48<22:51,  1.80s/it, loss=0.396]
Epoch 4/4:  39%|███▉      | 489/1250 [14:49<22:51,  1.80s/it, loss=0.379]
Epoch 4/4:  39%|███▉      | 490/1250 [14:49<21:19,  1.68s/it, loss=0.379]
Epoch 4/4:  39%|███▉      | 490/1250 [14:51<21:19,  1.68s/it, loss=0.414]
Epoch 4/4:  39%|███▉      | 491/1250 [14:51<20:36,  1.63s/it, loss=0.414]
Epoch 4/4:  39%|███▉      | 491/1250 [14:52<20:36,  1.63s/it, loss=0.313]
Epoch 4/4:  39%|███▉      | 492/1250 [14:52<20:30,  1.62s/it, loss=0.313]
Epoch 4/4:  39%|███▉      | 492/1250 [14:54<20:30,  1.62s/it, loss=0.679]
Epoch 4/4:  39%|███▉      | 493/1250 [14:54<20:12,  1.60s/it, loss=0.679]
Epoch 4/4:  39%|███▉      | 493/1250 [14:56<20:12,  1.60s/it, loss=0.484]
Epoch 4/4:  40%|███▉      | 494/1250 [14:56<21:11,  1.68s/it, loss=0.484]
Epoch 4/4:  40%|███▉      | 494/1250 [14:57<21:11,  1.68s/it, loss=0.388]
Epoch 4/4:  40%|███▉      | 495/1250 [14:57<19:34,  1.56s/it, loss=0.388]
Epoch 4/4:  40%|███▉      | 495/1250 [14:59<19:34,  1.56s/it, loss=0.357]
Epoch 4/4:  40%|███▉      | 496/1250 [14:59<19:41,  1.57s/it, loss=0.357]
Epoch 4/4:  40%|███▉      | 496/1250 [15:00<19:41,  1.57s/it, loss=0.544]
Epoch 4/4:  40%|███▉      | 497/1250 [15:00<19:55,  1.59s/it, loss=0.544]
Epoch 4/4:  40%|███▉      | 497/1250 [15:02<19:55,  1.59s/it, loss=0.444]
Epoch 4/4:  40%|███▉      | 498/1250 [15:02<19:30,  1.56s/it, loss=0.444]
Epoch 4/4:  40%|███▉      | 498/1250 [15:03<19:30,  1.56s/it, loss=0.565]
Epoch 4/4:  40%|███▉      | 499/1250 [15:03<19:25,  1.55s/it, loss=0.565]
Epoch 4/4:  40%|███▉      | 499/1250 [15:05<19:25,  1.55s/it, loss=0.327]
Epoch 4/4:  40%|████      | 500/1250 [15:05<19:23,  1.55s/it, loss=0.327]
Epoch 4/4:  40%|████      | 500/1250 [15:06<19:23,  1.55s/it, loss=0.548]
Epoch 4/4:  40%|████      | 501/1250 [15:06<19:10,  1.54s/it, loss=0.548]
Epoch 4/4:  40%|████      | 501/1250 [15:08<19:10,  1.54s/it, loss=0.455]
Epoch 4/4:  40%|████      | 502/1250 [15:08<19:29,  1.56s/it, loss=0.455]
Epoch 4/4:  40%|████      | 502/1250 [15:10<19:29,  1.56s/it, loss=0.516]
Epoch 4/4:  40%|████      | 503/1250 [15:10<19:43,  1.58s/it, loss=0.516]
Epoch 4/4:  40%|████      | 503/1250 [15:11<19:43,  1.58s/it, loss=0.472]
Epoch 4/4:  40%|████      | 504/1250 [15:11<19:31,  1.57s/it, loss=0.472]
Epoch 4/4:  40%|████      | 504/1250 [15:13<19:31,  1.57s/it, loss=0.621]
Epoch 4/4:  40%|████      | 505/1250 [15:13<19:26,  1.57s/it, loss=0.621]
Epoch 4/4:  40%|████      | 505/1250 [15:14<19:26,  1.57s/it, loss=0.45] 
Epoch 4/4:  40%|████      | 506/1250 [15:14<19:14,  1.55s/it, loss=0.45]
Epoch 4/4:  40%|████      | 506/1250 [15:16<19:14,  1.55s/it, loss=0.384]
Epoch 4/4:  41%|████      | 507/1250 [15:16<19:20,  1.56s/it, loss=0.384]
Epoch 4/4:  41%|████      | 507/1250 [15:18<19:20,  1.56s/it, loss=0.397]
Epoch 4/4:  41%|████      | 508/1250 [15:18<19:57,  1.61s/it, loss=0.397]
Epoch 4/4:  41%|████      | 508/1250 [15:19<19:57,  1.61s/it, loss=0.3]  
Epoch 4/4:  41%|████      | 509/1250 [15:19<19:57,  1.62s/it, loss=0.3]
Epoch 4/4:  41%|████      | 509/1250 [15:21<19:57,  1.62s/it, loss=0.503]
Epoch 4/4:  41%|████      | 510/1250 [15:21<20:19,  1.65s/it, loss=0.503]
Epoch 4/4:  41%|████      | 510/1250 [15:23<20:19,  1.65s/it, loss=0.382]
Epoch 4/4:  41%|████      | 511/1250 [15:23<19:49,  1.61s/it, loss=0.382]
Epoch 4/4:  41%|████      | 511/1250 [15:24<19:49,  1.61s/it, loss=0.397]
Epoch 4/4:  41%|████      | 512/1250 [15:24<20:51,  1.70s/it, loss=0.397]
Epoch 4/4:  41%|████      | 512/1250 [15:26<20:51,  1.70s/it, loss=0.325]
Epoch 4/4:  41%|████      | 513/1250 [15:26<20:33,  1.67s/it, loss=0.325]
Epoch 4/4:  41%|████      | 513/1250 [15:28<20:33,  1.67s/it, loss=0.427]
Epoch 4/4:  41%|████      | 514/1250 [15:28<20:38,  1.68s/it, loss=0.427]
Epoch 4/4:  41%|████      | 514/1250 [15:29<20:38,  1.68s/it, loss=0.548]
Epoch 4/4:  41%|████      | 515/1250 [15:29<20:00,  1.63s/it, loss=0.548]
Epoch 4/4:  41%|████      | 515/1250 [15:31<20:00,  1.63s/it, loss=0.403]
Epoch 4/4:  41%|████▏     | 516/1250 [15:31<20:00,  1.64s/it, loss=0.403]
Epoch 4/4:  41%|████▏     | 516/1250 [15:32<20:00,  1.64s/it, loss=0.451]
Epoch 4/4:  41%|████▏     | 517/1250 [15:32<19:20,  1.58s/it, loss=0.451]
Epoch 4/4:  41%|████▏     | 517/1250 [15:34<19:20,  1.58s/it, loss=0.528]
Epoch 4/4:  41%|████▏     | 518/1250 [15:34<19:05,  1.56s/it, loss=0.528]
Epoch 4/4:  41%|████▏     | 518/1250 [15:35<19:05,  1.56s/it, loss=0.453]
Epoch 4/4:  42%|████▏     | 519/1250 [15:35<18:54,  1.55s/it, loss=0.453]
Epoch 4/4:  42%|████▏     | 519/1250 [15:37<18:54,  1.55s/it, loss=0.569]
Epoch 4/4:  42%|████▏     | 520/1250 [15:37<18:32,  1.52s/it, loss=0.569]
Epoch 4/4:  42%|████▏     | 520/1250 [15:38<18:32,  1.52s/it, loss=0.324]
Epoch 4/4:  42%|████▏     | 521/1250 [15:38<18:24,  1.51s/it, loss=0.324]
Epoch 4/4:  42%|████▏     | 521/1250 [15:40<18:24,  1.51s/it, loss=0.482]
Epoch 4/4:  42%|████▏     | 522/1250 [15:40<19:00,  1.57s/it, loss=0.482]
Epoch 4/4:  42%|████▏     | 522/1250 [15:42<19:00,  1.57s/it, loss=0.376]
Epoch 4/4:  42%|████▏     | 523/1250 [15:42<18:52,  1.56s/it, loss=0.376]
Epoch 4/4:  42%|████▏     | 523/1250 [15:43<18:52,  1.56s/it, loss=0.382]
Epoch 4/4:  42%|████▏     | 524/1250 [15:43<18:34,  1.53s/it, loss=0.382]
Epoch 4/4:  42%|████▏     | 524/1250 [15:45<18:34,  1.53s/it, loss=0.457]
Epoch 4/4:  42%|████▏     | 525/1250 [15:45<18:51,  1.56s/it, loss=0.457]
Epoch 4/4:  42%|████▏     | 525/1250 [15:47<18:51,  1.56s/it, loss=0.525]
Epoch 4/4:  42%|████▏     | 526/1250 [15:47<20:03,  1.66s/it, loss=0.525]
Epoch 4/4:  42%|████▏     | 526/1250 [15:48<20:03,  1.66s/it, loss=0.486]
Epoch 4/4:  42%|████▏     | 527/1250 [15:48<19:53,  1.65s/it, loss=0.486]
Epoch 4/4:  42%|████▏     | 527/1250 [15:50<19:53,  1.65s/it, loss=0.379]
Epoch 4/4:  42%|████▏     | 528/1250 [15:50<19:21,  1.61s/it, loss=0.379]
Epoch 4/4:  42%|████▏     | 528/1250 [15:52<19:21,  1.61s/it, loss=0.43] 
Epoch 4/4:  42%|████▏     | 529/1250 [15:52<20:08,  1.68s/it, loss=0.43]
Epoch 4/4:  42%|████▏     | 529/1250 [15:53<20:08,  1.68s/it, loss=0.367]
Epoch 4/4:  42%|████▏     | 530/1250 [15:53<18:54,  1.58s/it, loss=0.367]
Epoch 4/4:  42%|████▏     | 530/1250 [15:55<18:54,  1.58s/it, loss=0.5]  
Epoch 4/4:  42%|████▏     | 531/1250 [15:55<20:29,  1.71s/it, loss=0.5]
Epoch 4/4:  42%|████▏     | 531/1250 [15:58<20:29,  1.71s/it, loss=0.692]
Epoch 4/4:  43%|████▎     | 532/1250 [15:58<23:46,  1.99s/it, loss=0.692]
Epoch 4/4:  43%|████▎     | 532/1250 [16:02<23:46,  1.99s/it, loss=0.46] 
Epoch 4/4:  43%|████▎     | 533/1250 [16:02<33:31,  2.80s/it, loss=0.46]
Epoch 4/4:  43%|████▎     | 533/1250 [16:06<33:31,  2.80s/it, loss=0.373]
Epoch 4/4:  43%|████▎     | 534/1250 [16:06<35:43,  2.99s/it, loss=0.373]
Epoch 4/4:  43%|████▎     | 534/1250 [16:07<35:43,  2.99s/it, loss=0.614]
Epoch 4/4:  43%|████▎     | 535/1250 [16:07<31:14,  2.62s/it, loss=0.614]
Epoch 4/4:  43%|████▎     | 535/1250 [16:09<31:14,  2.62s/it, loss=0.614]
Epoch 4/4:  43%|████▎     | 536/1250 [16:09<27:46,  2.33s/it, loss=0.614]
Epoch 4/4:  43%|████▎     | 536/1250 [16:12<27:46,  2.33s/it, loss=0.286]
Epoch 4/4:  43%|████▎     | 537/1250 [16:12<29:27,  2.48s/it, loss=0.286]
Epoch 4/4:  43%|████▎     | 537/1250 [16:14<29:27,  2.48s/it, loss=0.558]
Epoch 4/4:  43%|████▎     | 538/1250 [16:14<28:16,  2.38s/it, loss=0.558]
Epoch 4/4:  43%|████▎     | 538/1250 [16:16<28:16,  2.38s/it, loss=0.361]
Epoch 4/4:  43%|████▎     | 539/1250 [16:16<28:02,  2.37s/it, loss=0.361]
Epoch 4/4:  43%|████▎     | 539/1250 [16:18<28:02,  2.37s/it, loss=0.371]
Epoch 4/4:  43%|████▎     | 540/1250 [16:18<26:16,  2.22s/it, loss=0.371]
Epoch 4/4:  43%|████▎     | 540/1250 [16:20<26:16,  2.22s/it, loss=0.285]
Epoch 4/4:  43%|████▎     | 541/1250 [16:20<24:12,  2.05s/it, loss=0.285]
Epoch 4/4:  43%|████▎     | 541/1250 [16:22<24:12,  2.05s/it, loss=0.403]
Epoch 4/4:  43%|████▎     | 542/1250 [16:22<22:36,  1.92s/it, loss=0.403]
Epoch 4/4:  43%|████▎     | 542/1250 [16:24<22:36,  1.92s/it, loss=0.433]
Epoch 4/4:  43%|████▎     | 543/1250 [16:24<24:07,  2.05s/it, loss=0.433]
Epoch 4/4:  43%|████▎     | 543/1250 [16:27<24:07,  2.05s/it, loss=0.411]
Epoch 4/4:  44%|████▎     | 544/1250 [16:27<27:55,  2.37s/it, loss=0.411]
Epoch 4/4:  44%|████▎     | 544/1250 [16:32<27:55,  2.37s/it, loss=0.611]
Epoch 4/4:  44%|████▎     | 545/1250 [16:32<36:13,  3.08s/it, loss=0.611]
Epoch 4/4:  44%|████▎     | 545/1250 [16:34<36:13,  3.08s/it, loss=0.35] 
Epoch 4/4:  44%|████▎     | 546/1250 [16:34<33:23,  2.85s/it, loss=0.35]
Epoch 4/4:  44%|████▎     | 546/1250 [16:36<33:23,  2.85s/it, loss=0.596]
Epoch 4/4:  44%|████▍     | 547/1250 [16:36<31:45,  2.71s/it, loss=0.596]
Epoch 4/4:  44%|████▍     | 547/1250 [16:39<31:45,  2.71s/it, loss=0.651]
Epoch 4/4:  44%|████▍     | 548/1250 [16:39<30:13,  2.58s/it, loss=0.651]
Epoch 4/4:  44%|████▍     | 548/1250 [16:41<30:13,  2.58s/it, loss=0.349]
Epoch 4/4:  44%|████▍     | 549/1250 [16:41<28:23,  2.43s/it, loss=0.349]
Epoch 4/4:  44%|████▍     | 549/1250 [16:44<28:23,  2.43s/it, loss=0.666]
Epoch 4/4:  44%|████▍     | 550/1250 [16:44<29:12,  2.50s/it, loss=0.666]
Epoch 4/4:  44%|████▍     | 550/1250 [16:47<29:12,  2.50s/it, loss=0.508]
Epoch 4/4:  44%|████▍     | 551/1250 [16:47<31:48,  2.73s/it, loss=0.508]
Epoch 4/4:  44%|████▍     | 551/1250 [16:49<31:48,  2.73s/it, loss=0.431]
Epoch 4/4:  44%|████▍     | 552/1250 [16:49<30:07,  2.59s/it, loss=0.431]
Epoch 4/4:  44%|████▍     | 552/1250 [16:52<30:07,  2.59s/it, loss=0.384]
Epoch 4/4:  44%|████▍     | 553/1250 [16:52<29:54,  2.57s/it, loss=0.384]
Epoch 4/4:  44%|████▍     | 553/1250 [16:53<29:54,  2.57s/it, loss=0.27] 
Epoch 4/4:  44%|████▍     | 554/1250 [16:53<26:53,  2.32s/it, loss=0.27]
Epoch 4/4:  44%|████▍     | 554/1250 [16:56<26:53,  2.32s/it, loss=0.526]
Epoch 4/4:  44%|████▍     | 555/1250 [16:56<26:50,  2.32s/it, loss=0.526]
Epoch 4/4:  44%|████▍     | 555/1250 [16:57<26:50,  2.32s/it, loss=0.337]
Epoch 4/4:  44%|████▍     | 556/1250 [16:57<25:14,  2.18s/it, loss=0.337]
Epoch 4/4:  44%|████▍     | 556/1250 [16:59<25:14,  2.18s/it, loss=0.375]
Epoch 4/4:  45%|████▍     | 557/1250 [16:59<24:02,  2.08s/it, loss=0.375]
Epoch 4/4:  45%|████▍     | 557/1250 [17:01<24:02,  2.08s/it, loss=0.385]
Epoch 4/4:  45%|████▍     | 558/1250 [17:01<24:14,  2.10s/it, loss=0.385]
Epoch 4/4:  45%|████▍     | 558/1250 [17:03<24:14,  2.10s/it, loss=0.481]
Epoch 4/4:  45%|████▍     | 559/1250 [17:03<23:57,  2.08s/it, loss=0.481]
Epoch 4/4:  45%|████▍     | 559/1250 [17:05<23:57,  2.08s/it, loss=0.439]
Epoch 4/4:  45%|████▍     | 560/1250 [17:05<23:14,  2.02s/it, loss=0.439]
Epoch 4/4:  45%|████▍     | 560/1250 [17:07<23:14,  2.02s/it, loss=0.344]
Epoch 4/4:  45%|████▍     | 561/1250 [17:07<22:53,  1.99s/it, loss=0.344]
Epoch 4/4:  45%|████▍     | 561/1250 [17:09<22:53,  1.99s/it, loss=0.539]
Epoch 4/4:  45%|████▍     | 562/1250 [17:09<22:39,  1.98s/it, loss=0.539]
Epoch 4/4:  45%|████▍     | 562/1250 [17:11<22:39,  1.98s/it, loss=0.481]
Epoch 4/4:  45%|████▌     | 563/1250 [17:11<22:38,  1.98s/it, loss=0.481]
Epoch 4/4:  45%|████▌     | 563/1250 [17:13<22:38,  1.98s/it, loss=0.369]
Epoch 4/4:  45%|████▌     | 564/1250 [17:13<22:21,  1.96s/it, loss=0.369]
Epoch 4/4:  45%|████▌     | 564/1250 [17:15<22:21,  1.96s/it, loss=0.322]
Epoch 4/4:  45%|████▌     | 565/1250 [17:15<22:32,  1.98s/it, loss=0.322]
Epoch 4/4:  45%|████▌     | 565/1250 [17:17<22:32,  1.98s/it, loss=0.446]
Epoch 4/4:  45%|████▌     | 566/1250 [17:17<22:56,  2.01s/it, loss=0.446]
Epoch 4/4:  45%|████▌     | 566/1250 [17:19<22:56,  2.01s/it, loss=0.58] 
Epoch 4/4:  45%|████▌     | 567/1250 [17:19<21:54,  1.92s/it, loss=0.58]
Epoch 4/4:  45%|████▌     | 567/1250 [17:21<21:54,  1.92s/it, loss=0.401]
Epoch 4/4:  45%|████▌     | 568/1250 [17:21<21:37,  1.90s/it, loss=0.401]
Epoch 4/4:  45%|████▌     | 568/1250 [17:23<21:37,  1.90s/it, loss=0.473]
Epoch 4/4:  46%|████▌     | 569/1250 [17:23<21:41,  1.91s/it, loss=0.473]
Epoch 4/4:  46%|████▌     | 569/1250 [17:24<21:41,  1.91s/it, loss=0.371]
Epoch 4/4:  46%|████▌     | 570/1250 [17:24<20:23,  1.80s/it, loss=0.371]
Epoch 4/4:  46%|████▌     | 570/1250 [17:26<20:23,  1.80s/it, loss=0.347]
Epoch 4/4:  46%|████▌     | 571/1250 [17:26<21:40,  1.92s/it, loss=0.347]
Epoch 4/4:  46%|████▌     | 571/1250 [17:30<21:40,  1.92s/it, loss=0.367]
Epoch 4/4:  46%|████▌     | 572/1250 [17:30<28:14,  2.50s/it, loss=0.367]
Epoch 4/4:  46%|████▌     | 572/1250 [17:35<28:14,  2.50s/it, loss=0.469]
Epoch 4/4:  46%|████▌     | 573/1250 [17:35<34:05,  3.02s/it, loss=0.469]
Epoch 4/4:  46%|████▌     | 573/1250 [17:36<34:05,  3.02s/it, loss=0.528]
Epoch 4/4:  46%|████▌     | 574/1250 [17:36<28:23,  2.52s/it, loss=0.528]
Epoch 4/4:  46%|████▌     | 574/1250 [17:38<28:23,  2.52s/it, loss=0.487]
Epoch 4/4:  46%|████▌     | 575/1250 [17:38<25:50,  2.30s/it, loss=0.487]
Epoch 4/4:  46%|████▌     | 575/1250 [17:39<25:50,  2.30s/it, loss=0.494]
Epoch 4/4:  46%|████▌     | 576/1250 [17:39<24:04,  2.14s/it, loss=0.494]
Epoch 4/4:  46%|████▌     | 576/1250 [17:41<24:04,  2.14s/it, loss=0.41] 
Epoch 4/4:  46%|████▌     | 577/1250 [17:41<23:30,  2.10s/it, loss=0.41]
Epoch 4/4:  46%|████▌     | 577/1250 [17:44<23:30,  2.10s/it, loss=0.412]
Epoch 4/4:  46%|████▌     | 578/1250 [17:44<24:22,  2.18s/it, loss=0.412]
Epoch 4/4:  46%|████▌     | 578/1250 [17:47<24:22,  2.18s/it, loss=0.397]
Epoch 4/4:  46%|████▋     | 579/1250 [17:47<26:19,  2.35s/it, loss=0.397]
Epoch 4/4:  46%|████▋     | 579/1250 [17:48<26:19,  2.35s/it, loss=0.416]
Epoch 4/4:  46%|████▋     | 580/1250 [17:48<23:54,  2.14s/it, loss=0.416]
Epoch 4/4:  46%|████▋     | 580/1250 [17:50<23:54,  2.14s/it, loss=0.434]
Epoch 4/4:  46%|████▋     | 581/1250 [17:50<23:08,  2.08s/it, loss=0.434]
Epoch 4/4:  46%|████▋     | 581/1250 [17:52<23:08,  2.08s/it, loss=0.502]
Epoch 4/4:  47%|████▋     | 582/1250 [17:52<21:10,  1.90s/it, loss=0.502]
Epoch 4/4:  47%|████▋     | 582/1250 [17:53<21:10,  1.90s/it, loss=0.451]
Epoch 4/4:  47%|████▋     | 583/1250 [17:53<20:26,  1.84s/it, loss=0.451]
Epoch 4/4:  47%|████▋     | 583/1250 [17:55<20:26,  1.84s/it, loss=0.446]
Epoch 4/4:  47%|████▋     | 584/1250 [17:55<20:28,  1.85s/it, loss=0.446]
Epoch 4/4:  47%|████▋     | 584/1250 [17:57<20:28,  1.85s/it, loss=0.642]
Epoch 4/4:  47%|████▋     | 585/1250 [17:57<20:43,  1.87s/it, loss=0.642]
Epoch 4/4:  47%|████▋     | 585/1250 [17:59<20:43,  1.87s/it, loss=0.371]
Epoch 4/4:  47%|████▋     | 586/1250 [17:59<21:07,  1.91s/it, loss=0.371]
Epoch 4/4:  47%|████▋     | 586/1250 [18:01<21:07,  1.91s/it, loss=0.627]
Epoch 4/4:  47%|████▋     | 587/1250 [18:01<19:47,  1.79s/it, loss=0.627]
Epoch 4/4:  47%|████▋     | 587/1250 [18:02<19:47,  1.79s/it, loss=0.446]
Epoch 4/4:  47%|████▋     | 588/1250 [18:02<19:31,  1.77s/it, loss=0.446]
Epoch 4/4:  47%|████▋     | 588/1250 [18:04<19:31,  1.77s/it, loss=0.313]
Epoch 4/4:  47%|████▋     | 589/1250 [18:04<18:39,  1.69s/it, loss=0.313]
Epoch 4/4:  47%|████▋     | 589/1250 [18:06<18:39,  1.69s/it, loss=0.413]
Epoch 4/4:  47%|████▋     | 590/1250 [18:06<19:05,  1.74s/it, loss=0.413]
Epoch 4/4:  47%|████▋     | 590/1250 [18:07<19:05,  1.74s/it, loss=0.499]
Epoch 4/4:  47%|████▋     | 591/1250 [18:07<18:42,  1.70s/it, loss=0.499]
Epoch 4/4:  47%|████▋     | 591/1250 [18:09<18:42,  1.70s/it, loss=0.389]
Epoch 4/4:  47%|████▋     | 592/1250 [18:09<19:22,  1.77s/it, loss=0.389]
Epoch 4/4:  47%|████▋     | 592/1250 [18:11<19:22,  1.77s/it, loss=0.479]
Epoch 4/4:  47%|████▋     | 593/1250 [18:11<20:40,  1.89s/it, loss=0.479]
Epoch 4/4:  47%|████▋     | 593/1250 [18:13<20:40,  1.89s/it, loss=0.511]
Epoch 4/4:  48%|████▊     | 594/1250 [18:14<21:11,  1.94s/it, loss=0.511]
Epoch 4/4:  48%|████▊     | 594/1250 [18:16<21:11,  1.94s/it, loss=0.286]
Epoch 4/4:  48%|████▊     | 595/1250 [18:16<21:59,  2.01s/it, loss=0.286]
Epoch 4/4:  48%|████▊     | 595/1250 [18:18<21:59,  2.01s/it, loss=0.484]
Epoch 4/4:  48%|████▊     | 596/1250 [18:18<21:33,  1.98s/it, loss=0.484]
Epoch 4/4:  48%|████▊     | 596/1250 [18:20<21:33,  1.98s/it, loss=0.582]
Epoch 4/4:  48%|████▊     | 597/1250 [18:20<24:31,  2.25s/it, loss=0.582]
Epoch 4/4:  48%|████▊     | 597/1250 [18:24<24:31,  2.25s/it, loss=0.433]
Epoch 4/4:  48%|████▊     | 598/1250 [18:24<27:04,  2.49s/it, loss=0.433]
Epoch 4/4:  48%|████▊     | 598/1250 [18:25<27:04,  2.49s/it, loss=0.504]
Epoch 4/4:  48%|████▊     | 599/1250 [18:25<24:26,  2.25s/it, loss=0.504]
Epoch 4/4:  48%|████▊     | 599/1250 [18:27<24:26,  2.25s/it, loss=0.473]
Epoch 4/4:  48%|████▊     | 600/1250 [18:27<24:00,  2.22s/it, loss=0.473]
Epoch 4/4:  48%|████▊     | 600/1250 [18:29<24:00,  2.22s/it, loss=0.419]
Epoch 4/4:  48%|████▊     | 601/1250 [18:29<22:44,  2.10s/it, loss=0.419]
Epoch 4/4:  48%|████▊     | 601/1250 [18:31<22:44,  2.10s/it, loss=0.379]
Epoch 4/4:  48%|████▊     | 602/1250 [18:31<21:56,  2.03s/it, loss=0.379]
Epoch 4/4:  48%|████▊     | 602/1250 [18:33<21:56,  2.03s/it, loss=0.386]
Epoch 4/4:  48%|████▊     | 603/1250 [18:33<20:26,  1.90s/it, loss=0.386]
Epoch 4/4:  48%|████▊     | 603/1250 [18:35<20:26,  1.90s/it, loss=0.566]
Epoch 4/4:  48%|████▊     | 604/1250 [18:35<20:52,  1.94s/it, loss=0.566]
Epoch 4/4:  48%|████▊     | 604/1250 [18:36<20:52,  1.94s/it, loss=0.566]
Epoch 4/4:  48%|████▊     | 605/1250 [18:36<20:27,  1.90s/it, loss=0.566]
Epoch 4/4:  48%|████▊     | 605/1250 [18:38<20:27,  1.90s/it, loss=0.467]
Epoch 4/4:  48%|████▊     | 606/1250 [18:38<20:07,  1.87s/it, loss=0.467]
Epoch 4/4:  48%|████▊     | 606/1250 [18:40<20:07,  1.87s/it, loss=0.289]
Epoch 4/4:  49%|████▊     | 607/1250 [18:40<18:53,  1.76s/it, loss=0.289]
Epoch 4/4:  49%|████▊     | 607/1250 [18:42<18:53,  1.76s/it, loss=0.363]
Epoch 4/4:  49%|████▊     | 608/1250 [18:42<19:39,  1.84s/it, loss=0.363]
Epoch 4/4:  49%|████▊     | 608/1250 [18:43<19:39,  1.84s/it, loss=0.34] 
Epoch 4/4:  49%|████▊     | 609/1250 [18:43<18:37,  1.74s/it, loss=0.34]
Epoch 4/4:  49%|████▊     | 609/1250 [18:45<18:37,  1.74s/it, loss=0.297]
Epoch 4/4:  49%|████▉     | 610/1250 [18:45<18:07,  1.70s/it, loss=0.297]
Epoch 4/4:  49%|████▉     | 610/1250 [18:47<18:07,  1.70s/it, loss=0.545]
Epoch 4/4:  49%|████▉     | 611/1250 [18:47<18:33,  1.74s/it, loss=0.545]
Epoch 4/4:  49%|████▉     | 611/1250 [18:48<18:33,  1.74s/it, loss=0.324]
Epoch 4/4:  49%|████▉     | 612/1250 [18:48<17:57,  1.69s/it, loss=0.324]
Epoch 4/4:  49%|████▉     | 612/1250 [18:50<17:57,  1.69s/it, loss=0.508]
Epoch 4/4:  49%|████▉     | 613/1250 [18:50<18:44,  1.76s/it, loss=0.508]
Epoch 4/4:  49%|████▉     | 613/1250 [18:52<18:44,  1.76s/it, loss=0.347]
Epoch 4/4:  49%|████▉     | 614/1250 [18:52<18:17,  1.73s/it, loss=0.347]
Epoch 4/4:  49%|████▉     | 614/1250 [18:54<18:17,  1.73s/it, loss=0.563]
Epoch 4/4:  49%|████▉     | 615/1250 [18:54<18:15,  1.72s/it, loss=0.563]
Epoch 4/4:  49%|████▉     | 615/1250 [18:55<18:15,  1.72s/it, loss=0.478]
Epoch 4/4:  49%|████▉     | 616/1250 [18:55<17:38,  1.67s/it, loss=0.478]
Epoch 4/4:  49%|████▉     | 616/1250 [18:57<17:38,  1.67s/it, loss=0.447]
Epoch 4/4:  49%|████▉     | 617/1250 [18:57<18:15,  1.73s/it, loss=0.447]
Epoch 4/4:  49%|████▉     | 617/1250 [18:59<18:15,  1.73s/it, loss=0.617]
Epoch 4/4:  49%|████▉     | 618/1250 [18:59<18:16,  1.73s/it, loss=0.617]
Epoch 4/4:  49%|████▉     | 618/1250 [19:00<18:16,  1.73s/it, loss=0.362]
Epoch 4/4:  50%|████▉     | 619/1250 [19:00<17:26,  1.66s/it, loss=0.362]
Epoch 4/4:  50%|████▉     | 619/1250 [19:02<17:26,  1.66s/it, loss=0.317]
Epoch 4/4:  50%|████▉     | 620/1250 [19:02<17:59,  1.71s/it, loss=0.317]
Epoch 4/4:  50%|████▉     | 620/1250 [19:04<17:59,  1.71s/it, loss=0.656]
Epoch 4/4:  50%|████▉     | 621/1250 [19:04<17:33,  1.67s/it, loss=0.656]
Epoch 4/4:  50%|████▉     | 621/1250 [19:05<17:33,  1.67s/it, loss=0.455]
Epoch 4/4:  50%|████▉     | 622/1250 [19:05<17:33,  1.68s/it, loss=0.455]
Epoch 4/4:  50%|████▉     | 622/1250 [19:07<17:33,  1.68s/it, loss=0.451]
Epoch 4/4:  50%|████▉     | 623/1250 [19:07<17:24,  1.67s/it, loss=0.451]
Epoch 4/4:  50%|████▉     | 623/1250 [19:09<17:24,  1.67s/it, loss=0.507]
Epoch 4/4:  50%|████▉     | 624/1250 [19:09<16:56,  1.62s/it, loss=0.507]
Epoch 4/4:  50%|████▉     | 624/1250 [19:10<16:56,  1.62s/it, loss=0.323]
Epoch 4/4:  50%|█████     | 625/1250 [19:10<17:28,  1.68s/it, loss=0.323]
Epoch 4/4:  50%|█████     | 625/1250 [19:12<17:28,  1.68s/it, loss=0.388]
Epoch 4/4:  50%|█████     | 626/1250 [19:12<16:43,  1.61s/it, loss=0.388]
Epoch 4/4:  50%|█████     | 626/1250 [19:14<16:43,  1.61s/it, loss=0.369]
Epoch 4/4:  50%|█████     | 627/1250 [19:14<17:15,  1.66s/it, loss=0.369]
Epoch 4/4:  50%|█████     | 627/1250 [19:15<17:15,  1.66s/it, loss=0.357]
Epoch 4/4:  50%|█████     | 628/1250 [19:15<16:44,  1.61s/it, loss=0.357]
Epoch 4/4:  50%|█████     | 628/1250 [19:17<16:44,  1.61s/it, loss=0.389]
Epoch 4/4:  50%|█████     | 629/1250 [19:17<16:30,  1.59s/it, loss=0.389]
Epoch 4/4:  50%|█████     | 629/1250 [19:18<16:30,  1.59s/it, loss=0.581]
Epoch 4/4:  50%|█████     | 630/1250 [19:18<16:30,  1.60s/it, loss=0.581]
Epoch 4/4:  50%|█████     | 630/1250 [19:20<16:30,  1.60s/it, loss=0.412]
Epoch 4/4:  50%|█████     | 631/1250 [19:20<16:28,  1.60s/it, loss=0.412]
Epoch 4/4:  50%|█████     | 631/1250 [19:21<16:28,  1.60s/it, loss=0.394]
Epoch 4/4:  51%|█████     | 632/1250 [19:21<16:16,  1.58s/it, loss=0.394]
Epoch 4/4:  51%|█████     | 632/1250 [19:23<16:16,  1.58s/it, loss=0.425]
Epoch 4/4:  51%|█████     | 633/1250 [19:23<16:13,  1.58s/it, loss=0.425]
Epoch 4/4:  51%|█████     | 633/1250 [19:25<16:13,  1.58s/it, loss=0.433]
Epoch 4/4:  51%|█████     | 634/1250 [19:25<16:09,  1.57s/it, loss=0.433]
Epoch 4/4:  51%|█████     | 634/1250 [19:26<16:09,  1.57s/it, loss=0.399]
Epoch 4/4:  51%|█████     | 635/1250 [19:26<16:06,  1.57s/it, loss=0.399]
Epoch 4/4:  51%|█████     | 635/1250 [19:28<16:06,  1.57s/it, loss=0.38] 
Epoch 4/4:  51%|█████     | 636/1250 [19:28<16:23,  1.60s/it, loss=0.38]
Epoch 4/4:  51%|█████     | 636/1250 [19:29<16:23,  1.60s/it, loss=0.519]
Epoch 4/4:  51%|█████     | 637/1250 [19:29<16:10,  1.58s/it, loss=0.519]
Epoch 4/4:  51%|█████     | 637/1250 [19:31<16:10,  1.58s/it, loss=0.481]
Epoch 4/4:  51%|█████     | 638/1250 [19:31<15:54,  1.56s/it, loss=0.481]
Epoch 4/4:  51%|█████     | 638/1250 [19:32<15:54,  1.56s/it, loss=0.433]
Epoch 4/4:  51%|█████     | 639/1250 [19:32<16:01,  1.57s/it, loss=0.433]
Epoch 4/4:  51%|█████     | 639/1250 [19:34<16:01,  1.57s/it, loss=0.331]
Epoch 4/4:  51%|█████     | 640/1250 [19:34<16:59,  1.67s/it, loss=0.331]
Epoch 4/4:  51%|█████     | 640/1250 [19:36<16:59,  1.67s/it, loss=0.633]
Epoch 4/4:  51%|█████▏    | 641/1250 [19:36<16:40,  1.64s/it, loss=0.633]
Epoch 4/4:  51%|█████▏    | 641/1250 [19:38<16:40,  1.64s/it, loss=0.443]
Epoch 4/4:  51%|█████▏    | 642/1250 [19:38<16:54,  1.67s/it, loss=0.443]
Epoch 4/4:  51%|█████▏    | 642/1250 [19:39<16:54,  1.67s/it, loss=0.48] 
Epoch 4/4:  51%|█████▏    | 643/1250 [19:39<16:38,  1.64s/it, loss=0.48]
Epoch 4/4:  51%|█████▏    | 643/1250 [19:41<16:38,  1.64s/it, loss=0.639]
Epoch 4/4:  52%|█████▏    | 644/1250 [19:41<17:32,  1.74s/it, loss=0.639]
Epoch 4/4:  52%|█████▏    | 644/1250 [19:43<17:32,  1.74s/it, loss=0.32] 
Epoch 4/4:  52%|█████▏    | 645/1250 [19:43<17:16,  1.71s/it, loss=0.32]
Epoch 4/4:  52%|█████▏    | 645/1250 [19:45<17:16,  1.71s/it, loss=0.568]
Epoch 4/4:  52%|█████▏    | 646/1250 [19:45<17:33,  1.74s/it, loss=0.568]
Epoch 4/4:  52%|█████▏    | 646/1250 [19:46<17:33,  1.74s/it, loss=0.454]
Epoch 4/4:  52%|█████▏    | 647/1250 [19:46<16:45,  1.67s/it, loss=0.454]
Epoch 4/4:  52%|█████▏    | 647/1250 [19:48<16:45,  1.67s/it, loss=0.574]
Epoch 4/4:  52%|█████▏    | 648/1250 [19:48<17:08,  1.71s/it, loss=0.574]
Epoch 4/4:  52%|█████▏    | 648/1250 [19:50<17:08,  1.71s/it, loss=0.41] 
Epoch 4/4:  52%|█████▏    | 649/1250 [19:50<16:43,  1.67s/it, loss=0.41]
Epoch 4/4:  52%|█████▏    | 649/1250 [19:51<16:43,  1.67s/it, loss=0.526]
Epoch 4/4:  52%|█████▏    | 650/1250 [19:51<16:15,  1.63s/it, loss=0.526]
Epoch 4/4:  52%|█████▏    | 650/1250 [19:53<16:15,  1.63s/it, loss=0.564]
Epoch 4/4:  52%|█████▏    | 651/1250 [19:53<16:29,  1.65s/it, loss=0.564]
Epoch 4/4:  52%|█████▏    | 651/1250 [19:54<16:29,  1.65s/it, loss=0.342]
Epoch 4/4:  52%|█████▏    | 652/1250 [19:54<16:12,  1.63s/it, loss=0.342]
Epoch 4/4:  52%|█████▏    | 652/1250 [19:56<16:12,  1.63s/it, loss=0.462]
Epoch 4/4:  52%|█████▏    | 653/1250 [19:56<16:53,  1.70s/it, loss=0.462]
Epoch 4/4:  52%|█████▏    | 653/1250 [19:58<16:53,  1.70s/it, loss=0.524]
Epoch 4/4:  52%|█████▏    | 654/1250 [19:58<16:27,  1.66s/it, loss=0.524]
Epoch 4/4:  52%|█████▏    | 654/1250 [19:59<16:27,  1.66s/it, loss=0.52] 
Epoch 4/4:  52%|█████▏    | 655/1250 [19:59<16:09,  1.63s/it, loss=0.52]
Epoch 4/4:  52%|█████▏    | 655/1250 [20:02<16:09,  1.63s/it, loss=0.282]
Epoch 4/4:  52%|█████▏    | 656/1250 [20:02<19:56,  2.01s/it, loss=0.282]
Epoch 4/4:  52%|█████▏    | 656/1250 [20:10<19:56,  2.01s/it, loss=0.408]
Epoch 4/4:  53%|█████▎    | 657/1250 [20:10<35:35,  3.60s/it, loss=0.408]
Epoch 4/4:  53%|█████▎    | 657/1250 [20:12<35:35,  3.60s/it, loss=0.404]
Epoch 4/4:  53%|█████▎    | 658/1250 [20:12<31:52,  3.23s/it, loss=0.404]
Epoch 4/4:  53%|█████▎    | 658/1250 [20:14<31:52,  3.23s/it, loss=0.497]
Epoch 4/4:  53%|█████▎    | 659/1250 [20:14<28:19,  2.88s/it, loss=0.497]
Epoch 4/4:  53%|█████▎    | 659/1250 [20:16<28:19,  2.88s/it, loss=0.525]
Epoch 4/4:  53%|█████▎    | 660/1250 [20:16<25:17,  2.57s/it, loss=0.525]
Epoch 4/4:  53%|█████▎    | 660/1250 [20:18<25:17,  2.57s/it, loss=0.424]
Epoch 4/4:  53%|█████▎    | 661/1250 [20:18<25:31,  2.60s/it, loss=0.424]
Epoch 4/4:  53%|█████▎    | 661/1250 [20:21<25:31,  2.60s/it, loss=0.325]
Epoch 4/4:  53%|█████▎    | 662/1250 [20:21<26:13,  2.68s/it, loss=0.325]
Epoch 4/4:  53%|█████▎    | 662/1250 [20:23<26:13,  2.68s/it, loss=0.492]
Epoch 4/4:  53%|█████▎    | 663/1250 [20:23<24:37,  2.52s/it, loss=0.492]
Epoch 4/4:  53%|█████▎    | 663/1250 [20:25<24:37,  2.52s/it, loss=0.486]
Epoch 4/4:  53%|█████▎    | 664/1250 [20:25<23:01,  2.36s/it, loss=0.486]
Epoch 4/4:  53%|█████▎    | 664/1250 [20:28<23:01,  2.36s/it, loss=0.337]
Epoch 4/4:  53%|█████▎    | 665/1250 [20:28<22:42,  2.33s/it, loss=0.337]
Epoch 4/4:  53%|█████▎    | 665/1250 [20:30<22:42,  2.33s/it, loss=0.532]
Epoch 4/4:  53%|█████▎    | 666/1250 [20:30<23:37,  2.43s/it, loss=0.532]
Epoch 4/4:  53%|█████▎    | 666/1250 [20:32<23:37,  2.43s/it, loss=0.592]
Epoch 4/4:  53%|█████▎    | 667/1250 [20:32<21:48,  2.24s/it, loss=0.592]
Epoch 4/4:  53%|█████▎    | 667/1250 [20:35<21:48,  2.24s/it, loss=0.423]
Epoch 4/4:  53%|█████▎    | 668/1250 [20:35<22:47,  2.35s/it, loss=0.423]
Epoch 4/4:  53%|█████▎    | 668/1250 [20:38<22:47,  2.35s/it, loss=0.439]
Epoch 4/4:  54%|█████▎    | 669/1250 [20:38<25:22,  2.62s/it, loss=0.439]
Epoch 4/4:  54%|█████▎    | 669/1250 [20:42<25:22,  2.62s/it, loss=0.403]
Epoch 4/4:  54%|█████▎    | 670/1250 [20:42<29:47,  3.08s/it, loss=0.403]
Epoch 4/4:  54%|█████▎    | 670/1250 [20:45<29:47,  3.08s/it, loss=0.468]
Epoch 4/4:  54%|█████▎    | 671/1250 [20:45<29:21,  3.04s/it, loss=0.468]
Epoch 4/4:  54%|█████▎    | 671/1250 [20:48<29:21,  3.04s/it, loss=0.496]
Epoch 4/4:  54%|█████▍    | 672/1250 [20:48<27:35,  2.86s/it, loss=0.496]
Epoch 4/4:  54%|█████▍    | 672/1250 [20:50<27:35,  2.86s/it, loss=0.429]
Epoch 4/4:  54%|█████▍    | 673/1250 [20:50<25:35,  2.66s/it, loss=0.429]
Epoch 4/4:  54%|█████▍    | 673/1250 [20:52<25:35,  2.66s/it, loss=0.318]
Epoch 4/4:  54%|█████▍    | 674/1250 [20:52<23:10,  2.41s/it, loss=0.318]
Epoch 4/4:  54%|█████▍    | 674/1250 [20:54<23:10,  2.41s/it, loss=0.371]
Epoch 4/4:  54%|█████▍    | 675/1250 [20:54<21:39,  2.26s/it, loss=0.371]
Epoch 4/4:  54%|█████▍    | 675/1250 [20:55<21:39,  2.26s/it, loss=0.45] 
Epoch 4/4:  54%|█████▍    | 676/1250 [20:55<20:28,  2.14s/it, loss=0.45]
Epoch 4/4:  54%|█████▍    | 676/1250 [20:57<20:28,  2.14s/it, loss=0.464]
Epoch 4/4:  54%|█████▍    | 677/1250 [20:57<20:07,  2.11s/it, loss=0.464]
Epoch 4/4:  54%|█████▍    | 677/1250 [20:59<20:07,  2.11s/it, loss=0.317]
Epoch 4/4:  54%|█████▍    | 678/1250 [20:59<19:30,  2.05s/it, loss=0.317]
Epoch 4/4:  54%|█████▍    | 678/1250 [21:01<19:30,  2.05s/it, loss=0.431]
Epoch 4/4:  54%|█████▍    | 679/1250 [21:01<18:31,  1.95s/it, loss=0.431]
Epoch 4/4:  54%|█████▍    | 679/1250 [21:03<18:31,  1.95s/it, loss=0.353]
Epoch 4/4:  54%|█████▍    | 680/1250 [21:03<18:37,  1.96s/it, loss=0.353]
Epoch 4/4:  54%|█████▍    | 680/1250 [21:05<18:37,  1.96s/it, loss=0.306]
Epoch 4/4:  54%|█████▍    | 681/1250 [21:05<17:56,  1.89s/it, loss=0.306]
Epoch 4/4:  54%|█████▍    | 681/1250 [21:06<17:56,  1.89s/it, loss=0.473]
Epoch 4/4:  55%|█████▍    | 682/1250 [21:06<17:11,  1.82s/it, loss=0.473]
Epoch 4/4:  55%|█████▍    | 682/1250 [21:08<17:11,  1.82s/it, loss=0.516]
Epoch 4/4:  55%|█████▍    | 683/1250 [21:08<17:42,  1.87s/it, loss=0.516]
Epoch 4/4:  55%|█████▍    | 683/1250 [21:10<17:42,  1.87s/it, loss=0.379]
Epoch 4/4:  55%|█████▍    | 684/1250 [21:10<16:41,  1.77s/it, loss=0.379]
Epoch 4/4:  55%|█████▍    | 684/1250 [21:12<16:41,  1.77s/it, loss=0.554]
Epoch 4/4:  55%|█████▍    | 685/1250 [21:12<17:31,  1.86s/it, loss=0.554]
Epoch 4/4:  55%|█████▍    | 685/1250 [21:14<17:31,  1.86s/it, loss=0.546]
Epoch 4/4:  55%|█████▍    | 686/1250 [21:14<17:33,  1.87s/it, loss=0.546]
Epoch 4/4:  55%|█████▍    | 686/1250 [21:15<17:33,  1.87s/it, loss=0.448]
Epoch 4/4:  55%|█████▍    | 687/1250 [21:15<16:48,  1.79s/it, loss=0.448]
Epoch 4/4:  55%|█████▍    | 687/1250 [21:18<16:48,  1.79s/it, loss=0.509]
Epoch 4/4:  55%|█████▌    | 688/1250 [21:18<18:34,  1.98s/it, loss=0.509]
Epoch 4/4:  55%|█████▌    | 688/1250 [21:20<18:34,  1.98s/it, loss=0.581]
Epoch 4/4:  55%|█████▌    | 689/1250 [21:20<19:19,  2.07s/it, loss=0.581]
Epoch 4/4:  55%|█████▌    | 689/1250 [21:23<19:19,  2.07s/it, loss=0.479]
Epoch 4/4:  55%|█████▌    | 690/1250 [21:23<20:05,  2.15s/it, loss=0.479]
Epoch 4/4:  55%|█████▌    | 690/1250 [21:25<20:05,  2.15s/it, loss=0.504]
Epoch 4/4:  55%|█████▌    | 691/1250 [21:25<20:03,  2.15s/it, loss=0.504]
Epoch 4/4:  55%|█████▌    | 691/1250 [21:27<20:03,  2.15s/it, loss=0.467]
Epoch 4/4:  55%|█████▌    | 692/1250 [21:27<20:03,  2.16s/it, loss=0.467]
Epoch 4/4:  55%|█████▌    | 692/1250 [21:29<20:03,  2.16s/it, loss=0.391]
Epoch 4/4:  55%|█████▌    | 693/1250 [21:29<19:45,  2.13s/it, loss=0.391]
Epoch 4/4:  55%|█████▌    | 693/1250 [21:31<19:45,  2.13s/it, loss=0.468]
Epoch 4/4:  56%|█████▌    | 694/1250 [21:31<19:17,  2.08s/it, loss=0.468]
Epoch 4/4:  56%|█████▌    | 694/1250 [21:33<19:17,  2.08s/it, loss=0.445]
Epoch 4/4:  56%|█████▌    | 695/1250 [21:33<19:22,  2.09s/it, loss=0.445]
Epoch 4/4:  56%|█████▌    | 695/1250 [21:35<19:22,  2.09s/it, loss=0.633]
Epoch 4/4:  56%|█████▌    | 696/1250 [21:35<18:50,  2.04s/it, loss=0.633]
Epoch 4/4:  56%|█████▌    | 696/1250 [21:37<18:50,  2.04s/it, loss=0.455]
Epoch 4/4:  56%|█████▌    | 697/1250 [21:37<18:31,  2.01s/it, loss=0.455]
Epoch 4/4:  56%|█████▌    | 697/1250 [21:39<18:31,  2.01s/it, loss=0.498]
Epoch 4/4:  56%|█████▌    | 698/1250 [21:39<18:17,  1.99s/it, loss=0.498]
Epoch 4/4:  56%|█████▌    | 698/1250 [21:41<18:17,  1.99s/it, loss=0.345]
Epoch 4/4:  56%|█████▌    | 699/1250 [21:41<17:41,  1.93s/it, loss=0.345]
Epoch 4/4:  56%|█████▌    | 699/1250 [21:44<17:41,  1.93s/it, loss=0.356]
Epoch 4/4:  56%|█████▌    | 700/1250 [21:44<22:00,  2.40s/it, loss=0.356]
Epoch 4/4:  56%|█████▌    | 700/1250 [21:46<22:00,  2.40s/it, loss=0.479]
Epoch 4/4:  56%|█████▌    | 701/1250 [21:46<21:20,  2.33s/it, loss=0.479]
Epoch 4/4:  56%|█████▌    | 701/1250 [21:49<21:20,  2.33s/it, loss=0.572]
Epoch 4/4:  56%|█████▌    | 702/1250 [21:49<21:38,  2.37s/it, loss=0.572]
Epoch 4/4:  56%|█████▌    | 702/1250 [21:51<21:38,  2.37s/it, loss=0.493]
Epoch 4/4:  56%|█████▌    | 703/1250 [21:51<20:43,  2.27s/it, loss=0.493]
Epoch 4/4:  56%|█████▌    | 703/1250 [21:53<20:43,  2.27s/it, loss=0.435]
Epoch 4/4:  56%|█████▋    | 704/1250 [21:53<20:38,  2.27s/it, loss=0.435]
Epoch 4/4:  56%|█████▋    | 704/1250 [21:56<20:38,  2.27s/it, loss=0.515]
Epoch 4/4:  56%|█████▋    | 705/1250 [21:56<22:41,  2.50s/it, loss=0.515]
Epoch 4/4:  56%|█████▋    | 705/1250 [22:00<22:41,  2.50s/it, loss=0.34] 
Epoch 4/4:  56%|█████▋    | 706/1250 [22:00<25:58,  2.86s/it, loss=0.34]
Epoch 4/4:  56%|█████▋    | 706/1250 [22:03<25:58,  2.86s/it, loss=0.438]
Epoch 4/4:  57%|█████▋    | 707/1250 [22:03<25:43,  2.84s/it, loss=0.438]
Epoch 4/4:  57%|█████▋    | 707/1250 [22:05<25:43,  2.84s/it, loss=0.491]
Epoch 4/4:  57%|█████▋    | 708/1250 [22:05<24:26,  2.71s/it, loss=0.491]
Epoch 4/4:  57%|█████▋    | 708/1250 [22:07<24:26,  2.71s/it, loss=0.363]
Epoch 4/4:  57%|█████▋    | 709/1250 [22:07<22:51,  2.53s/it, loss=0.363]
Epoch 4/4:  57%|█████▋    | 709/1250 [22:09<22:51,  2.53s/it, loss=0.501]
Epoch 4/4:  57%|█████▋    | 710/1250 [22:09<22:03,  2.45s/it, loss=0.501]
Epoch 4/4:  57%|█████▋    | 710/1250 [22:11<22:03,  2.45s/it, loss=0.292]
Epoch 4/4:  57%|█████▋    | 711/1250 [22:11<20:45,  2.31s/it, loss=0.292]
Epoch 4/4:  57%|█████▋    | 711/1250 [22:13<20:45,  2.31s/it, loss=0.513]
Epoch 4/4:  57%|█████▋    | 712/1250 [22:13<20:09,  2.25s/it, loss=0.513]
Epoch 4/4:  57%|█████▋    | 712/1250 [22:15<20:09,  2.25s/it, loss=0.344]
Epoch 4/4:  57%|█████▋    | 713/1250 [22:15<19:09,  2.14s/it, loss=0.344]
Epoch 4/4:  57%|█████▋    | 713/1250 [22:17<19:09,  2.14s/it, loss=0.489]
Epoch 4/4:  57%|█████▋    | 714/1250 [22:17<17:55,  2.01s/it, loss=0.489]
Epoch 4/4:  57%|█████▋    | 714/1250 [22:19<17:55,  2.01s/it, loss=0.476]
Epoch 4/4:  57%|█████▋    | 715/1250 [22:19<18:08,  2.03s/it, loss=0.476]
Epoch 4/4:  57%|█████▋    | 715/1250 [22:21<18:08,  2.03s/it, loss=0.464]
Epoch 4/4:  57%|█████▋    | 716/1250 [22:21<17:58,  2.02s/it, loss=0.464]
Epoch 4/4:  57%|█████▋    | 716/1250 [22:23<17:58,  2.02s/it, loss=0.946]
Epoch 4/4:  57%|█████▋    | 717/1250 [22:23<17:56,  2.02s/it, loss=0.946]
Epoch 4/4:  57%|█████▋    | 717/1250 [22:25<17:56,  2.02s/it, loss=0.418]
Epoch 4/4:  57%|█████▋    | 718/1250 [22:25<17:38,  1.99s/it, loss=0.418]
Epoch 4/4:  57%|█████▋    | 718/1250 [22:27<17:38,  1.99s/it, loss=0.491]
Epoch 4/4:  58%|█████▊    | 719/1250 [22:27<17:32,  1.98s/it, loss=0.491]
Epoch 4/4:  58%|█████▊    | 719/1250 [22:29<17:32,  1.98s/it, loss=0.407]
Epoch 4/4:  58%|█████▊    | 720/1250 [22:29<17:45,  2.01s/it, loss=0.407]
Epoch 4/4:  58%|█████▊    | 720/1250 [22:31<17:45,  2.01s/it, loss=0.41] 
Epoch 4/4:  58%|█████▊    | 721/1250 [22:31<17:58,  2.04s/it, loss=0.41]
Epoch 4/4:  58%|█████▊    | 721/1250 [22:33<17:58,  2.04s/it, loss=0.473]
Epoch 4/4:  58%|█████▊    | 722/1250 [22:33<18:06,  2.06s/it, loss=0.473]
Epoch 4/4:  58%|█████▊    | 722/1250 [22:36<18:06,  2.06s/it, loss=0.428]
Epoch 4/4:  58%|█████▊    | 723/1250 [22:36<18:46,  2.14s/it, loss=0.428]
Epoch 4/4:  58%|█████▊    | 723/1250 [22:38<18:46,  2.14s/it, loss=0.426]
Epoch 4/4:  58%|█████▊    | 724/1250 [22:38<18:40,  2.13s/it, loss=0.426]
Epoch 4/4:  58%|█████▊    | 724/1250 [22:40<18:40,  2.13s/it, loss=0.715]
Epoch 4/4:  58%|█████▊    | 725/1250 [22:40<17:47,  2.03s/it, loss=0.715]
Epoch 4/4:  58%|█████▊    | 725/1250 [22:42<17:47,  2.03s/it, loss=0.424]
Epoch 4/4:  58%|█████▊    | 726/1250 [22:42<17:37,  2.02s/it, loss=0.424]
Epoch 4/4:  58%|█████▊    | 726/1250 [22:43<17:37,  2.02s/it, loss=0.341]
Epoch 4/4:  58%|█████▊    | 727/1250 [22:43<17:13,  1.98s/it, loss=0.341]
Epoch 4/4:  58%|█████▊    | 727/1250 [22:45<17:13,  1.98s/it, loss=0.436]
Epoch 4/4:  58%|█████▊    | 728/1250 [22:45<16:19,  1.88s/it, loss=0.436]
Epoch 4/4:  58%|█████▊    | 728/1250 [22:47<16:19,  1.88s/it, loss=0.349]
Epoch 4/4:  58%|█████▊    | 729/1250 [22:47<16:31,  1.90s/it, loss=0.349]
Epoch 4/4:  58%|█████▊    | 729/1250 [22:49<16:31,  1.90s/it, loss=0.437]
Epoch 4/4:  58%|█████▊    | 730/1250 [22:49<16:19,  1.88s/it, loss=0.437]
Epoch 4/4:  58%|█████▊    | 730/1250 [22:51<16:19,  1.88s/it, loss=0.501]
Epoch 4/4:  58%|█████▊    | 731/1250 [22:51<16:06,  1.86s/it, loss=0.501]
Epoch 4/4:  58%|█████▊    | 731/1250 [22:53<16:06,  1.86s/it, loss=0.361]
Epoch 4/4:  59%|█████▊    | 732/1250 [22:53<16:31,  1.92s/it, loss=0.361]
Epoch 4/4:  59%|█████▊    | 732/1250 [22:55<16:31,  1.92s/it, loss=0.521]
Epoch 4/4:  59%|█████▊    | 733/1250 [22:55<16:14,  1.88s/it, loss=0.521]
Epoch 4/4:  59%|█████▊    | 733/1250 [22:56<16:14,  1.88s/it, loss=0.366]
Epoch 4/4:  59%|█████▊    | 734/1250 [22:56<16:09,  1.88s/it, loss=0.366]
Epoch 4/4:  59%|█████▊    | 734/1250 [22:59<16:09,  1.88s/it, loss=0.496]
Epoch 4/4:  59%|█████▉    | 735/1250 [22:59<16:46,  1.95s/it, loss=0.496]
Epoch 4/4:  59%|█████▉    | 735/1250 [23:00<16:46,  1.95s/it, loss=0.4]  
Epoch 4/4:  59%|█████▉    | 736/1250 [23:00<16:01,  1.87s/it, loss=0.4]
Epoch 4/4:  59%|█████▉    | 736/1250 [23:02<16:01,  1.87s/it, loss=0.447]
Epoch 4/4:  59%|█████▉    | 737/1250 [23:02<16:04,  1.88s/it, loss=0.447]
Epoch 4/4:  59%|█████▉    | 737/1250 [23:04<16:04,  1.88s/it, loss=0.371]
Epoch 4/4:  59%|█████▉    | 738/1250 [23:04<16:09,  1.89s/it, loss=0.371]
Epoch 4/4:  59%|█████▉    | 738/1250 [23:06<16:09,  1.89s/it, loss=0.456]
Epoch 4/4:  59%|█████▉    | 739/1250 [23:06<15:54,  1.87s/it, loss=0.456]
Epoch 4/4:  59%|█████▉    | 739/1250 [23:08<15:54,  1.87s/it, loss=0.413]
Epoch 4/4:  59%|█████▉    | 740/1250 [23:08<15:27,  1.82s/it, loss=0.413]
Epoch 4/4:  59%|█████▉    | 740/1250 [23:10<15:27,  1.82s/it, loss=0.64] 
Epoch 4/4:  59%|█████▉    | 741/1250 [23:10<16:21,  1.93s/it, loss=0.64]
Epoch 4/4:  59%|█████▉    | 741/1250 [23:11<16:21,  1.93s/it, loss=0.383]
Epoch 4/4:  59%|█████▉    | 742/1250 [23:11<15:34,  1.84s/it, loss=0.383]
Epoch 4/4:  59%|█████▉    | 742/1250 [23:13<15:34,  1.84s/it, loss=0.45] 
Epoch 4/4:  59%|█████▉    | 743/1250 [23:13<16:06,  1.91s/it, loss=0.45]
Epoch 4/4:  59%|█████▉    | 743/1250 [23:15<16:06,  1.91s/it, loss=0.42]
Epoch 4/4:  60%|█████▉    | 744/1250 [23:15<16:02,  1.90s/it, loss=0.42]
Epoch 4/4:  60%|█████▉    | 744/1250 [23:17<16:02,  1.90s/it, loss=0.455]
Epoch 4/4:  60%|█████▉    | 745/1250 [23:17<15:55,  1.89s/it, loss=0.455]
Epoch 4/4:  60%|█████▉    | 745/1250 [23:19<15:55,  1.89s/it, loss=0.469]
Epoch 4/4:  60%|█████▉    | 746/1250 [23:19<15:40,  1.87s/it, loss=0.469]
Epoch 4/4:  60%|█████▉    | 746/1250 [23:21<15:40,  1.87s/it, loss=0.409]
Epoch 4/4:  60%|█████▉    | 747/1250 [23:21<15:43,  1.88s/it, loss=0.409]
Epoch 4/4:  60%|█████▉    | 747/1250 [23:23<15:43,  1.88s/it, loss=0.378]
Epoch 4/4:  60%|█████▉    | 748/1250 [23:23<15:35,  1.86s/it, loss=0.378]
Epoch 4/4:  60%|█████▉    | 748/1250 [23:25<15:35,  1.86s/it, loss=0.625]
Epoch 4/4:  60%|█████▉    | 749/1250 [23:25<15:56,  1.91s/it, loss=0.625]
Epoch 4/4:  60%|█████▉    | 749/1250 [23:26<15:56,  1.91s/it, loss=0.379]
Epoch 4/4:  60%|██████    | 750/1250 [23:26<15:17,  1.83s/it, loss=0.379]
Epoch 4/4:  60%|██████    | 750/1250 [23:28<15:17,  1.83s/it, loss=0.508]
Epoch 4/4:  60%|██████    | 751/1250 [23:28<15:45,  1.89s/it, loss=0.508]
Epoch 4/4:  60%|██████    | 751/1250 [23:30<15:45,  1.89s/it, loss=0.431]
Epoch 4/4:  60%|██████    | 752/1250 [23:30<15:30,  1.87s/it, loss=0.431]
Epoch 4/4:  60%|██████    | 752/1250 [23:32<15:30,  1.87s/it, loss=0.328]
Epoch 4/4:  60%|██████    | 753/1250 [23:32<14:50,  1.79s/it, loss=0.328]
Epoch 4/4:  60%|██████    | 753/1250 [23:34<14:50,  1.79s/it, loss=0.486]
Epoch 4/4:  60%|██████    | 754/1250 [23:34<15:36,  1.89s/it, loss=0.486]
Epoch 4/4:  60%|██████    | 754/1250 [23:36<15:36,  1.89s/it, loss=0.449]
Epoch 4/4:  60%|██████    | 755/1250 [23:36<15:39,  1.90s/it, loss=0.449]
Epoch 4/4:  60%|██████    | 755/1250 [23:38<15:39,  1.90s/it, loss=0.602]
Epoch 4/4:  60%|██████    | 756/1250 [23:38<15:32,  1.89s/it, loss=0.602]
Epoch 4/4:  60%|██████    | 756/1250 [23:39<15:32,  1.89s/it, loss=0.494]
Epoch 4/4:  61%|██████    | 757/1250 [23:39<14:49,  1.80s/it, loss=0.494]
Epoch 4/4:  61%|██████    | 757/1250 [23:41<14:49,  1.80s/it, loss=0.616]
Epoch 4/4:  61%|██████    | 758/1250 [23:41<15:13,  1.86s/it, loss=0.616]
Epoch 4/4:  61%|██████    | 758/1250 [23:43<15:13,  1.86s/it, loss=0.499]
Epoch 4/4:  61%|██████    | 759/1250 [23:43<14:32,  1.78s/it, loss=0.499]
Epoch 4/4:  61%|██████    | 759/1250 [23:44<14:32,  1.78s/it, loss=0.515]
Epoch 4/4:  61%|██████    | 760/1250 [23:44<13:56,  1.71s/it, loss=0.515]
Epoch 4/4:  61%|██████    | 760/1250 [23:46<13:56,  1.71s/it, loss=0.547]
Epoch 4/4:  61%|██████    | 761/1250 [23:46<14:18,  1.76s/it, loss=0.547]
Epoch 4/4:  61%|██████    | 761/1250 [23:48<14:18,  1.76s/it, loss=0.612]
Epoch 4/4:  61%|██████    | 762/1250 [23:48<14:07,  1.74s/it, loss=0.612]
Epoch 4/4:  61%|██████    | 762/1250 [23:50<14:07,  1.74s/it, loss=0.447]
Epoch 4/4:  61%|██████    | 763/1250 [23:50<14:59,  1.85s/it, loss=0.447]
Epoch 4/4:  61%|██████    | 763/1250 [23:52<14:59,  1.85s/it, loss=0.486]
Epoch 4/4:  61%|██████    | 764/1250 [23:52<14:33,  1.80s/it, loss=0.486]
Epoch 4/4:  61%|██████    | 764/1250 [23:54<14:33,  1.80s/it, loss=0.454]
Epoch 4/4:  61%|██████    | 765/1250 [23:54<15:06,  1.87s/it, loss=0.454]
Epoch 4/4:  61%|██████    | 765/1250 [23:56<15:06,  1.87s/it, loss=0.413]
Epoch 4/4:  61%|██████▏   | 766/1250 [23:56<15:08,  1.88s/it, loss=0.413]
Epoch 4/4:  61%|██████▏   | 766/1250 [23:57<15:08,  1.88s/it, loss=0.524]
Epoch 4/4:  61%|██████▏   | 767/1250 [23:57<14:15,  1.77s/it, loss=0.524]
Epoch 4/4:  61%|██████▏   | 767/1250 [23:59<14:15,  1.77s/it, loss=0.504]
Epoch 4/4:  61%|██████▏   | 768/1250 [23:59<14:44,  1.84s/it, loss=0.504]
Epoch 4/4:  61%|██████▏   | 768/1250 [24:01<14:44,  1.84s/it, loss=0.401]
Epoch 4/4:  62%|██████▏   | 769/1250 [24:01<14:48,  1.85s/it, loss=0.401]
Epoch 4/4:  62%|██████▏   | 769/1250 [24:03<14:48,  1.85s/it, loss=0.363]
Epoch 4/4:  62%|██████▏   | 770/1250 [24:03<14:46,  1.85s/it, loss=0.363]
Epoch 4/4:  62%|██████▏   | 770/1250 [24:05<14:46,  1.85s/it, loss=0.447]
Epoch 4/4:  62%|██████▏   | 771/1250 [24:05<14:29,  1.81s/it, loss=0.447]
Epoch 4/4:  62%|██████▏   | 771/1250 [24:07<14:29,  1.81s/it, loss=0.474]
Epoch 4/4:  62%|██████▏   | 772/1250 [24:07<14:47,  1.86s/it, loss=0.474]
Epoch 4/4:  62%|██████▏   | 772/1250 [24:09<14:47,  1.86s/it, loss=0.483]
Epoch 4/4:  62%|██████▏   | 773/1250 [24:09<14:52,  1.87s/it, loss=0.483]
Epoch 4/4:  62%|██████▏   | 773/1250 [24:10<14:52,  1.87s/it, loss=0.421]
Epoch 4/4:  62%|██████▏   | 774/1250 [24:10<14:07,  1.78s/it, loss=0.421]
Epoch 4/4:  62%|██████▏   | 774/1250 [24:12<14:07,  1.78s/it, loss=0.341]
Epoch 4/4:  62%|██████▏   | 775/1250 [24:12<13:42,  1.73s/it, loss=0.341]
Epoch 4/4:  62%|██████▏   | 775/1250 [24:14<13:42,  1.73s/it, loss=0.616]
Epoch 4/4:  62%|██████▏   | 776/1250 [24:14<14:28,  1.83s/it, loss=0.616]
Epoch 4/4:  62%|██████▏   | 776/1250 [24:15<14:28,  1.83s/it, loss=0.53] 
Epoch 4/4:  62%|██████▏   | 777/1250 [24:15<13:40,  1.73s/it, loss=0.53]
Epoch 4/4:  62%|██████▏   | 777/1250 [24:17<13:40,  1.73s/it, loss=0.387]
Epoch 4/4:  62%|██████▏   | 778/1250 [24:17<13:35,  1.73s/it, loss=0.387]
Epoch 4/4:  62%|██████▏   | 778/1250 [24:19<13:35,  1.73s/it, loss=0.405]
Epoch 4/4:  62%|██████▏   | 779/1250 [24:19<14:20,  1.83s/it, loss=0.405]
Epoch 4/4:  62%|██████▏   | 779/1250 [24:21<14:20,  1.83s/it, loss=0.627]
Epoch 4/4:  62%|██████▏   | 780/1250 [24:21<14:34,  1.86s/it, loss=0.627]
Epoch 4/4:  62%|██████▏   | 780/1250 [24:23<14:34,  1.86s/it, loss=0.448]
Epoch 4/4:  62%|██████▏   | 781/1250 [24:23<14:05,  1.80s/it, loss=0.448]
Epoch 4/4:  62%|██████▏   | 781/1250 [24:25<14:05,  1.80s/it, loss=0.261]
Epoch 4/4:  63%|██████▎   | 782/1250 [24:25<14:32,  1.86s/it, loss=0.261]
Epoch 4/4:  63%|██████▎   | 782/1250 [24:26<14:32,  1.86s/it, loss=0.559]
Epoch 4/4:  63%|██████▎   | 783/1250 [24:26<14:20,  1.84s/it, loss=0.559]
Epoch 4/4:  63%|██████▎   | 783/1250 [24:28<14:20,  1.84s/it, loss=0.372]
Epoch 4/4:  63%|██████▎   | 784/1250 [24:28<14:25,  1.86s/it, loss=0.372]
Epoch 4/4:  63%|██████▎   | 784/1250 [24:30<14:25,  1.86s/it, loss=0.535]
Epoch 4/4:  63%|██████▎   | 785/1250 [24:30<14:26,  1.86s/it, loss=0.535]
Epoch 4/4:  63%|██████▎   | 785/1250 [24:32<14:26,  1.86s/it, loss=0.582]
Epoch 4/4:  63%|██████▎   | 786/1250 [24:32<15:10,  1.96s/it, loss=0.582]
Epoch 4/4:  63%|██████▎   | 786/1250 [24:35<15:10,  1.96s/it, loss=0.4]  
Epoch 4/4:  63%|██████▎   | 787/1250 [24:35<15:41,  2.03s/it, loss=0.4]
Epoch 4/4:  63%|██████▎   | 787/1250 [24:37<15:41,  2.03s/it, loss=0.467]
Epoch 4/4:  63%|██████▎   | 788/1250 [24:37<15:35,  2.03s/it, loss=0.467]
Epoch 4/4:  63%|██████▎   | 788/1250 [24:39<15:35,  2.03s/it, loss=0.555]
Epoch 4/4:  63%|██████▎   | 789/1250 [24:39<16:00,  2.08s/it, loss=0.555]
Epoch 4/4:  63%|██████▎   | 789/1250 [24:41<16:00,  2.08s/it, loss=0.382]
Epoch 4/4:  63%|██████▎   | 790/1250 [24:41<16:09,  2.11s/it, loss=0.382]
Epoch 4/4:  63%|██████▎   | 790/1250 [24:43<16:09,  2.11s/it, loss=0.42] 
Epoch 4/4:  63%|██████▎   | 791/1250 [24:43<15:41,  2.05s/it, loss=0.42]
Epoch 4/4:  63%|██████▎   | 791/1250 [24:45<15:41,  2.05s/it, loss=0.501]
Epoch 4/4:  63%|██████▎   | 792/1250 [24:45<15:01,  1.97s/it, loss=0.501]
Epoch 4/4:  63%|██████▎   | 792/1250 [24:47<15:01,  1.97s/it, loss=0.509]
Epoch 4/4:  63%|██████▎   | 793/1250 [24:47<14:51,  1.95s/it, loss=0.509]
Epoch 4/4:  63%|██████▎   | 793/1250 [24:48<14:51,  1.95s/it, loss=0.354]
Epoch 4/4:  64%|██████▎   | 794/1250 [24:48<14:31,  1.91s/it, loss=0.354]
Epoch 4/4:  64%|██████▎   | 794/1250 [24:50<14:31,  1.91s/it, loss=0.432]
Epoch 4/4:  64%|██████▎   | 795/1250 [24:50<14:07,  1.86s/it, loss=0.432]
Epoch 4/4:  64%|██████▎   | 795/1250 [24:52<14:07,  1.86s/it, loss=0.494]
Epoch 4/4:  64%|██████▎   | 796/1250 [24:52<13:41,  1.81s/it, loss=0.494]
Epoch 4/4:  64%|██████▎   | 796/1250 [24:54<13:41,  1.81s/it, loss=0.614]
Epoch 4/4:  64%|██████▍   | 797/1250 [24:54<13:46,  1.82s/it, loss=0.614]
Epoch 4/4:  64%|██████▍   | 797/1250 [24:56<13:46,  1.82s/it, loss=0.39] 
Epoch 4/4:  64%|██████▍   | 798/1250 [24:56<14:20,  1.90s/it, loss=0.39]
Epoch 4/4:  64%|██████▍   | 798/1250 [24:57<14:20,  1.90s/it, loss=0.523]
Epoch 4/4:  64%|██████▍   | 799/1250 [24:57<13:17,  1.77s/it, loss=0.523]
Epoch 4/4:  64%|██████▍   | 799/1250 [24:59<13:17,  1.77s/it, loss=0.489]
Epoch 4/4:  64%|██████▍   | 800/1250 [24:59<13:50,  1.84s/it, loss=0.489]
Epoch 4/4:  64%|██████▍   | 800/1250 [25:01<13:50,  1.84s/it, loss=0.59] 
Epoch 4/4:  64%|██████▍   | 801/1250 [25:01<13:20,  1.78s/it, loss=0.59]
Epoch 4/4:  64%|██████▍   | 801/1250 [25:03<13:20,  1.78s/it, loss=0.464]
Epoch 4/4:  64%|██████▍   | 802/1250 [25:03<13:42,  1.84s/it, loss=0.464]
Epoch 4/4:  64%|██████▍   | 802/1250 [25:05<13:42,  1.84s/it, loss=0.48] 
Epoch 4/4:  64%|██████▍   | 803/1250 [25:05<13:33,  1.82s/it, loss=0.48]
Epoch 4/4:  64%|██████▍   | 803/1250 [25:06<13:33,  1.82s/it, loss=0.372]
Epoch 4/4:  64%|██████▍   | 804/1250 [25:06<13:05,  1.76s/it, loss=0.372]
Epoch 4/4:  64%|██████▍   | 804/1250 [25:08<13:05,  1.76s/it, loss=0.415]
Epoch 4/4:  64%|██████▍   | 805/1250 [25:08<12:41,  1.71s/it, loss=0.415]
Epoch 4/4:  64%|██████▍   | 805/1250 [25:10<12:41,  1.71s/it, loss=0.528]
Epoch 4/4:  64%|██████▍   | 806/1250 [25:10<13:00,  1.76s/it, loss=0.528]
Epoch 4/4:  64%|██████▍   | 806/1250 [25:11<13:00,  1.76s/it, loss=0.496]
Epoch 4/4:  65%|██████▍   | 807/1250 [25:11<12:39,  1.71s/it, loss=0.496]
Epoch 4/4:  65%|██████▍   | 807/1250 [25:13<12:39,  1.71s/it, loss=0.405]
Epoch 4/4:  65%|██████▍   | 808/1250 [25:13<12:27,  1.69s/it, loss=0.405]
Epoch 4/4:  65%|██████▍   | 808/1250 [25:15<12:27,  1.69s/it, loss=0.481]
Epoch 4/4:  65%|██████▍   | 809/1250 [25:15<13:15,  1.80s/it, loss=0.481]
Epoch 4/4:  65%|██████▍   | 809/1250 [25:17<13:15,  1.80s/it, loss=0.424]
Epoch 4/4:  65%|██████▍   | 810/1250 [25:17<12:42,  1.73s/it, loss=0.424]
Epoch 4/4:  65%|██████▍   | 810/1250 [25:19<12:42,  1.73s/it, loss=0.419]
Epoch 4/4:  65%|██████▍   | 811/1250 [25:19<13:09,  1.80s/it, loss=0.419]
Epoch 4/4:  65%|██████▍   | 811/1250 [25:20<13:09,  1.80s/it, loss=0.316]
Epoch 4/4:  65%|██████▍   | 812/1250 [25:20<13:11,  1.81s/it, loss=0.316]
Epoch 4/4:  65%|██████▍   | 812/1250 [25:23<13:11,  1.81s/it, loss=0.43] 
Epoch 4/4:  65%|██████▌   | 813/1250 [25:23<14:12,  1.95s/it, loss=0.43]
Epoch 4/4:  65%|██████▌   | 813/1250 [25:25<14:12,  1.95s/it, loss=0.336]
Epoch 4/4:  65%|██████▌   | 814/1250 [25:25<14:41,  2.02s/it, loss=0.336]
Epoch 4/4:  65%|██████▌   | 814/1250 [25:27<14:41,  2.02s/it, loss=0.463]
Epoch 4/4:  65%|██████▌   | 815/1250 [25:27<15:03,  2.08s/it, loss=0.463]
Epoch 4/4:  65%|██████▌   | 815/1250 [25:30<15:03,  2.08s/it, loss=0.445]
Epoch 4/4:  65%|██████▌   | 816/1250 [25:30<17:41,  2.45s/it, loss=0.445]
Epoch 4/4:  65%|██████▌   | 816/1250 [25:34<17:41,  2.45s/it, loss=0.433]
Epoch 4/4:  65%|██████▌   | 817/1250 [25:34<19:31,  2.71s/it, loss=0.433]
Epoch 4/4:  65%|██████▌   | 817/1250 [25:36<19:31,  2.71s/it, loss=0.337]
Epoch 4/4:  65%|██████▌   | 818/1250 [25:36<19:20,  2.69s/it, loss=0.337]
Epoch 4/4:  65%|██████▌   | 818/1250 [25:38<19:20,  2.69s/it, loss=0.258]
Epoch 4/4:  66%|██████▌   | 819/1250 [25:38<17:23,  2.42s/it, loss=0.258]
Epoch 4/4:  66%|██████▌   | 819/1250 [25:41<17:23,  2.42s/it, loss=0.706]
Epoch 4/4:  66%|██████▌   | 820/1250 [25:41<17:12,  2.40s/it, loss=0.706]
Epoch 4/4:  66%|██████▌   | 820/1250 [25:43<17:12,  2.40s/it, loss=0.424]
Epoch 4/4:  66%|██████▌   | 821/1250 [25:43<18:18,  2.56s/it, loss=0.424]
Epoch 4/4:  66%|██████▌   | 821/1250 [25:46<18:18,  2.56s/it, loss=0.521]
Epoch 4/4:  66%|██████▌   | 822/1250 [25:46<19:04,  2.67s/it, loss=0.521]
Epoch 4/4:  66%|██████▌   | 822/1250 [25:50<19:04,  2.67s/it, loss=0.603]
Epoch 4/4:  66%|██████▌   | 823/1250 [25:50<20:03,  2.82s/it, loss=0.603]
Epoch 4/4:  66%|██████▌   | 823/1250 [25:52<20:03,  2.82s/it, loss=0.564]
Epoch 4/4:  66%|██████▌   | 824/1250 [25:52<19:25,  2.74s/it, loss=0.564]
Epoch 4/4:  66%|██████▌   | 824/1250 [25:55<19:25,  2.74s/it, loss=0.293]
Epoch 4/4:  66%|██████▌   | 825/1250 [25:55<19:19,  2.73s/it, loss=0.293]
Epoch 4/4:  66%|██████▌   | 825/1250 [25:57<19:19,  2.73s/it, loss=0.643]
Epoch 4/4:  66%|██████▌   | 826/1250 [25:57<18:15,  2.58s/it, loss=0.643]
Epoch 4/4:  66%|██████▌   | 826/1250 [25:59<18:15,  2.58s/it, loss=0.375]
Epoch 4/4:  66%|██████▌   | 827/1250 [25:59<16:46,  2.38s/it, loss=0.375]
Epoch 4/4:  66%|██████▌   | 827/1250 [26:01<16:46,  2.38s/it, loss=0.337]
Epoch 4/4:  66%|██████▌   | 828/1250 [26:01<15:06,  2.15s/it, loss=0.337]
Epoch 4/4:  66%|██████▌   | 828/1250 [26:02<15:06,  2.15s/it, loss=0.435]
Epoch 4/4:  66%|██████▋   | 829/1250 [26:02<13:57,  1.99s/it, loss=0.435]
Epoch 4/4:  66%|██████▋   | 829/1250 [26:04<13:57,  1.99s/it, loss=0.397]
Epoch 4/4:  66%|██████▋   | 830/1250 [26:04<13:09,  1.88s/it, loss=0.397]
Epoch 4/4:  66%|██████▋   | 830/1250 [26:06<13:09,  1.88s/it, loss=0.293]
Epoch 4/4:  66%|██████▋   | 831/1250 [26:06<12:57,  1.86s/it, loss=0.293]
Epoch 4/4:  66%|██████▋   | 831/1250 [26:07<12:57,  1.86s/it, loss=0.549]
Epoch 4/4:  67%|██████▋   | 832/1250 [26:07<12:25,  1.78s/it, loss=0.549]
Epoch 4/4:  67%|██████▋   | 832/1250 [26:09<12:25,  1.78s/it, loss=0.33] 
Epoch 4/4:  67%|██████▋   | 833/1250 [26:09<11:54,  1.71s/it, loss=0.33]
Epoch 4/4:  67%|██████▋   | 833/1250 [26:11<11:54,  1.71s/it, loss=0.567]
Epoch 4/4:  67%|██████▋   | 834/1250 [26:11<12:09,  1.75s/it, loss=0.567]
Epoch 4/4:  67%|██████▋   | 834/1250 [26:12<12:09,  1.75s/it, loss=0.308]
Epoch 4/4:  67%|██████▋   | 835/1250 [26:12<11:51,  1.71s/it, loss=0.308]
Epoch 4/4:  67%|██████▋   | 835/1250 [26:14<11:51,  1.71s/it, loss=0.444]
Epoch 4/4:  67%|██████▋   | 836/1250 [26:14<12:21,  1.79s/it, loss=0.444]
Epoch 4/4:  67%|██████▋   | 836/1250 [26:16<12:21,  1.79s/it, loss=0.375]
Epoch 4/4:  67%|██████▋   | 837/1250 [26:16<12:54,  1.87s/it, loss=0.375]
Epoch 4/4:  67%|██████▋   | 837/1250 [26:18<12:54,  1.87s/it, loss=0.435]
Epoch 4/4:  67%|██████▋   | 838/1250 [26:18<13:26,  1.96s/it, loss=0.435]
Epoch 4/4:  67%|██████▋   | 838/1250 [26:21<13:26,  1.96s/it, loss=0.442]
Epoch 4/4:  67%|██████▋   | 839/1250 [26:21<13:56,  2.03s/it, loss=0.442]
Epoch 4/4:  67%|██████▋   | 839/1250 [26:23<13:56,  2.03s/it, loss=0.396]
Epoch 4/4:  67%|██████▋   | 840/1250 [26:23<13:31,  1.98s/it, loss=0.396]
Epoch 4/4:  67%|██████▋   | 840/1250 [26:24<13:31,  1.98s/it, loss=0.3]  
Epoch 4/4:  67%|██████▋   | 841/1250 [26:24<13:02,  1.91s/it, loss=0.3]
Epoch 4/4:  67%|██████▋   | 841/1250 [26:26<13:02,  1.91s/it, loss=0.522]
Epoch 4/4:  67%|██████▋   | 842/1250 [26:26<12:59,  1.91s/it, loss=0.522]
Epoch 4/4:  67%|██████▋   | 842/1250 [26:28<12:59,  1.91s/it, loss=0.437]
Epoch 4/4:  67%|██████▋   | 843/1250 [26:28<12:11,  1.80s/it, loss=0.437]
Epoch 4/4:  67%|██████▋   | 843/1250 [26:30<12:11,  1.80s/it, loss=0.36] 
Epoch 4/4:  68%|██████▊   | 844/1250 [26:30<13:03,  1.93s/it, loss=0.36]
Epoch 4/4:  68%|██████▊   | 844/1250 [26:32<13:03,  1.93s/it, loss=0.484]
Epoch 4/4:  68%|██████▊   | 845/1250 [26:32<13:03,  1.93s/it, loss=0.484]
Epoch 4/4:  68%|██████▊   | 845/1250 [26:34<13:03,  1.93s/it, loss=0.329]
Epoch 4/4:  68%|██████▊   | 846/1250 [26:34<13:10,  1.96s/it, loss=0.329]
Epoch 4/4:  68%|██████▊   | 846/1250 [26:36<13:10,  1.96s/it, loss=0.598]
Epoch 4/4:  68%|██████▊   | 847/1250 [26:36<13:12,  1.97s/it, loss=0.598]
Epoch 4/4:  68%|██████▊   | 847/1250 [26:38<13:12,  1.97s/it, loss=0.446]
Epoch 4/4:  68%|██████▊   | 848/1250 [26:38<13:50,  2.07s/it, loss=0.446]
Epoch 4/4:  68%|██████▊   | 848/1250 [26:41<13:50,  2.07s/it, loss=0.38] 
Epoch 4/4:  68%|██████▊   | 849/1250 [26:41<14:27,  2.16s/it, loss=0.38]
Epoch 4/4:  68%|██████▊   | 849/1250 [26:43<14:27,  2.16s/it, loss=0.408]
Epoch 4/4:  68%|██████▊   | 850/1250 [26:43<15:03,  2.26s/it, loss=0.408]
Epoch 4/4:  68%|██████▊   | 850/1250 [26:49<15:03,  2.26s/it, loss=0.36] 
Epoch 4/4:  68%|██████▊   | 851/1250 [26:49<21:34,  3.24s/it, loss=0.36]
Epoch 4/4:  68%|██████▊   | 851/1250 [26:50<21:34,  3.24s/it, loss=0.495]
Epoch 4/4:  68%|██████▊   | 852/1250 [26:50<18:29,  2.79s/it, loss=0.495]
Epoch 4/4:  68%|██████▊   | 852/1250 [26:52<18:29,  2.79s/it, loss=0.453]
Epoch 4/4:  68%|██████▊   | 853/1250 [26:52<16:49,  2.54s/it, loss=0.453]
Epoch 4/4:  68%|██████▊   | 853/1250 [26:54<16:49,  2.54s/it, loss=0.461]
Epoch 4/4:  68%|██████▊   | 854/1250 [26:54<15:17,  2.32s/it, loss=0.461]
Epoch 4/4:  68%|██████▊   | 854/1250 [26:56<15:17,  2.32s/it, loss=0.487]
Epoch 4/4:  68%|██████▊   | 855/1250 [26:56<14:39,  2.23s/it, loss=0.487]
Epoch 4/4:  68%|██████▊   | 855/1250 [26:58<14:39,  2.23s/it, loss=0.439]
Epoch 4/4:  68%|██████▊   | 856/1250 [26:58<14:23,  2.19s/it, loss=0.439]
Epoch 4/4:  68%|██████▊   | 856/1250 [27:00<14:23,  2.19s/it, loss=0.463]
Epoch 4/4:  69%|██████▊   | 857/1250 [27:00<14:09,  2.16s/it, loss=0.463]
Epoch 4/4:  69%|██████▊   | 857/1250 [27:02<14:09,  2.16s/it, loss=0.637]
Epoch 4/4:  69%|██████▊   | 858/1250 [27:02<13:33,  2.07s/it, loss=0.637]
Epoch 4/4:  69%|██████▊   | 858/1250 [27:04<13:33,  2.07s/it, loss=0.473]
Epoch 4/4:  69%|██████▊   | 859/1250 [27:04<13:27,  2.07s/it, loss=0.473]
Epoch 4/4:  69%|██████▊   | 859/1250 [27:06<13:27,  2.07s/it, loss=0.456]
Epoch 4/4:  69%|██████▉   | 860/1250 [27:06<13:10,  2.03s/it, loss=0.456]
Epoch 4/4:  69%|██████▉   | 860/1250 [27:08<13:10,  2.03s/it, loss=0.366]
Epoch 4/4:  69%|██████▉   | 861/1250 [27:08<12:10,  1.88s/it, loss=0.366]
Epoch 4/4:  69%|██████▉   | 861/1250 [27:09<12:10,  1.88s/it, loss=0.511]
Epoch 4/4:  69%|██████▉   | 862/1250 [27:09<11:37,  1.80s/it, loss=0.511]
Epoch 4/4:  69%|██████▉   | 862/1250 [27:11<11:37,  1.80s/it, loss=0.439]
Epoch 4/4:  69%|██████▉   | 863/1250 [27:11<11:11,  1.74s/it, loss=0.439]
Epoch 4/4:  69%|██████▉   | 863/1250 [27:13<11:11,  1.74s/it, loss=0.621]
Epoch 4/4:  69%|██████▉   | 864/1250 [27:13<11:10,  1.74s/it, loss=0.621]
Epoch 4/4:  69%|██████▉   | 864/1250 [27:14<11:10,  1.74s/it, loss=0.473]
Epoch 4/4:  69%|██████▉   | 865/1250 [27:14<10:51,  1.69s/it, loss=0.473]
Epoch 4/4:  69%|██████▉   | 865/1250 [27:16<10:51,  1.69s/it, loss=0.637]
Epoch 4/4:  69%|██████▉   | 866/1250 [27:16<11:27,  1.79s/it, loss=0.637]
Epoch 4/4:  69%|██████▉   | 866/1250 [27:18<11:27,  1.79s/it, loss=0.379]
Epoch 4/4:  69%|██████▉   | 867/1250 [27:18<11:36,  1.82s/it, loss=0.379]
Epoch 4/4:  69%|██████▉   | 867/1250 [27:20<11:36,  1.82s/it, loss=0.349]
Epoch 4/4:  69%|██████▉   | 868/1250 [27:20<11:05,  1.74s/it, loss=0.349]
Epoch 4/4:  69%|██████▉   | 868/1250 [27:22<11:05,  1.74s/it, loss=0.406]
Epoch 4/4:  70%|██████▉   | 869/1250 [27:22<11:34,  1.82s/it, loss=0.406]
Epoch 4/4:  70%|██████▉   | 869/1250 [27:24<11:34,  1.82s/it, loss=0.55] 
Epoch 4/4:  70%|██████▉   | 870/1250 [27:24<11:42,  1.85s/it, loss=0.55]
Epoch 4/4:  70%|██████▉   | 870/1250 [27:25<11:42,  1.85s/it, loss=0.531]
Epoch 4/4:  70%|██████▉   | 871/1250 [27:25<11:30,  1.82s/it, loss=0.531]
Epoch 4/4:  70%|██████▉   | 871/1250 [27:27<11:30,  1.82s/it, loss=0.376]
Epoch 4/4:  70%|██████▉   | 872/1250 [27:27<11:50,  1.88s/it, loss=0.376]
Epoch 4/4:  70%|██████▉   | 872/1250 [27:29<11:50,  1.88s/it, loss=0.429]
Epoch 4/4:  70%|██████▉   | 873/1250 [27:29<11:45,  1.87s/it, loss=0.429]
Epoch 4/4:  70%|██████▉   | 873/1250 [27:31<11:45,  1.87s/it, loss=0.359]
Epoch 4/4:  70%|██████▉   | 874/1250 [27:31<11:43,  1.87s/it, loss=0.359]
Epoch 4/4:  70%|██████▉   | 874/1250 [27:33<11:43,  1.87s/it, loss=0.361]
Epoch 4/4:  70%|███████   | 875/1250 [27:33<11:11,  1.79s/it, loss=0.361]
Epoch 4/4:  70%|███████   | 875/1250 [27:35<11:11,  1.79s/it, loss=0.568]
Epoch 4/4:  70%|███████   | 876/1250 [27:35<11:26,  1.84s/it, loss=0.568]
Epoch 4/4:  70%|███████   | 876/1250 [27:37<11:26,  1.84s/it, loss=0.492]
Epoch 4/4:  70%|███████   | 877/1250 [27:37<11:34,  1.86s/it, loss=0.492]
Epoch 4/4:  70%|███████   | 877/1250 [27:38<11:34,  1.86s/it, loss=0.62] 
Epoch 4/4:  70%|███████   | 878/1250 [27:38<11:28,  1.85s/it, loss=0.62]
Epoch 4/4:  70%|███████   | 878/1250 [27:40<11:28,  1.85s/it, loss=0.355]
Epoch 4/4:  70%|███████   | 879/1250 [27:40<11:24,  1.84s/it, loss=0.355]
Epoch 4/4:  70%|███████   | 879/1250 [27:43<11:24,  1.84s/it, loss=0.5]  
Epoch 4/4:  70%|███████   | 880/1250 [27:43<12:24,  2.01s/it, loss=0.5]
Epoch 4/4:  70%|███████   | 880/1250 [27:44<12:24,  2.01s/it, loss=0.439]
Epoch 4/4:  70%|███████   | 881/1250 [27:44<11:51,  1.93s/it, loss=0.439]
Epoch 4/4:  70%|███████   | 881/1250 [27:47<11:51,  1.93s/it, loss=0.556]
Epoch 4/4:  71%|███████   | 882/1250 [27:47<12:39,  2.07s/it, loss=0.556]
Epoch 4/4:  71%|███████   | 882/1250 [27:48<12:39,  2.07s/it, loss=0.622]
Epoch 4/4:  71%|███████   | 883/1250 [27:48<11:53,  1.94s/it, loss=0.622]
Epoch 4/4:  71%|███████   | 883/1250 [27:50<11:53,  1.94s/it, loss=0.344]
Epoch 4/4:  71%|███████   | 884/1250 [27:50<11:19,  1.86s/it, loss=0.344]
Epoch 4/4:  71%|███████   | 884/1250 [27:52<11:19,  1.86s/it, loss=0.479]
Epoch 4/4:  71%|███████   | 885/1250 [27:52<11:24,  1.87s/it, loss=0.479]
Epoch 4/4:  71%|███████   | 885/1250 [27:54<11:24,  1.87s/it, loss=0.469]
Epoch 4/4:  71%|███████   | 886/1250 [27:54<11:14,  1.85s/it, loss=0.469]
Epoch 4/4:  71%|███████   | 886/1250 [27:56<11:14,  1.85s/it, loss=0.409]
Epoch 4/4:  71%|███████   | 887/1250 [27:56<11:23,  1.88s/it, loss=0.409]
Epoch 4/4:  71%|███████   | 887/1250 [27:58<11:23,  1.88s/it, loss=0.45] 
Epoch 4/4:  71%|███████   | 888/1250 [27:58<11:32,  1.91s/it, loss=0.45]
Epoch 4/4:  71%|███████   | 888/1250 [27:59<11:32,  1.91s/it, loss=0.405]
Epoch 4/4:  71%|███████   | 889/1250 [27:59<11:06,  1.85s/it, loss=0.405]
Epoch 4/4:  71%|███████   | 889/1250 [28:01<11:06,  1.85s/it, loss=0.519]
Epoch 4/4:  71%|███████   | 890/1250 [28:01<11:16,  1.88s/it, loss=0.519]
Epoch 4/4:  71%|███████   | 890/1250 [28:03<11:16,  1.88s/it, loss=0.412]
Epoch 4/4:  71%|███████▏  | 891/1250 [28:03<10:47,  1.80s/it, loss=0.412]
Epoch 4/4:  71%|███████▏  | 891/1250 [28:05<10:47,  1.80s/it, loss=0.411]
Epoch 4/4:  71%|███████▏  | 892/1250 [28:05<11:07,  1.87s/it, loss=0.411]
Epoch 4/4:  71%|███████▏  | 892/1250 [28:07<11:07,  1.87s/it, loss=0.513]
Epoch 4/4:  71%|███████▏  | 893/1250 [28:07<11:10,  1.88s/it, loss=0.513]
Epoch 4/4:  71%|███████▏  | 893/1250 [28:09<11:10,  1.88s/it, loss=0.39] 
Epoch 4/4:  72%|███████▏  | 894/1250 [28:09<10:53,  1.83s/it, loss=0.39]
Epoch 4/4:  72%|███████▏  | 894/1250 [28:11<10:53,  1.83s/it, loss=0.469]
Epoch 4/4:  72%|███████▏  | 895/1250 [28:11<10:54,  1.84s/it, loss=0.469]
Epoch 4/4:  72%|███████▏  | 895/1250 [28:12<10:54,  1.84s/it, loss=0.257]
Epoch 4/4:  72%|███████▏  | 896/1250 [28:12<10:23,  1.76s/it, loss=0.257]
Epoch 4/4:  72%|███████▏  | 896/1250 [28:14<10:23,  1.76s/it, loss=0.511]
Epoch 4/4:  72%|███████▏  | 897/1250 [28:14<10:33,  1.80s/it, loss=0.511]
Epoch 4/4:  72%|███████▏  | 897/1250 [28:16<10:33,  1.80s/it, loss=0.487]
Epoch 4/4:  72%|███████▏  | 898/1250 [28:16<10:10,  1.73s/it, loss=0.487]
Epoch 4/4:  72%|███████▏  | 898/1250 [28:18<10:10,  1.73s/it, loss=0.419]
Epoch 4/4:  72%|███████▏  | 899/1250 [28:18<10:41,  1.83s/it, loss=0.419]
Epoch 4/4:  72%|███████▏  | 899/1250 [28:19<10:41,  1.83s/it, loss=0.502]
Epoch 4/4:  72%|███████▏  | 900/1250 [28:19<10:08,  1.74s/it, loss=0.502]
Epoch 4/4:  72%|███████▏  | 900/1250 [28:21<10:08,  1.74s/it, loss=0.489]
Epoch 4/4:  72%|███████▏  | 901/1250 [28:21<10:04,  1.73s/it, loss=0.489]
Epoch 4/4:  72%|███████▏  | 901/1250 [28:23<10:04,  1.73s/it, loss=0.413]
Epoch 4/4:  72%|███████▏  | 902/1250 [28:23<10:25,  1.80s/it, loss=0.413]
Epoch 4/4:  72%|███████▏  | 902/1250 [28:24<10:25,  1.80s/it, loss=0.661]
Epoch 4/4:  72%|███████▏  | 903/1250 [28:24<09:59,  1.73s/it, loss=0.661]
Epoch 4/4:  72%|███████▏  | 903/1250 [28:26<09:59,  1.73s/it, loss=0.438]
Epoch 4/4:  72%|███████▏  | 904/1250 [28:26<10:17,  1.79s/it, loss=0.438]
Epoch 4/4:  72%|███████▏  | 904/1250 [28:28<10:17,  1.79s/it, loss=0.492]
Epoch 4/4:  72%|███████▏  | 905/1250 [28:28<09:58,  1.74s/it, loss=0.492]
Epoch 4/4:  72%|███████▏  | 905/1250 [28:29<09:58,  1.74s/it, loss=0.502]
Epoch 4/4:  72%|███████▏  | 906/1250 [28:29<09:39,  1.68s/it, loss=0.502]
Epoch 4/4:  72%|███████▏  | 906/1250 [28:31<09:39,  1.68s/it, loss=0.524]
Epoch 4/4:  73%|███████▎  | 907/1250 [28:31<09:28,  1.66s/it, loss=0.524]
Epoch 4/4:  73%|███████▎  | 907/1250 [28:33<09:28,  1.66s/it, loss=0.657]
Epoch 4/4:  73%|███████▎  | 908/1250 [28:33<09:54,  1.74s/it, loss=0.657]
Epoch 4/4:  73%|███████▎  | 908/1250 [28:35<09:54,  1.74s/it, loss=0.33] 
Epoch 4/4:  73%|███████▎  | 909/1250 [28:35<10:44,  1.89s/it, loss=0.33]
Epoch 4/4:  73%|███████▎  | 909/1250 [28:37<10:44,  1.89s/it, loss=0.456]
Epoch 4/4:  73%|███████▎  | 910/1250 [28:37<10:42,  1.89s/it, loss=0.456]
Epoch 4/4:  73%|███████▎  | 910/1250 [28:39<10:42,  1.89s/it, loss=0.63] 
Epoch 4/4:  73%|███████▎  | 911/1250 [28:39<10:23,  1.84s/it, loss=0.63]
Epoch 4/4:  73%|███████▎  | 911/1250 [28:40<10:23,  1.84s/it, loss=0.64]
Epoch 4/4:  73%|███████▎  | 912/1250 [28:40<10:01,  1.78s/it, loss=0.64]
Epoch 4/4:  73%|███████▎  | 912/1250 [28:42<10:01,  1.78s/it, loss=0.545]
Epoch 4/4:  73%|███████▎  | 913/1250 [28:42<10:18,  1.83s/it, loss=0.545]
Epoch 4/4:  73%|███████▎  | 913/1250 [28:44<10:18,  1.83s/it, loss=0.417]
Epoch 4/4:  73%|███████▎  | 914/1250 [28:44<10:18,  1.84s/it, loss=0.417]
Epoch 4/4:  73%|███████▎  | 914/1250 [28:46<10:18,  1.84s/it, loss=0.4]  
Epoch 4/4:  73%|███████▎  | 915/1250 [28:46<10:35,  1.90s/it, loss=0.4]
Epoch 4/4:  73%|███████▎  | 915/1250 [28:49<10:35,  1.90s/it, loss=0.518]
Epoch 4/4:  73%|███████▎  | 916/1250 [28:49<11:17,  2.03s/it, loss=0.518]
Epoch 4/4:  73%|███████▎  | 916/1250 [28:50<11:17,  2.03s/it, loss=0.445]
Epoch 4/4:  73%|███████▎  | 917/1250 [28:50<10:53,  1.96s/it, loss=0.445]
Epoch 4/4:  73%|███████▎  | 917/1250 [28:53<10:53,  1.96s/it, loss=0.368]
Epoch 4/4:  73%|███████▎  | 918/1250 [28:53<11:06,  2.01s/it, loss=0.368]
Epoch 4/4:  73%|███████▎  | 918/1250 [28:55<11:06,  2.01s/it, loss=0.514]
Epoch 4/4:  74%|███████▎  | 919/1250 [28:55<11:15,  2.04s/it, loss=0.514]
Epoch 4/4:  74%|███████▎  | 919/1250 [28:57<11:15,  2.04s/it, loss=0.41] 
Epoch 4/4:  74%|███████▎  | 920/1250 [28:57<11:42,  2.13s/it, loss=0.41]
Epoch 4/4:  74%|███████▎  | 920/1250 [28:59<11:42,  2.13s/it, loss=0.477]
Epoch 4/4:  74%|███████▎  | 921/1250 [28:59<11:17,  2.06s/it, loss=0.477]
Epoch 4/4:  74%|███████▎  | 921/1250 [29:01<11:17,  2.06s/it, loss=0.399]
Epoch 4/4:  74%|███████▍  | 922/1250 [29:01<10:39,  1.95s/it, loss=0.399]
Epoch 4/4:  74%|███████▍  | 922/1250 [29:03<10:39,  1.95s/it, loss=0.469]
Epoch 4/4:  74%|███████▍  | 923/1250 [29:03<10:44,  1.97s/it, loss=0.469]
Epoch 4/4:  74%|███████▍  | 923/1250 [29:05<10:44,  1.97s/it, loss=0.512]
Epoch 4/4:  74%|███████▍  | 924/1250 [29:05<10:41,  1.97s/it, loss=0.512]
Epoch 4/4:  74%|███████▍  | 924/1250 [29:06<10:41,  1.97s/it, loss=0.544]
Epoch 4/4:  74%|███████▍  | 925/1250 [29:06<10:02,  1.85s/it, loss=0.544]
Epoch 4/4:  74%|███████▍  | 925/1250 [29:08<10:02,  1.85s/it, loss=0.494]
Epoch 4/4:  74%|███████▍  | 926/1250 [29:08<09:41,  1.80s/it, loss=0.494]
Epoch 4/4:  74%|███████▍  | 926/1250 [29:10<09:41,  1.80s/it, loss=0.585]
Epoch 4/4:  74%|███████▍  | 927/1250 [29:10<09:42,  1.80s/it, loss=0.585]
Epoch 4/4:  74%|███████▍  | 927/1250 [29:12<09:42,  1.80s/it, loss=0.468]
Epoch 4/4:  74%|███████▍  | 928/1250 [29:12<10:09,  1.89s/it, loss=0.468]
Epoch 4/4:  74%|███████▍  | 928/1250 [29:14<10:09,  1.89s/it, loss=0.613]
Epoch 4/4:  74%|███████▍  | 929/1250 [29:14<10:07,  1.89s/it, loss=0.613]
Epoch 4/4:  74%|███████▍  | 929/1250 [29:16<10:07,  1.89s/it, loss=0.531]
Epoch 4/4:  74%|███████▍  | 930/1250 [29:16<10:06,  1.89s/it, loss=0.531]
Epoch 4/4:  74%|███████▍  | 930/1250 [29:17<10:06,  1.89s/it, loss=0.459]
Epoch 4/4:  74%|███████▍  | 931/1250 [29:17<09:47,  1.84s/it, loss=0.459]
Epoch 4/4:  74%|███████▍  | 931/1250 [29:19<09:47,  1.84s/it, loss=0.554]
Epoch 4/4:  75%|███████▍  | 932/1250 [29:19<09:47,  1.85s/it, loss=0.554]
Epoch 4/4:  75%|███████▍  | 932/1250 [29:21<09:47,  1.85s/it, loss=0.524]
Epoch 4/4:  75%|███████▍  | 933/1250 [29:21<09:28,  1.79s/it, loss=0.524]
Epoch 4/4:  75%|███████▍  | 933/1250 [29:23<09:28,  1.79s/it, loss=0.488]
Epoch 4/4:  75%|███████▍  | 934/1250 [29:23<09:40,  1.84s/it, loss=0.488]
Epoch 4/4:  75%|███████▍  | 934/1250 [29:25<09:40,  1.84s/it, loss=0.4]  
Epoch 4/4:  75%|███████▍  | 935/1250 [29:25<10:12,  1.94s/it, loss=0.4]
Epoch 4/4:  75%|███████▍  | 935/1250 [29:27<10:12,  1.94s/it, loss=0.574]
Epoch 4/4:  75%|███████▍  | 936/1250 [29:27<09:37,  1.84s/it, loss=0.574]
Epoch 4/4:  75%|███████▍  | 936/1250 [29:28<09:37,  1.84s/it, loss=0.484]
Epoch 4/4:  75%|███████▍  | 937/1250 [29:28<09:31,  1.83s/it, loss=0.484]
Epoch 4/4:  75%|███████▍  | 937/1250 [29:30<09:31,  1.83s/it, loss=0.472]
Epoch 4/4:  75%|███████▌  | 938/1250 [29:30<09:35,  1.85s/it, loss=0.472]
Epoch 4/4:  75%|███████▌  | 938/1250 [29:32<09:35,  1.85s/it, loss=0.348]
Epoch 4/4:  75%|███████▌  | 939/1250 [29:32<09:43,  1.88s/it, loss=0.348]
Epoch 4/4:  75%|███████▌  | 939/1250 [29:34<09:43,  1.88s/it, loss=0.553]
Epoch 4/4:  75%|███████▌  | 940/1250 [29:34<09:46,  1.89s/it, loss=0.553]
Epoch 4/4:  75%|███████▌  | 940/1250 [29:36<09:46,  1.89s/it, loss=0.309]
Epoch 4/4:  75%|███████▌  | 941/1250 [29:36<09:46,  1.90s/it, loss=0.309]
Epoch 4/4:  75%|███████▌  | 941/1250 [29:38<09:46,  1.90s/it, loss=0.424]
Epoch 4/4:  75%|███████▌  | 942/1250 [29:38<10:14,  2.00s/it, loss=0.424]
Epoch 4/4:  75%|███████▌  | 942/1250 [29:41<10:14,  2.00s/it, loss=0.574]
Epoch 4/4:  75%|███████▌  | 943/1250 [29:41<10:59,  2.15s/it, loss=0.574]
Epoch 4/4:  75%|███████▌  | 943/1250 [29:43<10:59,  2.15s/it, loss=0.465]
Epoch 4/4:  76%|███████▌  | 944/1250 [29:43<11:01,  2.16s/it, loss=0.465]
Epoch 4/4:  76%|███████▌  | 944/1250 [29:45<11:01,  2.16s/it, loss=0.603]
Epoch 4/4:  76%|███████▌  | 945/1250 [29:45<10:14,  2.01s/it, loss=0.603]
Epoch 4/4:  76%|███████▌  | 945/1250 [29:46<10:14,  2.01s/it, loss=0.416]
Epoch 4/4:  76%|███████▌  | 946/1250 [29:46<09:27,  1.87s/it, loss=0.416]
Epoch 4/4:  76%|███████▌  | 946/1250 [29:48<09:27,  1.87s/it, loss=0.544]
Epoch 4/4:  76%|███████▌  | 947/1250 [29:48<09:01,  1.79s/it, loss=0.544]
Epoch 4/4:  76%|███████▌  | 947/1250 [29:50<09:01,  1.79s/it, loss=0.542]
Epoch 4/4:  76%|███████▌  | 948/1250 [29:50<09:05,  1.80s/it, loss=0.542]
Epoch 4/4:  76%|███████▌  | 948/1250 [29:51<09:05,  1.80s/it, loss=0.674]
Epoch 4/4:  76%|███████▌  | 949/1250 [29:51<08:33,  1.71s/it, loss=0.674]
Epoch 4/4:  76%|███████▌  | 949/1250 [29:53<08:33,  1.71s/it, loss=0.612]
Epoch 4/4:  76%|███████▌  | 950/1250 [29:53<08:13,  1.65s/it, loss=0.612]
Epoch 4/4:  76%|███████▌  | 950/1250 [29:54<08:13,  1.65s/it, loss=0.683]
Epoch 4/4:  76%|███████▌  | 951/1250 [29:54<08:18,  1.67s/it, loss=0.683]
Epoch 4/4:  76%|███████▌  | 951/1250 [29:56<08:18,  1.67s/it, loss=0.424]
Epoch 4/4:  76%|███████▌  | 952/1250 [29:56<08:00,  1.61s/it, loss=0.424]
Epoch 4/4:  76%|███████▌  | 952/1250 [29:57<08:00,  1.61s/it, loss=0.447]
Epoch 4/4:  76%|███████▌  | 953/1250 [29:57<07:59,  1.62s/it, loss=0.447]
Epoch 4/4:  76%|███████▌  | 953/1250 [29:59<07:59,  1.62s/it, loss=0.546]
Epoch 4/4:  76%|███████▋  | 954/1250 [29:59<07:48,  1.58s/it, loss=0.546]
Epoch 4/4:  76%|███████▋  | 954/1250 [30:00<07:48,  1.58s/it, loss=0.636]
Epoch 4/4:  76%|███████▋  | 955/1250 [30:00<07:49,  1.59s/it, loss=0.636]
Epoch 4/4:  76%|███████▋  | 955/1250 [30:02<07:49,  1.59s/it, loss=0.44] 
Epoch 4/4:  76%|███████▋  | 956/1250 [30:02<07:52,  1.61s/it, loss=0.44]
Epoch 4/4:  76%|███████▋  | 956/1250 [30:04<07:52,  1.61s/it, loss=0.683]
Epoch 4/4:  77%|███████▋  | 957/1250 [30:04<07:39,  1.57s/it, loss=0.683]
Epoch 4/4:  77%|███████▋  | 957/1250 [30:05<07:39,  1.57s/it, loss=0.426]
Epoch 4/4:  77%|███████▋  | 958/1250 [30:05<07:50,  1.61s/it, loss=0.426]
Epoch 4/4:  77%|███████▋  | 958/1250 [30:07<07:50,  1.61s/it, loss=0.424]
Epoch 4/4:  77%|███████▋  | 959/1250 [30:07<07:43,  1.59s/it, loss=0.424]
Epoch 4/4:  77%|███████▋  | 959/1250 [30:08<07:43,  1.59s/it, loss=0.455]
Epoch 4/4:  77%|███████▋  | 960/1250 [30:08<07:41,  1.59s/it, loss=0.455]
Epoch 4/4:  77%|███████▋  | 960/1250 [30:10<07:41,  1.59s/it, loss=0.472]
Epoch 4/4:  77%|███████▋  | 961/1250 [30:10<07:42,  1.60s/it, loss=0.472]
Epoch 4/4:  77%|███████▋  | 961/1250 [30:12<07:42,  1.60s/it, loss=0.396]
Epoch 4/4:  77%|███████▋  | 962/1250 [30:12<07:55,  1.65s/it, loss=0.396]
Epoch 4/4:  77%|███████▋  | 962/1250 [30:14<07:55,  1.65s/it, loss=0.448]
Epoch 4/4:  77%|███████▋  | 963/1250 [30:14<08:20,  1.74s/it, loss=0.448]
Epoch 4/4:  77%|███████▋  | 963/1250 [30:16<08:20,  1.74s/it, loss=0.47] 
Epoch 4/4:  77%|███████▋  | 964/1250 [30:16<09:08,  1.92s/it, loss=0.47]
Epoch 4/4:  77%|███████▋  | 964/1250 [30:20<09:08,  1.92s/it, loss=0.509]
Epoch 4/4:  77%|███████▋  | 965/1250 [30:20<12:26,  2.62s/it, loss=0.509]
Epoch 4/4:  77%|███████▋  | 965/1250 [30:26<12:26,  2.62s/it, loss=0.427]
Epoch 4/4:  77%|███████▋  | 966/1250 [30:26<17:04,  3.61s/it, loss=0.427]
Epoch 4/4:  77%|███████▋  | 966/1250 [30:29<17:04,  3.61s/it, loss=0.461]
Epoch 4/4:  77%|███████▋  | 967/1250 [30:29<15:55,  3.38s/it, loss=0.461]
Epoch 4/4:  77%|███████▋  | 967/1250 [30:31<15:55,  3.38s/it, loss=0.442]
Epoch 4/4:  77%|███████▋  | 968/1250 [30:31<14:22,  3.06s/it, loss=0.442]
Epoch 4/4:  77%|███████▋  | 968/1250 [30:33<14:22,  3.06s/it, loss=0.845]
Epoch 4/4:  78%|███████▊  | 969/1250 [30:33<12:43,  2.72s/it, loss=0.845]
Epoch 4/4:  78%|███████▊  | 969/1250 [30:35<12:43,  2.72s/it, loss=0.468]
Epoch 4/4:  78%|███████▊  | 970/1250 [30:35<11:06,  2.38s/it, loss=0.468]
Epoch 4/4:  78%|███████▊  | 970/1250 [30:37<11:06,  2.38s/it, loss=0.481]
Epoch 4/4:  78%|███████▊  | 971/1250 [30:37<09:59,  2.15s/it, loss=0.481]
Epoch 4/4:  78%|███████▊  | 971/1250 [30:38<09:59,  2.15s/it, loss=0.457]
Epoch 4/4:  78%|███████▊  | 972/1250 [30:38<09:08,  1.97s/it, loss=0.457]
Epoch 4/4:  78%|███████▊  | 972/1250 [30:40<09:08,  1.97s/it, loss=0.642]
Epoch 4/4:  78%|███████▊  | 973/1250 [30:40<08:23,  1.82s/it, loss=0.642]
Epoch 4/4:  78%|███████▊  | 973/1250 [30:42<08:23,  1.82s/it, loss=0.522]
Epoch 4/4:  78%|███████▊  | 974/1250 [30:42<08:47,  1.91s/it, loss=0.522]
Epoch 4/4:  78%|███████▊  | 974/1250 [30:44<08:47,  1.91s/it, loss=0.392]
Epoch 4/4:  78%|███████▊  | 975/1250 [30:44<08:40,  1.89s/it, loss=0.392]
Epoch 4/4:  78%|███████▊  | 975/1250 [30:46<08:40,  1.89s/it, loss=0.262]
Epoch 4/4:  78%|███████▊  | 976/1250 [30:46<08:57,  1.96s/it, loss=0.262]
Epoch 4/4:  78%|███████▊  | 976/1250 [30:48<08:57,  1.96s/it, loss=0.506]
Epoch 4/4:  78%|███████▊  | 977/1250 [30:48<09:13,  2.03s/it, loss=0.506]
Epoch 4/4:  78%|███████▊  | 977/1250 [30:50<09:13,  2.03s/it, loss=0.475]
Epoch 4/4:  78%|███████▊  | 978/1250 [30:50<09:06,  2.01s/it, loss=0.475]
Epoch 4/4:  78%|███████▊  | 978/1250 [30:52<09:06,  2.01s/it, loss=0.549]
Epoch 4/4:  78%|███████▊  | 979/1250 [30:52<09:19,  2.06s/it, loss=0.549]
Epoch 4/4:  78%|███████▊  | 979/1250 [30:56<09:19,  2.06s/it, loss=0.322]
Epoch 4/4:  78%|███████▊  | 980/1250 [30:56<12:27,  2.77s/it, loss=0.322]
Epoch 4/4:  78%|███████▊  | 980/1250 [30:59<12:27,  2.77s/it, loss=0.558]
Epoch 4/4:  78%|███████▊  | 981/1250 [30:59<11:26,  2.55s/it, loss=0.558]
Epoch 4/4:  78%|███████▊  | 981/1250 [31:00<11:26,  2.55s/it, loss=0.584]
Epoch 4/4:  79%|███████▊  | 982/1250 [31:00<10:12,  2.28s/it, loss=0.584]
Epoch 4/4:  79%|███████▊  | 982/1250 [31:02<10:12,  2.28s/it, loss=0.47] 
Epoch 4/4:  79%|███████▊  | 983/1250 [31:02<09:09,  2.06s/it, loss=0.47]
Epoch 4/4:  79%|███████▊  | 983/1250 [31:03<09:09,  2.06s/it, loss=0.626]
Epoch 4/4:  79%|███████▊  | 984/1250 [31:03<08:37,  1.95s/it, loss=0.626]
Epoch 4/4:  79%|███████▊  | 984/1250 [31:05<08:37,  1.95s/it, loss=0.662]
Epoch 4/4:  79%|███████▉  | 985/1250 [31:05<08:05,  1.83s/it, loss=0.662]
Epoch 4/4:  79%|███████▉  | 985/1250 [31:07<08:05,  1.83s/it, loss=0.395]
Epoch 4/4:  79%|███████▉  | 986/1250 [31:07<07:47,  1.77s/it, loss=0.395]
Epoch 4/4:  79%|███████▉  | 986/1250 [31:09<07:47,  1.77s/it, loss=0.414]
Epoch 4/4:  79%|███████▉  | 987/1250 [31:09<08:00,  1.83s/it, loss=0.414]
Epoch 4/4:  79%|███████▉  | 987/1250 [31:10<08:00,  1.83s/it, loss=0.42] 
Epoch 4/4:  79%|███████▉  | 988/1250 [31:10<07:36,  1.74s/it, loss=0.42]
Epoch 4/4:  79%|███████▉  | 988/1250 [31:12<07:36,  1.74s/it, loss=0.287]
Epoch 4/4:  79%|███████▉  | 989/1250 [31:12<07:20,  1.69s/it, loss=0.287]
Epoch 4/4:  79%|███████▉  | 989/1250 [31:14<07:20,  1.69s/it, loss=0.355]
Epoch 4/4:  79%|███████▉  | 990/1250 [31:14<07:37,  1.76s/it, loss=0.355]
Epoch 4/4:  79%|███████▉  | 990/1250 [31:15<07:37,  1.76s/it, loss=0.636]
Epoch 4/4:  79%|███████▉  | 991/1250 [31:15<07:40,  1.78s/it, loss=0.636]
Epoch 4/4:  79%|███████▉  | 991/1250 [31:17<07:40,  1.78s/it, loss=0.346]
Epoch 4/4:  79%|███████▉  | 992/1250 [31:17<07:36,  1.77s/it, loss=0.346]
Epoch 4/4:  79%|███████▉  | 992/1250 [31:20<07:36,  1.77s/it, loss=0.421]
Epoch 4/4:  79%|███████▉  | 993/1250 [31:20<09:32,  2.23s/it, loss=0.421]
Epoch 4/4:  79%|███████▉  | 993/1250 [31:22<09:32,  2.23s/it, loss=0.513]
Epoch 4/4:  80%|███████▉  | 994/1250 [31:22<08:14,  1.93s/it, loss=0.513]
Epoch 4/4:  80%|███████▉  | 994/1250 [31:23<08:14,  1.93s/it, loss=0.544]
Epoch 4/4:  80%|███████▉  | 995/1250 [31:23<07:48,  1.84s/it, loss=0.544]
Epoch 4/4:  80%|███████▉  | 995/1250 [31:25<07:48,  1.84s/it, loss=0.494]
Epoch 4/4:  80%|███████▉  | 996/1250 [31:25<07:32,  1.78s/it, loss=0.494]
Epoch 4/4:  80%|███████▉  | 996/1250 [31:28<07:32,  1.78s/it, loss=0.462]
Epoch 4/4:  80%|███████▉  | 997/1250 [31:28<08:37,  2.04s/it, loss=0.462]
Epoch 4/4:  80%|███████▉  | 997/1250 [31:29<08:37,  2.04s/it, loss=0.436]
Epoch 4/4:  80%|███████▉  | 998/1250 [31:29<07:30,  1.79s/it, loss=0.436]
Epoch 4/4:  80%|███████▉  | 998/1250 [31:29<07:30,  1.79s/it, loss=0.425]
Epoch 4/4:  80%|███████▉  | 999/1250 [31:29<06:06,  1.46s/it, loss=0.425]
Epoch 4/4:  80%|███████▉  | 999/1250 [31:30<06:06,  1.46s/it, loss=0.537]
Epoch 4/4:  80%|████████  | 1000/1250 [31:30<05:14,  1.26s/it, loss=0.537]
Epoch 4/4:  80%|████████  | 1000/1250 [31:31<05:14,  1.26s/it, loss=0.515]
Epoch 4/4:  80%|████████  | 1001/1250 [31:31<04:26,  1.07s/it, loss=0.515]
Epoch 4/4:  80%|████████  | 1001/1250 [31:32<04:26,  1.07s/it, loss=0.485]
Epoch 4/4:  80%|████████  | 1002/1250 [31:32<03:53,  1.06it/s, loss=0.485]
Epoch 4/4:  80%|████████  | 1002/1250 [31:32<03:53,  1.06it/s, loss=0.418]
Epoch 4/4:  80%|████████  | 1003/1250 [31:32<03:28,  1.18it/s, loss=0.418]
Epoch 4/4:  80%|████████  | 1003/1250 [31:33<03:28,  1.18it/s, loss=0.47] 
Epoch 4/4:  80%|████████  | 1004/1250 [31:33<03:11,  1.29it/s, loss=0.47]
Epoch 4/4:  80%|████████  | 1004/1250 [31:33<03:11,  1.29it/s, loss=0.424]
Epoch 4/4:  80%|████████  | 1005/1250 [31:33<03:00,  1.36it/s, loss=0.424]
Epoch 4/4:  80%|████████  | 1005/1250 [31:34<03:00,  1.36it/s, loss=0.417]
Epoch 4/4:  80%|████████  | 1006/1250 [31:34<02:51,  1.42it/s, loss=0.417]
Epoch 4/4:  80%|████████  | 1006/1250 [31:35<02:51,  1.42it/s, loss=0.372]
Epoch 4/4:  81%|████████  | 1007/1250 [31:35<02:43,  1.48it/s, loss=0.372]
Epoch 4/4:  81%|████████  | 1007/1250 [31:35<02:43,  1.48it/s, loss=0.714]
Epoch 4/4:  81%|████████  | 1008/1250 [31:35<02:38,  1.52it/s, loss=0.714]
Epoch 4/4:  81%|████████  | 1008/1250 [31:36<02:38,  1.52it/s, loss=0.489]
Epoch 4/4:  81%|████████  | 1009/1250 [31:36<02:34,  1.56it/s, loss=0.489]
Epoch 4/4:  81%|████████  | 1009/1250 [31:37<02:34,  1.56it/s, loss=0.523]
Epoch 4/4:  81%|████████  | 1010/1250 [31:37<02:35,  1.55it/s, loss=0.523]
Epoch 4/4:  81%|████████  | 1010/1250 [31:37<02:35,  1.55it/s, loss=0.591]
Epoch 4/4:  81%|████████  | 1011/1250 [31:37<02:39,  1.50it/s, loss=0.591]
Epoch 4/4:  81%|████████  | 1011/1250 [31:38<02:39,  1.50it/s, loss=0.414]
Epoch 4/4:  81%|████████  | 1012/1250 [31:38<02:43,  1.45it/s, loss=0.414]
Epoch 4/4:  81%|████████  | 1012/1250 [31:39<02:43,  1.45it/s, loss=0.368]
Epoch 4/4:  81%|████████  | 1013/1250 [31:39<02:40,  1.48it/s, loss=0.368]
Epoch 4/4:  81%|████████  | 1013/1250 [31:39<02:40,  1.48it/s, loss=0.549]
Epoch 4/4:  81%|████████  | 1014/1250 [31:39<02:37,  1.50it/s, loss=0.549]
Epoch 4/4:  81%|████████  | 1014/1250 [31:40<02:37,  1.50it/s, loss=0.54] 
Epoch 4/4:  81%|████████  | 1015/1250 [31:40<02:32,  1.54it/s, loss=0.54]
Epoch 4/4:  81%|████████  | 1015/1250 [31:41<02:32,  1.54it/s, loss=0.563]
Epoch 4/4:  81%|████████▏ | 1016/1250 [31:41<02:29,  1.56it/s, loss=0.563]
Epoch 4/4:  81%|████████▏ | 1016/1250 [31:41<02:29,  1.56it/s, loss=0.662]
Epoch 4/4:  81%|████████▏ | 1017/1250 [31:41<02:30,  1.54it/s, loss=0.662]
Epoch 4/4:  81%|████████▏ | 1017/1250 [31:42<02:30,  1.54it/s, loss=0.425]
Epoch 4/4:  81%|████████▏ | 1018/1250 [31:42<02:32,  1.52it/s, loss=0.425]
Epoch 4/4:  81%|████████▏ | 1018/1250 [31:43<02:32,  1.52it/s, loss=0.5]  
Epoch 4/4:  82%|████████▏ | 1019/1250 [31:43<02:37,  1.46it/s, loss=0.5]
Epoch 4/4:  82%|████████▏ | 1019/1250 [31:43<02:37,  1.46it/s, loss=0.486]
Epoch 4/4:  82%|████████▏ | 1020/1250 [31:43<02:43,  1.41it/s, loss=0.486]
Epoch 4/4:  82%|████████▏ | 1020/1250 [31:44<02:43,  1.41it/s, loss=0.279]
Epoch 4/4:  82%|████████▏ | 1021/1250 [31:44<02:45,  1.39it/s, loss=0.279]
Epoch 4/4:  82%|████████▏ | 1021/1250 [31:45<02:45,  1.39it/s, loss=0.621]
Epoch 4/4:  82%|████████▏ | 1022/1250 [31:45<02:39,  1.43it/s, loss=0.621]
Epoch 4/4:  82%|████████▏ | 1022/1250 [31:45<02:39,  1.43it/s, loss=0.507]
Epoch 4/4:  82%|████████▏ | 1023/1250 [31:45<02:33,  1.48it/s, loss=0.507]
Epoch 4/4:  82%|████████▏ | 1023/1250 [31:46<02:33,  1.48it/s, loss=0.505]
Epoch 4/4:  82%|████████▏ | 1024/1250 [31:46<02:29,  1.51it/s, loss=0.505]
Epoch 4/4:  82%|████████▏ | 1024/1250 [31:47<02:29,  1.51it/s, loss=0.65] 
Epoch 4/4:  82%|████████▏ | 1025/1250 [31:47<02:26,  1.53it/s, loss=0.65]
Epoch 4/4:  82%|████████▏ | 1025/1250 [31:47<02:26,  1.53it/s, loss=0.466]
Epoch 4/4:  82%|████████▏ | 1026/1250 [31:47<02:25,  1.54it/s, loss=0.466]
Epoch 4/4:  82%|████████▏ | 1026/1250 [31:48<02:25,  1.54it/s, loss=0.38] 
Epoch 4/4:  82%|████████▏ | 1027/1250 [31:48<02:26,  1.52it/s, loss=0.38]
Epoch 4/4:  82%|████████▏ | 1027/1250 [31:49<02:26,  1.52it/s, loss=0.368]
Epoch 4/4:  82%|████████▏ | 1028/1250 [31:49<02:26,  1.51it/s, loss=0.368]
Epoch 4/4:  82%|████████▏ | 1028/1250 [31:49<02:26,  1.51it/s, loss=0.374]
Epoch 4/4:  82%|████████▏ | 1029/1250 [31:49<02:24,  1.53it/s, loss=0.374]
Epoch 4/4:  82%|████████▏ | 1029/1250 [31:50<02:24,  1.53it/s, loss=0.687]
Epoch 4/4:  82%|████████▏ | 1030/1250 [31:50<02:21,  1.55it/s, loss=0.687]
Epoch 4/4:  82%|████████▏ | 1030/1250 [31:51<02:21,  1.55it/s, loss=0.462]
Epoch 4/4:  82%|████████▏ | 1031/1250 [31:51<02:19,  1.57it/s, loss=0.462]
Epoch 4/4:  82%|████████▏ | 1031/1250 [31:51<02:19,  1.57it/s, loss=0.448]
Epoch 4/4:  83%|████████▎ | 1032/1250 [31:51<02:19,  1.56it/s, loss=0.448]
Epoch 4/4:  83%|████████▎ | 1032/1250 [31:52<02:19,  1.56it/s, loss=0.297]
Epoch 4/4:  83%|████████▎ | 1033/1250 [31:52<02:19,  1.55it/s, loss=0.297]
Epoch 4/4:  83%|████████▎ | 1033/1250 [31:52<02:19,  1.55it/s, loss=0.339]
Epoch 4/4:  83%|████████▎ | 1034/1250 [31:52<02:21,  1.53it/s, loss=0.339]
Epoch 4/4:  83%|████████▎ | 1034/1250 [31:53<02:21,  1.53it/s, loss=0.424]
Epoch 4/4:  83%|████████▎ | 1035/1250 [31:53<02:20,  1.53it/s, loss=0.424]
Epoch 4/4:  83%|████████▎ | 1035/1250 [31:54<02:20,  1.53it/s, loss=0.468]
Epoch 4/4:  83%|████████▎ | 1036/1250 [31:54<02:19,  1.53it/s, loss=0.468]
Epoch 4/4:  83%|████████▎ | 1036/1250 [31:54<02:19,  1.53it/s, loss=0.478]
Epoch 4/4:  83%|████████▎ | 1037/1250 [31:54<02:18,  1.54it/s, loss=0.478]
Epoch 4/4:  83%|████████▎ | 1037/1250 [31:55<02:18,  1.54it/s, loss=0.496]
Epoch 4/4:  83%|████████▎ | 1038/1250 [31:55<02:17,  1.54it/s, loss=0.496]
Epoch 4/4:  83%|████████▎ | 1038/1250 [31:56<02:17,  1.54it/s, loss=0.387]
Epoch 4/4:  83%|████████▎ | 1039/1250 [31:56<02:16,  1.55it/s, loss=0.387]
Epoch 4/4:  83%|████████▎ | 1039/1250 [31:56<02:16,  1.55it/s, loss=0.458]
Epoch 4/4:  83%|████████▎ | 1040/1250 [31:56<02:15,  1.55it/s, loss=0.458]
Epoch 4/4:  83%|████████▎ | 1040/1250 [31:57<02:15,  1.55it/s, loss=0.442]
Epoch 4/4:  83%|████████▎ | 1041/1250 [31:57<02:15,  1.54it/s, loss=0.442]
Epoch 4/4:  83%|████████▎ | 1041/1250 [31:58<02:15,  1.54it/s, loss=0.502]
Epoch 4/4:  83%|████████▎ | 1042/1250 [31:58<02:17,  1.51it/s, loss=0.502]
Epoch 4/4:  83%|████████▎ | 1042/1250 [31:58<02:17,  1.51it/s, loss=0.334]
Epoch 4/4:  83%|████████▎ | 1043/1250 [31:58<02:18,  1.49it/s, loss=0.334]
Epoch 4/4:  83%|████████▎ | 1043/1250 [31:59<02:18,  1.49it/s, loss=0.354]
Epoch 4/4:  84%|████████▎ | 1044/1250 [31:59<02:20,  1.47it/s, loss=0.354]
Epoch 4/4:  84%|████████▎ | 1044/1250 [32:00<02:20,  1.47it/s, loss=0.561]
Epoch 4/4:  84%|████████▎ | 1045/1250 [32:00<02:18,  1.48it/s, loss=0.561]
Epoch 4/4:  84%|████████▎ | 1045/1250 [32:00<02:18,  1.48it/s, loss=0.339]
Epoch 4/4:  84%|████████▎ | 1046/1250 [32:00<02:18,  1.47it/s, loss=0.339]
Epoch 4/4:  84%|████████▎ | 1046/1250 [32:01<02:18,  1.47it/s, loss=0.575]
Epoch 4/4:  84%|████████▍ | 1047/1250 [32:01<02:17,  1.48it/s, loss=0.575]
Epoch 4/4:  84%|████████▍ | 1047/1250 [32:02<02:17,  1.48it/s, loss=0.441]
Epoch 4/4:  84%|████████▍ | 1048/1250 [32:02<02:16,  1.48it/s, loss=0.441]
Epoch 4/4:  84%|████████▍ | 1048/1250 [32:03<02:16,  1.48it/s, loss=0.755]
Epoch 4/4:  84%|████████▍ | 1049/1250 [32:03<02:17,  1.46it/s, loss=0.755]
Epoch 4/4:  84%|████████▍ | 1049/1250 [32:03<02:17,  1.46it/s, loss=0.462]
Epoch 4/4:  84%|████████▍ | 1050/1250 [32:03<02:19,  1.44it/s, loss=0.462]
Epoch 4/4:  84%|████████▍ | 1050/1250 [32:04<02:19,  1.44it/s, loss=0.376]
Epoch 4/4:  84%|████████▍ | 1051/1250 [32:04<02:19,  1.43it/s, loss=0.376]
Epoch 4/4:  84%|████████▍ | 1051/1250 [32:05<02:19,  1.43it/s, loss=0.541]
Epoch 4/4:  84%|████████▍ | 1052/1250 [32:05<02:21,  1.40it/s, loss=0.541]
Epoch 4/4:  84%|████████▍ | 1052/1250 [32:05<02:21,  1.40it/s, loss=0.467]
Epoch 4/4:  84%|████████▍ | 1053/1250 [32:05<02:18,  1.42it/s, loss=0.467]
Epoch 4/4:  84%|████████▍ | 1053/1250 [32:06<02:18,  1.42it/s, loss=0.493]
Epoch 4/4:  84%|████████▍ | 1054/1250 [32:06<02:17,  1.43it/s, loss=0.493]
Epoch 4/4:  84%|████████▍ | 1054/1250 [32:07<02:17,  1.43it/s, loss=0.806]
Epoch 4/4:  84%|████████▍ | 1055/1250 [32:07<02:17,  1.42it/s, loss=0.806]
Epoch 4/4:  84%|████████▍ | 1055/1250 [32:08<02:17,  1.42it/s, loss=0.386]
Epoch 4/4:  84%|████████▍ | 1056/1250 [32:08<02:26,  1.33it/s, loss=0.386]
Epoch 4/4:  84%|████████▍ | 1056/1250 [32:08<02:26,  1.33it/s, loss=0.6]  
Epoch 4/4:  85%|████████▍ | 1057/1250 [32:08<02:25,  1.33it/s, loss=0.6]
Epoch 4/4:  85%|████████▍ | 1057/1250 [32:09<02:25,  1.33it/s, loss=0.387]
Epoch 4/4:  85%|████████▍ | 1058/1250 [32:09<02:25,  1.32it/s, loss=0.387]
Epoch 4/4:  85%|████████▍ | 1058/1250 [32:10<02:25,  1.32it/s, loss=0.442]
Epoch 4/4:  85%|████████▍ | 1059/1250 [32:10<02:20,  1.36it/s, loss=0.442]
Epoch 4/4:  85%|████████▍ | 1059/1250 [32:11<02:20,  1.36it/s, loss=0.355]
Epoch 4/4:  85%|████████▍ | 1060/1250 [32:11<02:16,  1.39it/s, loss=0.355]
Epoch 4/4:  85%|████████▍ | 1060/1250 [32:11<02:16,  1.39it/s, loss=0.347]
Epoch 4/4:  85%|████████▍ | 1061/1250 [32:11<02:13,  1.42it/s, loss=0.347]
Epoch 4/4:  85%|████████▍ | 1061/1250 [32:12<02:13,  1.42it/s, loss=0.485]
Epoch 4/4:  85%|████████▍ | 1062/1250 [32:12<02:10,  1.44it/s, loss=0.485]
Epoch 4/4:  85%|████████▍ | 1062/1250 [32:13<02:10,  1.44it/s, loss=0.365]
Epoch 4/4:  85%|████████▌ | 1063/1250 [32:13<02:10,  1.44it/s, loss=0.365]
Epoch 4/4:  85%|████████▌ | 1063/1250 [32:13<02:10,  1.44it/s, loss=0.498]
Epoch 4/4:  85%|████████▌ | 1064/1250 [32:13<02:10,  1.43it/s, loss=0.498]
Epoch 4/4:  85%|████████▌ | 1064/1250 [32:14<02:10,  1.43it/s, loss=0.52] 
Epoch 4/4:  85%|████████▌ | 1065/1250 [32:14<02:09,  1.43it/s, loss=0.52]
Epoch 4/4:  85%|████████▌ | 1065/1250 [32:15<02:09,  1.43it/s, loss=0.388]
Epoch 4/4:  85%|████████▌ | 1066/1250 [32:15<02:09,  1.42it/s, loss=0.388]
Epoch 4/4:  85%|████████▌ | 1066/1250 [32:15<02:09,  1.42it/s, loss=0.447]
Epoch 4/4:  85%|████████▌ | 1067/1250 [32:15<02:08,  1.43it/s, loss=0.447]
Epoch 4/4:  85%|████████▌ | 1067/1250 [32:16<02:08,  1.43it/s, loss=0.356]
Epoch 4/4:  85%|████████▌ | 1068/1250 [32:16<02:06,  1.44it/s, loss=0.356]
Epoch 4/4:  85%|████████▌ | 1068/1250 [32:17<02:06,  1.44it/s, loss=0.477]
Epoch 4/4:  86%|████████▌ | 1069/1250 [32:17<02:04,  1.46it/s, loss=0.477]
Epoch 4/4:  86%|████████▌ | 1069/1250 [32:17<02:04,  1.46it/s, loss=0.556]
Epoch 4/4:  86%|████████▌ | 1070/1250 [32:17<02:05,  1.44it/s, loss=0.556]
Epoch 4/4:  86%|████████▌ | 1070/1250 [32:18<02:05,  1.44it/s, loss=0.488]
Epoch 4/4:  86%|████████▌ | 1071/1250 [32:18<02:04,  1.44it/s, loss=0.488]
Epoch 4/4:  86%|████████▌ | 1071/1250 [32:19<02:04,  1.44it/s, loss=0.348]
Epoch 4/4:  86%|████████▌ | 1072/1250 [32:19<02:04,  1.43it/s, loss=0.348]
Epoch 4/4:  86%|████████▌ | 1072/1250 [32:20<02:04,  1.43it/s, loss=0.416]
Epoch 4/4:  86%|████████▌ | 1073/1250 [32:20<02:07,  1.39it/s, loss=0.416]
Epoch 4/4:  86%|████████▌ | 1073/1250 [32:20<02:07,  1.39it/s, loss=0.386]
Epoch 4/4:  86%|████████▌ | 1074/1250 [32:20<02:04,  1.42it/s, loss=0.386]
Epoch 4/4:  86%|████████▌ | 1074/1250 [32:21<02:04,  1.42it/s, loss=0.504]
Epoch 4/4:  86%|████████▌ | 1075/1250 [32:21<02:01,  1.44it/s, loss=0.504]
Epoch 4/4:  86%|████████▌ | 1075/1250 [32:22<02:01,  1.44it/s, loss=0.519]
Epoch 4/4:  86%|████████▌ | 1076/1250 [32:22<02:00,  1.45it/s, loss=0.519]
Epoch 4/4:  86%|████████▌ | 1076/1250 [32:22<02:00,  1.45it/s, loss=0.473]
Epoch 4/4:  86%|████████▌ | 1077/1250 [32:22<02:05,  1.37it/s, loss=0.473]
Epoch 4/4:  86%|████████▌ | 1077/1250 [32:23<02:05,  1.37it/s, loss=0.445]
Epoch 4/4:  86%|████████▌ | 1078/1250 [32:23<02:03,  1.39it/s, loss=0.445]
Epoch 4/4:  86%|████████▌ | 1078/1250 [32:24<02:03,  1.39it/s, loss=0.532]
Epoch 4/4:  86%|████████▋ | 1079/1250 [32:24<02:01,  1.40it/s, loss=0.532]
Epoch 4/4:  86%|████████▋ | 1079/1250 [32:25<02:01,  1.40it/s, loss=0.439]
Epoch 4/4:  86%|████████▋ | 1080/1250 [32:25<02:00,  1.41it/s, loss=0.439]
Epoch 4/4:  86%|████████▋ | 1080/1250 [32:25<02:00,  1.41it/s, loss=0.425]
Epoch 4/4:  86%|████████▋ | 1081/1250 [32:25<01:57,  1.43it/s, loss=0.425]
Epoch 4/4:  86%|████████▋ | 1081/1250 [32:26<01:57,  1.43it/s, loss=0.425]
Epoch 4/4:  87%|████████▋ | 1082/1250 [32:26<01:56,  1.44it/s, loss=0.425]
Epoch 4/4:  87%|████████▋ | 1082/1250 [32:27<01:56,  1.44it/s, loss=0.451]
Epoch 4/4:  87%|████████▋ | 1083/1250 [32:27<01:56,  1.43it/s, loss=0.451]
Epoch 4/4:  87%|████████▋ | 1083/1250 [32:27<01:56,  1.43it/s, loss=0.45] 
Epoch 4/4:  87%|████████▋ | 1084/1250 [32:27<01:56,  1.42it/s, loss=0.45]
Epoch 4/4:  87%|████████▋ | 1084/1250 [32:28<01:56,  1.42it/s, loss=0.274]
Epoch 4/4:  87%|████████▋ | 1085/1250 [32:28<01:58,  1.39it/s, loss=0.274]
Epoch 4/4:  87%|████████▋ | 1085/1250 [32:29<01:58,  1.39it/s, loss=0.57] 
Epoch 4/4:  87%|████████▋ | 1086/1250 [32:29<01:58,  1.38it/s, loss=0.57]
Epoch 4/4:  87%|████████▋ | 1086/1250 [32:30<01:58,  1.38it/s, loss=0.416]
Epoch 4/4:  87%|████████▋ | 1087/1250 [32:30<01:59,  1.37it/s, loss=0.416]
Epoch 4/4:  87%|████████▋ | 1087/1250 [32:30<01:59,  1.37it/s, loss=0.374]
Epoch 4/4:  87%|████████▋ | 1088/1250 [32:30<01:57,  1.37it/s, loss=0.374]
Epoch 4/4:  87%|████████▋ | 1088/1250 [32:31<01:57,  1.37it/s, loss=0.489]
Epoch 4/4:  87%|████████▋ | 1089/1250 [32:31<01:55,  1.39it/s, loss=0.489]
Epoch 4/4:  87%|████████▋ | 1089/1250 [32:32<01:55,  1.39it/s, loss=0.459]
Epoch 4/4:  87%|████████▋ | 1090/1250 [32:32<01:54,  1.40it/s, loss=0.459]
Epoch 4/4:  87%|████████▋ | 1090/1250 [32:32<01:54,  1.40it/s, loss=0.511]
Epoch 4/4:  87%|████████▋ | 1091/1250 [32:32<01:54,  1.39it/s, loss=0.511]
Epoch 4/4:  87%|████████▋ | 1091/1250 [32:33<01:54,  1.39it/s, loss=0.449]
Epoch 4/4:  87%|████████▋ | 1092/1250 [32:33<01:58,  1.34it/s, loss=0.449]
Epoch 4/4:  87%|████████▋ | 1092/1250 [32:34<01:58,  1.34it/s, loss=0.494]
Epoch 4/4:  87%|████████▋ | 1093/1250 [32:34<01:59,  1.32it/s, loss=0.494]
Epoch 4/4:  87%|████████▋ | 1093/1250 [32:35<01:59,  1.32it/s, loss=0.317]
Epoch 4/4:  88%|████████▊ | 1094/1250 [32:35<01:56,  1.33it/s, loss=0.317]
Epoch 4/4:  88%|████████▊ | 1094/1250 [32:35<01:56,  1.33it/s, loss=0.479]
Epoch 4/4:  88%|████████▊ | 1095/1250 [32:35<01:55,  1.34it/s, loss=0.479]
Epoch 4/4:  88%|████████▊ | 1095/1250 [32:36<01:55,  1.34it/s, loss=0.372]
Epoch 4/4:  88%|████████▊ | 1096/1250 [32:36<01:52,  1.37it/s, loss=0.372]
Epoch 4/4:  88%|████████▊ | 1096/1250 [32:37<01:52,  1.37it/s, loss=0.53] 
Epoch 4/4:  88%|████████▊ | 1097/1250 [32:37<01:49,  1.39it/s, loss=0.53]
Epoch 4/4:  88%|████████▊ | 1097/1250 [32:38<01:49,  1.39it/s, loss=0.662]
Epoch 4/4:  88%|████████▊ | 1098/1250 [32:38<01:48,  1.40it/s, loss=0.662]
Epoch 4/4:  88%|████████▊ | 1098/1250 [32:38<01:48,  1.40it/s, loss=0.467]
Epoch 4/4:  88%|████████▊ | 1099/1250 [32:38<01:48,  1.39it/s, loss=0.467]
Epoch 4/4:  88%|████████▊ | 1099/1250 [32:39<01:48,  1.39it/s, loss=0.471]
Epoch 4/4:  88%|████████▊ | 1100/1250 [32:39<01:48,  1.39it/s, loss=0.471]
Epoch 4/4:  88%|████████▊ | 1100/1250 [32:40<01:48,  1.39it/s, loss=0.539]
Epoch 4/4:  88%|████████▊ | 1101/1250 [32:40<01:48,  1.38it/s, loss=0.539]
Epoch 4/4:  88%|████████▊ | 1101/1250 [32:40<01:48,  1.38it/s, loss=0.389]
Epoch 4/4:  88%|████████▊ | 1102/1250 [32:40<01:47,  1.38it/s, loss=0.389]
Epoch 4/4:  88%|████████▊ | 1102/1250 [32:41<01:47,  1.38it/s, loss=0.462]
Epoch 4/4:  88%|████████▊ | 1103/1250 [32:41<01:50,  1.33it/s, loss=0.462]
Epoch 4/4:  88%|████████▊ | 1103/1250 [32:42<01:50,  1.33it/s, loss=0.284]
Epoch 4/4:  88%|████████▊ | 1104/1250 [32:42<01:48,  1.35it/s, loss=0.284]
Epoch 4/4:  88%|████████▊ | 1104/1250 [32:43<01:48,  1.35it/s, loss=0.412]
Epoch 4/4:  88%|████████▊ | 1105/1250 [32:43<01:47,  1.35it/s, loss=0.412]
Epoch 4/4:  88%|████████▊ | 1105/1250 [32:43<01:47,  1.35it/s, loss=0.453]
Epoch 4/4:  88%|████████▊ | 1106/1250 [32:43<01:46,  1.36it/s, loss=0.453]
Epoch 4/4:  88%|████████▊ | 1106/1250 [32:44<01:46,  1.36it/s, loss=0.516]
Epoch 4/4:  89%|████████▊ | 1107/1250 [32:44<01:44,  1.36it/s, loss=0.516]
Epoch 4/4:  89%|████████▊ | 1107/1250 [32:45<01:44,  1.36it/s, loss=0.371]
Epoch 4/4:  89%|████████▊ | 1108/1250 [32:45<01:43,  1.37it/s, loss=0.371]
Epoch 4/4:  89%|████████▊ | 1108/1250 [32:46<01:43,  1.37it/s, loss=0.633]
Epoch 4/4:  89%|████████▊ | 1109/1250 [32:46<01:42,  1.37it/s, loss=0.633]
Epoch 4/4:  89%|████████▊ | 1109/1250 [32:46<01:42,  1.37it/s, loss=0.368]
Epoch 4/4:  89%|████████▉ | 1110/1250 [32:46<01:41,  1.38it/s, loss=0.368]
Epoch 4/4:  89%|████████▉ | 1110/1250 [32:47<01:41,  1.38it/s, loss=0.385]
Epoch 4/4:  89%|████████▉ | 1111/1250 [32:47<01:40,  1.39it/s, loss=0.385]
Epoch 4/4:  89%|████████▉ | 1111/1250 [32:48<01:40,  1.39it/s, loss=0.39] 
Epoch 4/4:  89%|████████▉ | 1112/1250 [32:48<01:40,  1.38it/s, loss=0.39]
Epoch 4/4:  89%|████████▉ | 1112/1250 [32:49<01:40,  1.38it/s, loss=0.599]
Epoch 4/4:  89%|████████▉ | 1113/1250 [32:49<01:41,  1.34it/s, loss=0.599]
Epoch 4/4:  89%|████████▉ | 1113/1250 [32:49<01:41,  1.34it/s, loss=0.488]
Epoch 4/4:  89%|████████▉ | 1114/1250 [32:49<01:45,  1.29it/s, loss=0.488]
Epoch 4/4:  89%|████████▉ | 1114/1250 [32:50<01:45,  1.29it/s, loss=0.442]
Epoch 4/4:  89%|████████▉ | 1115/1250 [32:50<01:47,  1.26it/s, loss=0.442]
Epoch 4/4:  89%|████████▉ | 1115/1250 [32:51<01:47,  1.26it/s, loss=0.374]
Epoch 4/4:  89%|████████▉ | 1116/1250 [32:51<01:46,  1.26it/s, loss=0.374]
Epoch 4/4:  89%|████████▉ | 1116/1250 [32:52<01:46,  1.26it/s, loss=0.4]  
Epoch 4/4:  89%|████████▉ | 1117/1250 [32:52<01:43,  1.28it/s, loss=0.4]
Epoch 4/4:  89%|████████▉ | 1117/1250 [32:53<01:43,  1.28it/s, loss=0.434]
Epoch 4/4:  89%|████████▉ | 1118/1250 [32:53<01:41,  1.30it/s, loss=0.434]
Epoch 4/4:  89%|████████▉ | 1118/1250 [32:53<01:41,  1.30it/s, loss=0.509]
Epoch 4/4:  90%|████████▉ | 1119/1250 [32:53<01:40,  1.30it/s, loss=0.509]
Epoch 4/4:  90%|████████▉ | 1119/1250 [32:54<01:40,  1.30it/s, loss=0.395]
Epoch 4/4:  90%|████████▉ | 1120/1250 [32:54<01:40,  1.29it/s, loss=0.395]
Epoch 4/4:  90%|████████▉ | 1120/1250 [32:55<01:40,  1.29it/s, loss=0.523]
Epoch 4/4:  90%|████████▉ | 1121/1250 [32:55<01:40,  1.28it/s, loss=0.523]
Epoch 4/4:  90%|████████▉ | 1121/1250 [32:56<01:40,  1.28it/s, loss=0.508]
Epoch 4/4:  90%|████████▉ | 1122/1250 [32:56<01:40,  1.27it/s, loss=0.508]
Epoch 4/4:  90%|████████▉ | 1122/1250 [32:57<01:40,  1.27it/s, loss=0.339]
Epoch 4/4:  90%|████████▉ | 1123/1250 [32:57<01:39,  1.27it/s, loss=0.339]
Epoch 4/4:  90%|████████▉ | 1123/1250 [32:57<01:39,  1.27it/s, loss=0.414]
Epoch 4/4:  90%|████████▉ | 1124/1250 [32:57<01:39,  1.27it/s, loss=0.414]
Epoch 4/4:  90%|████████▉ | 1124/1250 [32:58<01:39,  1.27it/s, loss=0.337]
Epoch 4/4:  90%|█████████ | 1125/1250 [32:58<01:40,  1.25it/s, loss=0.337]
Epoch 4/4:  90%|█████████ | 1125/1250 [32:59<01:40,  1.25it/s, loss=0.507]
Epoch 4/4:  90%|█████████ | 1126/1250 [32:59<01:41,  1.22it/s, loss=0.507]
Epoch 4/4:  90%|█████████ | 1126/1250 [33:00<01:41,  1.22it/s, loss=0.608]
Epoch 4/4:  90%|█████████ | 1127/1250 [33:00<01:43,  1.19it/s, loss=0.608]
Epoch 4/4:  90%|█████████ | 1127/1250 [33:01<01:43,  1.19it/s, loss=0.407]
Epoch 4/4:  90%|█████████ | 1128/1250 [33:01<01:44,  1.17it/s, loss=0.407]
Epoch 4/4:  90%|█████████ | 1128/1250 [33:02<01:44,  1.17it/s, loss=0.6]  
Epoch 4/4:  90%|█████████ | 1129/1250 [33:02<01:43,  1.17it/s, loss=0.6]
Epoch 4/4:  90%|█████████ | 1129/1250 [33:02<01:43,  1.17it/s, loss=0.353]
Epoch 4/4:  90%|█████████ | 1130/1250 [33:02<01:41,  1.18it/s, loss=0.353]
Epoch 4/4:  90%|█████████ | 1130/1250 [33:03<01:41,  1.18it/s, loss=0.688]
Epoch 4/4:  90%|█████████ | 1131/1250 [33:03<01:43,  1.15it/s, loss=0.688]
Epoch 4/4:  90%|█████████ | 1131/1250 [33:04<01:43,  1.15it/s, loss=0.404]
Epoch 4/4:  91%|█████████ | 1132/1250 [33:04<01:45,  1.12it/s, loss=0.404]
Epoch 4/4:  91%|█████████ | 1132/1250 [33:05<01:45,  1.12it/s, loss=0.515]
Epoch 4/4:  91%|█████████ | 1133/1250 [33:05<01:45,  1.11it/s, loss=0.515]
Epoch 4/4:  91%|█████████ | 1133/1250 [33:06<01:45,  1.11it/s, loss=0.426]
Epoch 4/4:  91%|█████████ | 1134/1250 [33:06<01:43,  1.12it/s, loss=0.426]
Epoch 4/4:  91%|█████████ | 1134/1250 [33:07<01:43,  1.12it/s, loss=0.69] 
Epoch 4/4:  91%|█████████ | 1135/1250 [33:07<01:40,  1.14it/s, loss=0.69]
Epoch 4/4:  91%|█████████ | 1135/1250 [33:08<01:40,  1.14it/s, loss=0.535]
Epoch 4/4:  91%|█████████ | 1136/1250 [33:08<01:39,  1.15it/s, loss=0.535]
Epoch 4/4:  91%|█████████ | 1136/1250 [33:09<01:39,  1.15it/s, loss=0.588]
Epoch 4/4:  91%|█████████ | 1137/1250 [33:09<01:40,  1.13it/s, loss=0.588]
Epoch 4/4:  91%|█████████ | 1137/1250 [33:10<01:40,  1.13it/s, loss=0.584]
Epoch 4/4:  91%|█████████ | 1138/1250 [33:10<01:41,  1.11it/s, loss=0.584]
Epoch 4/4:  91%|█████████ | 1138/1250 [33:11<01:41,  1.11it/s, loss=0.484]
Epoch 4/4:  91%|█████████ | 1139/1250 [33:11<01:42,  1.09it/s, loss=0.484]
Epoch 4/4:  91%|█████████ | 1139/1250 [33:12<01:42,  1.09it/s, loss=0.483]
Epoch 4/4:  91%|█████████ | 1140/1250 [33:12<01:41,  1.09it/s, loss=0.483]
Epoch 4/4:  91%|█████████ | 1140/1250 [33:12<01:41,  1.09it/s, loss=0.574]
Epoch 4/4:  91%|█████████▏| 1141/1250 [33:12<01:38,  1.11it/s, loss=0.574]
Epoch 4/4:  91%|█████████▏| 1141/1250 [33:13<01:38,  1.11it/s, loss=0.386]
Epoch 4/4:  91%|█████████▏| 1142/1250 [33:13<01:36,  1.12it/s, loss=0.386]
Epoch 4/4:  91%|█████████▏| 1142/1250 [33:14<01:36,  1.12it/s, loss=0.453]
Epoch 4/4:  91%|█████████▏| 1143/1250 [33:14<01:35,  1.13it/s, loss=0.453]
Epoch 4/4:  91%|█████████▏| 1143/1250 [33:15<01:35,  1.13it/s, loss=0.44] 
Epoch 4/4:  92%|█████████▏| 1144/1250 [33:15<01:35,  1.11it/s, loss=0.44]
Epoch 4/4:  92%|█████████▏| 1144/1250 [33:16<01:35,  1.11it/s, loss=0.462]
Epoch 4/4:  92%|█████████▏| 1145/1250 [33:16<01:35,  1.09it/s, loss=0.462]
Epoch 4/4:  92%|█████████▏| 1145/1250 [33:17<01:35,  1.09it/s, loss=0.535]
Epoch 4/4:  92%|█████████▏| 1146/1250 [33:17<01:34,  1.10it/s, loss=0.535]
Epoch 4/4:  92%|█████████▏| 1146/1250 [33:18<01:34,  1.10it/s, loss=0.384]
Epoch 4/4:  92%|█████████▏| 1147/1250 [33:18<01:31,  1.12it/s, loss=0.384]
Epoch 4/4:  92%|█████████▏| 1147/1250 [33:19<01:31,  1.12it/s, loss=0.483]
Epoch 4/4:  92%|█████████▏| 1148/1250 [33:19<01:30,  1.12it/s, loss=0.483]
Epoch 4/4:  92%|█████████▏| 1148/1250 [33:20<01:30,  1.12it/s, loss=0.438]
Epoch 4/4:  92%|█████████▏| 1149/1250 [33:20<01:30,  1.11it/s, loss=0.438]
Epoch 4/4:  92%|█████████▏| 1149/1250 [33:21<01:30,  1.11it/s, loss=0.308]
Epoch 4/4:  92%|█████████▏| 1150/1250 [33:21<01:32,  1.08it/s, loss=0.308]
Epoch 4/4:  92%|█████████▏| 1150/1250 [33:22<01:32,  1.08it/s, loss=0.516]
Epoch 4/4:  92%|█████████▏| 1151/1250 [33:22<01:31,  1.08it/s, loss=0.516]
Epoch 4/4:  92%|█████████▏| 1151/1250 [33:22<01:31,  1.08it/s, loss=0.439]
Epoch 4/4:  92%|█████████▏| 1152/1250 [33:22<01:29,  1.10it/s, loss=0.439]
Epoch 4/4:  92%|█████████▏| 1152/1250 [33:23<01:29,  1.10it/s, loss=0.45] 
Epoch 4/4:  92%|█████████▏| 1153/1250 [33:23<01:27,  1.11it/s, loss=0.45]
Epoch 4/4:  92%|█████████▏| 1153/1250 [33:24<01:27,  1.11it/s, loss=0.519]
Epoch 4/4:  92%|█████████▏| 1154/1250 [33:24<01:28,  1.09it/s, loss=0.519]
Epoch 4/4:  92%|█████████▏| 1154/1250 [33:25<01:28,  1.09it/s, loss=0.438]
Epoch 4/4:  92%|█████████▏| 1155/1250 [33:25<01:29,  1.06it/s, loss=0.438]
Epoch 4/4:  92%|█████████▏| 1155/1250 [33:26<01:29,  1.06it/s, loss=0.478]
Epoch 4/4:  92%|█████████▏| 1156/1250 [33:26<01:29,  1.06it/s, loss=0.478]
Epoch 4/4:  92%|█████████▏| 1156/1250 [33:27<01:29,  1.06it/s, loss=0.382]
Epoch 4/4:  93%|█████████▎| 1157/1250 [33:27<01:27,  1.07it/s, loss=0.382]
Epoch 4/4:  93%|█████████▎| 1157/1250 [33:28<01:27,  1.07it/s, loss=0.464]
Epoch 4/4:  93%|█████████▎| 1158/1250 [33:28<01:25,  1.08it/s, loss=0.464]
Epoch 4/4:  93%|█████████▎| 1158/1250 [33:29<01:25,  1.08it/s, loss=0.505]
Epoch 4/4:  93%|█████████▎| 1159/1250 [33:29<01:26,  1.05it/s, loss=0.505]
Epoch 4/4:  93%|█████████▎| 1159/1250 [33:30<01:26,  1.05it/s, loss=0.394]
Epoch 4/4:  93%|█████████▎| 1160/1250 [33:30<01:32,  1.03s/it, loss=0.394]
Epoch 4/4:  93%|█████████▎| 1160/1250 [33:31<01:32,  1.03s/it, loss=0.431]
Epoch 4/4:  93%|█████████▎| 1161/1250 [33:31<01:33,  1.05s/it, loss=0.431]
Epoch 4/4:  93%|█████████▎| 1161/1250 [33:32<01:33,  1.05s/it, loss=0.408]
Epoch 4/4:  93%|█████████▎| 1162/1250 [33:32<01:29,  1.02s/it, loss=0.408]
Epoch 4/4:  93%|█████████▎| 1162/1250 [33:33<01:29,  1.02s/it, loss=0.404]
Epoch 4/4:  93%|█████████▎| 1163/1250 [33:33<01:26,  1.00it/s, loss=0.404]
Epoch 4/4:  93%|█████████▎| 1163/1250 [33:34<01:26,  1.00it/s, loss=0.429]
Epoch 4/4:  93%|█████████▎| 1164/1250 [33:34<01:25,  1.01it/s, loss=0.429]
Epoch 4/4:  93%|█████████▎| 1164/1250 [33:35<01:25,  1.01it/s, loss=0.386]
Epoch 4/4:  93%|█████████▎| 1165/1250 [33:35<01:25,  1.01s/it, loss=0.386]
Epoch 4/4:  93%|█████████▎| 1165/1250 [33:36<01:25,  1.01s/it, loss=0.471]
Epoch 4/4:  93%|█████████▎| 1166/1250 [33:36<01:24,  1.01s/it, loss=0.471]
Epoch 4/4:  93%|█████████▎| 1166/1250 [33:37<01:24,  1.01s/it, loss=0.395]
Epoch 4/4:  93%|█████████▎| 1167/1250 [33:37<01:21,  1.01it/s, loss=0.395]
Epoch 4/4:  93%|█████████▎| 1167/1250 [33:38<01:21,  1.01it/s, loss=0.364]
Epoch 4/4:  93%|█████████▎| 1168/1250 [33:38<01:20,  1.02it/s, loss=0.364]
Epoch 4/4:  93%|█████████▎| 1168/1250 [33:39<01:20,  1.02it/s, loss=0.324]
Epoch 4/4:  94%|█████████▎| 1169/1250 [33:39<01:21,  1.01s/it, loss=0.324]
Epoch 4/4:  94%|█████████▎| 1169/1250 [33:40<01:21,  1.01s/it, loss=0.378]
Epoch 4/4:  94%|█████████▎| 1170/1250 [33:40<01:25,  1.07s/it, loss=0.378]
Epoch 4/4:  94%|█████████▎| 1170/1250 [33:42<01:25,  1.07s/it, loss=0.484]
Epoch 4/4:  94%|█████████▎| 1171/1250 [33:42<01:27,  1.11s/it, loss=0.484]
Epoch 4/4:  94%|█████████▎| 1171/1250 [33:43<01:27,  1.11s/it, loss=0.307]
Epoch 4/4:  94%|█████████▍| 1172/1250 [33:43<01:24,  1.09s/it, loss=0.307]
Epoch 4/4:  94%|█████████▍| 1172/1250 [33:44<01:24,  1.09s/it, loss=0.379]
Epoch 4/4:  94%|█████████▍| 1173/1250 [33:44<01:21,  1.05s/it, loss=0.379]
Epoch 4/4:  94%|█████████▍| 1173/1250 [33:45<01:21,  1.05s/it, loss=0.418]
Epoch 4/4:  94%|█████████▍| 1174/1250 [33:45<01:18,  1.04s/it, loss=0.418]
Epoch 4/4:  94%|█████████▍| 1174/1250 [33:46<01:18,  1.04s/it, loss=0.581]
Epoch 4/4:  94%|█████████▍| 1175/1250 [33:46<01:19,  1.05s/it, loss=0.581]
Epoch 4/4:  94%|█████████▍| 1175/1250 [33:47<01:19,  1.05s/it, loss=0.601]
Epoch 4/4:  94%|█████████▍| 1176/1250 [33:47<01:18,  1.06s/it, loss=0.601]
Epoch 4/4:  94%|█████████▍| 1176/1250 [33:48<01:18,  1.06s/it, loss=0.525]
Epoch 4/4:  94%|█████████▍| 1177/1250 [33:48<01:15,  1.04s/it, loss=0.525]
Epoch 4/4:  94%|█████████▍| 1177/1250 [33:49<01:15,  1.04s/it, loss=0.356]
Epoch 4/4:  94%|█████████▍| 1178/1250 [33:49<01:13,  1.03s/it, loss=0.356]
Epoch 4/4:  94%|█████████▍| 1178/1250 [33:50<01:13,  1.03s/it, loss=0.308]
Epoch 4/4:  94%|█████████▍| 1179/1250 [33:50<01:16,  1.07s/it, loss=0.308]
Epoch 4/4:  94%|█████████▍| 1179/1250 [33:51<01:16,  1.07s/it, loss=0.434]
Epoch 4/4:  94%|█████████▍| 1180/1250 [33:51<01:20,  1.15s/it, loss=0.434]
Epoch 4/4:  94%|█████████▍| 1180/1250 [33:52<01:20,  1.15s/it, loss=0.61] 
Epoch 4/4:  94%|█████████▍| 1181/1250 [33:52<01:18,  1.14s/it, loss=0.61]
Epoch 4/4:  94%|█████████▍| 1181/1250 [33:53<01:18,  1.14s/it, loss=0.457]
Epoch 4/4:  95%|█████████▍| 1182/1250 [33:53<01:14,  1.10s/it, loss=0.457]
Epoch 4/4:  95%|█████████▍| 1182/1250 [33:55<01:14,  1.10s/it, loss=0.352]
Epoch 4/4:  95%|█████████▍| 1183/1250 [33:55<01:13,  1.10s/it, loss=0.352]
Epoch 4/4:  95%|█████████▍| 1183/1250 [33:56<01:13,  1.10s/it, loss=0.665]
Epoch 4/4:  95%|█████████▍| 1184/1250 [33:56<01:12,  1.09s/it, loss=0.665]
Epoch 4/4:  95%|█████████▍| 1184/1250 [33:57<01:12,  1.09s/it, loss=0.369]
Epoch 4/4:  95%|█████████▍| 1185/1250 [33:57<01:11,  1.11s/it, loss=0.369]
Epoch 4/4:  95%|█████████▍| 1185/1250 [33:58<01:11,  1.11s/it, loss=0.712]
Epoch 4/4:  95%|█████████▍| 1186/1250 [33:58<01:09,  1.09s/it, loss=0.712]
Epoch 4/4:  95%|█████████▍| 1186/1250 [33:59<01:09,  1.09s/it, loss=0.67] 
Epoch 4/4:  95%|█████████▍| 1187/1250 [33:59<01:08,  1.08s/it, loss=0.67]
Epoch 4/4:  95%|█████████▍| 1187/1250 [34:00<01:08,  1.08s/it, loss=0.354]
Epoch 4/4:  95%|█████████▌| 1188/1250 [34:00<01:06,  1.08s/it, loss=0.354]
Epoch 4/4:  95%|█████████▌| 1188/1250 [34:01<01:06,  1.08s/it, loss=0.497]
Epoch 4/4:  95%|█████████▌| 1189/1250 [34:01<01:05,  1.07s/it, loss=0.497]
Epoch 4/4:  95%|█████████▌| 1189/1250 [34:02<01:05,  1.07s/it, loss=0.474]
Epoch 4/4:  95%|█████████▌| 1190/1250 [34:02<01:04,  1.07s/it, loss=0.474]
Epoch 4/4:  95%|█████████▌| 1190/1250 [34:03<01:04,  1.07s/it, loss=0.718]
Epoch 4/4:  95%|█████████▌| 1191/1250 [34:03<01:01,  1.04s/it, loss=0.718]
Epoch 4/4:  95%|█████████▌| 1191/1250 [34:04<01:01,  1.04s/it, loss=0.527]
Epoch 4/4:  95%|█████████▌| 1192/1250 [34:04<01:00,  1.04s/it, loss=0.527]
Epoch 4/4:  95%|█████████▌| 1192/1250 [34:05<01:00,  1.04s/it, loss=0.512]
Epoch 4/4:  95%|█████████▌| 1193/1250 [34:05<01:00,  1.06s/it, loss=0.512]
Epoch 4/4:  95%|█████████▌| 1193/1250 [34:06<01:00,  1.06s/it, loss=0.602]
Epoch 4/4:  96%|█████████▌| 1194/1250 [34:06<01:00,  1.08s/it, loss=0.602]
Epoch 4/4:  96%|█████████▌| 1194/1250 [34:07<01:00,  1.08s/it, loss=0.483]
Epoch 4/4:  96%|█████████▌| 1195/1250 [34:07<01:00,  1.11s/it, loss=0.483]
Epoch 4/4:  96%|█████████▌| 1195/1250 [34:09<01:00,  1.11s/it, loss=0.519]
Epoch 4/4:  96%|█████████▌| 1196/1250 [34:09<01:07,  1.24s/it, loss=0.519]
Epoch 4/4:  96%|█████████▌| 1196/1250 [34:10<01:07,  1.24s/it, loss=0.544]
Epoch 4/4:  96%|█████████▌| 1197/1250 [34:10<01:05,  1.24s/it, loss=0.544]
Epoch 4/4:  96%|█████████▌| 1197/1250 [34:12<01:05,  1.24s/it, loss=0.448]
Epoch 4/4:  96%|█████████▌| 1198/1250 [34:12<01:05,  1.26s/it, loss=0.448]
Epoch 4/4:  96%|█████████▌| 1198/1250 [34:13<01:05,  1.26s/it, loss=0.488]
Epoch 4/4:  96%|█████████▌| 1199/1250 [34:13<01:01,  1.22s/it, loss=0.488]
Epoch 4/4:  96%|█████████▌| 1199/1250 [34:14<01:01,  1.22s/it, loss=0.449]
Epoch 4/4:  96%|█████████▌| 1200/1250 [34:14<00:57,  1.16s/it, loss=0.449]
Epoch 4/4:  96%|█████████▌| 1200/1250 [34:15<00:57,  1.16s/it, loss=0.484]
Epoch 4/4:  96%|█████████▌| 1201/1250 [34:15<00:53,  1.09s/it, loss=0.484]
Epoch 4/4:  96%|█████████▌| 1201/1250 [34:16<00:53,  1.09s/it, loss=0.455]
Epoch 4/4:  96%|█████████▌| 1202/1250 [34:16<00:54,  1.13s/it, loss=0.455]
Epoch 4/4:  96%|█████████▌| 1202/1250 [34:17<00:54,  1.13s/it, loss=0.392]
Epoch 4/4:  96%|█████████▌| 1203/1250 [34:17<00:55,  1.18s/it, loss=0.392]
Epoch 4/4:  96%|█████████▌| 1203/1250 [34:18<00:55,  1.18s/it, loss=0.392]
Epoch 4/4:  96%|█████████▋| 1204/1250 [34:18<00:54,  1.19s/it, loss=0.392]
Epoch 4/4:  96%|█████████▋| 1204/1250 [34:20<00:54,  1.19s/it, loss=0.739]
Epoch 4/4:  96%|█████████▋| 1205/1250 [34:20<00:53,  1.19s/it, loss=0.739]
Epoch 4/4:  96%|█████████▋| 1205/1250 [34:21<00:53,  1.19s/it, loss=0.437]
Epoch 4/4:  96%|█████████▋| 1206/1250 [34:21<00:55,  1.27s/it, loss=0.437]
Epoch 4/4:  96%|█████████▋| 1206/1250 [34:22<00:55,  1.27s/it, loss=0.28] 
Epoch 4/4:  97%|█████████▋| 1207/1250 [34:22<00:53,  1.26s/it, loss=0.28]
Epoch 4/4:  97%|█████████▋| 1207/1250 [34:23<00:53,  1.26s/it, loss=0.511]
Epoch 4/4:  97%|█████████▋| 1208/1250 [34:23<00:49,  1.18s/it, loss=0.511]
Epoch 4/4:  97%|█████████▋| 1208/1250 [34:24<00:49,  1.18s/it, loss=0.407]
Epoch 4/4:  97%|█████████▋| 1209/1250 [34:24<00:45,  1.12s/it, loss=0.407]
Epoch 4/4:  97%|█████████▋| 1209/1250 [34:25<00:45,  1.12s/it, loss=0.434]
Epoch 4/4:  97%|█████████▋| 1210/1250 [34:25<00:45,  1.13s/it, loss=0.434]
Epoch 4/4:  97%|█████████▋| 1210/1250 [34:27<00:45,  1.13s/it, loss=0.504]
Epoch 4/4:  97%|█████████▋| 1211/1250 [34:27<00:44,  1.14s/it, loss=0.504]
Epoch 4/4:  97%|█████████▋| 1211/1250 [34:28<00:44,  1.14s/it, loss=0.503]
Epoch 4/4:  97%|█████████▋| 1212/1250 [34:28<00:41,  1.10s/it, loss=0.503]
Epoch 4/4:  97%|█████████▋| 1212/1250 [34:28<00:41,  1.10s/it, loss=0.387]
Epoch 4/4:  97%|█████████▋| 1213/1250 [34:28<00:38,  1.05s/it, loss=0.387]
Epoch 4/4:  97%|█████████▋| 1213/1250 [34:29<00:38,  1.05s/it, loss=0.434]
Epoch 4/4:  97%|█████████▋| 1214/1250 [34:29<00:36,  1.02s/it, loss=0.434]
Epoch 4/4:  97%|█████████▋| 1214/1250 [34:31<00:36,  1.02s/it, loss=0.477]
Epoch 4/4:  97%|█████████▋| 1215/1250 [34:31<00:37,  1.06s/it, loss=0.477]
Epoch 4/4:  97%|█████████▋| 1215/1250 [34:32<00:37,  1.06s/it, loss=0.551]
Epoch 4/4:  97%|█████████▋| 1216/1250 [34:32<00:36,  1.08s/it, loss=0.551]
Epoch 4/4:  97%|█████████▋| 1216/1250 [34:33<00:36,  1.08s/it, loss=0.508]
Epoch 4/4:  97%|█████████▋| 1217/1250 [34:33<00:34,  1.03s/it, loss=0.508]
Epoch 4/4:  97%|█████████▋| 1217/1250 [34:34<00:34,  1.03s/it, loss=0.421]
Epoch 4/4:  97%|█████████▋| 1218/1250 [34:34<00:32,  1.01s/it, loss=0.421]
Epoch 4/4:  97%|█████████▋| 1218/1250 [34:35<00:32,  1.01s/it, loss=0.426]
Epoch 4/4:  98%|█████████▊| 1219/1250 [34:35<00:31,  1.02s/it, loss=0.426]
Epoch 4/4:  98%|█████████▊| 1219/1250 [34:36<00:31,  1.02s/it, loss=0.589]
Epoch 4/4:  98%|█████████▊| 1220/1250 [34:36<00:33,  1.10s/it, loss=0.589]
Epoch 4/4:  98%|█████████▊| 1220/1250 [34:37<00:33,  1.10s/it, loss=0.424]
Epoch 4/4:  98%|█████████▊| 1221/1250 [34:37<00:31,  1.09s/it, loss=0.424]
Epoch 4/4:  98%|█████████▊| 1221/1250 [34:38<00:31,  1.09s/it, loss=0.68] 
Epoch 4/4:  98%|█████████▊| 1222/1250 [34:38<00:28,  1.03s/it, loss=0.68]
Epoch 4/4:  98%|█████████▊| 1222/1250 [34:39<00:28,  1.03s/it, loss=0.578]
Epoch 4/4:  98%|█████████▊| 1223/1250 [34:39<00:26,  1.02it/s, loss=0.578]
Epoch 4/4:  98%|█████████▊| 1223/1250 [34:40<00:26,  1.02it/s, loss=0.355]
Epoch 4/4:  98%|█████████▊| 1224/1250 [34:40<00:24,  1.05it/s, loss=0.355]
Epoch 4/4:  98%|█████████▊| 1224/1250 [34:41<00:24,  1.05it/s, loss=0.375]
Epoch 4/4:  98%|█████████▊| 1225/1250 [34:41<00:25,  1.01s/it, loss=0.375]
Epoch 4/4:  98%|█████████▊| 1225/1250 [34:42<00:25,  1.01s/it, loss=0.615]
Epoch 4/4:  98%|█████████▊| 1226/1250 [34:42<00:24,  1.04s/it, loss=0.615]
Epoch 4/4:  98%|█████████▊| 1226/1250 [34:43<00:24,  1.04s/it, loss=0.365]
Epoch 4/4:  98%|█████████▊| 1227/1250 [34:43<00:23,  1.02s/it, loss=0.365]
Epoch 4/4:  98%|█████████▊| 1227/1250 [34:44<00:23,  1.02s/it, loss=0.316]
Epoch 4/4:  98%|█████████▊| 1228/1250 [34:44<00:21,  1.03it/s, loss=0.316]
Epoch 4/4:  98%|█████████▊| 1228/1250 [34:45<00:21,  1.03it/s, loss=0.596]
Epoch 4/4:  98%|█████████▊| 1229/1250 [34:45<00:19,  1.05it/s, loss=0.596]
Epoch 4/4:  98%|█████████▊| 1229/1250 [34:46<00:19,  1.05it/s, loss=0.461]
Epoch 4/4:  98%|█████████▊| 1230/1250 [34:46<00:20,  1.02s/it, loss=0.461]
Epoch 4/4:  98%|█████████▊| 1230/1250 [34:47<00:20,  1.02s/it, loss=0.437]
Epoch 4/4:  98%|█████████▊| 1231/1250 [34:47<00:20,  1.06s/it, loss=0.437]
Epoch 4/4:  98%|█████████▊| 1231/1250 [34:48<00:20,  1.06s/it, loss=0.387]
Epoch 4/4:  99%|█████████▊| 1232/1250 [34:48<00:18,  1.03s/it, loss=0.387]
Epoch 4/4:  99%|█████████▊| 1232/1250 [34:49<00:18,  1.03s/it, loss=0.499]
Epoch 4/4:  99%|█████████▊| 1233/1250 [34:49<00:16,  1.02it/s, loss=0.499]
Epoch 4/4:  99%|█████████▊| 1233/1250 [34:50<00:16,  1.02it/s, loss=0.493]
Epoch 4/4:  99%|█████████▊| 1234/1250 [34:50<00:15,  1.06it/s, loss=0.493]
Epoch 4/4:  99%|█████████▊| 1234/1250 [34:51<00:15,  1.06it/s, loss=0.475]
Epoch 4/4:  99%|█████████▉| 1235/1250 [34:51<00:14,  1.03it/s, loss=0.475]
Epoch 4/4:  99%|█████████▉| 1235/1250 [34:52<00:14,  1.03it/s, loss=0.41] 
Epoch 4/4:  99%|█████████▉| 1236/1250 [34:52<00:14,  1.02s/it, loss=0.41]
Epoch 4/4:  99%|█████████▉| 1236/1250 [34:53<00:14,  1.02s/it, loss=0.393]
Epoch 4/4:  99%|█████████▉| 1237/1250 [34:53<00:12,  1.01it/s, loss=0.393]
Epoch 4/4:  99%|█████████▉| 1237/1250 [34:54<00:12,  1.01it/s, loss=0.45] 
Epoch 4/4:  99%|█████████▉| 1238/1250 [34:54<00:11,  1.05it/s, loss=0.45]
Epoch 4/4:  99%|█████████▉| 1238/1250 [34:54<00:11,  1.05it/s, loss=0.464]
Epoch 4/4:  99%|█████████▉| 1239/1250 [34:54<00:10,  1.08it/s, loss=0.464]
Epoch 4/4:  99%|█████████▉| 1239/1250 [34:55<00:10,  1.08it/s, loss=0.469]
Epoch 4/4:  99%|█████████▉| 1240/1250 [34:55<00:09,  1.05it/s, loss=0.469]
Epoch 4/4:  99%|█████████▉| 1240/1250 [34:57<00:09,  1.05it/s, loss=0.435]
Epoch 4/4:  99%|█████████▉| 1241/1250 [34:57<00:09,  1.01s/it, loss=0.435]
Epoch 4/4:  99%|█████████▉| 1241/1250 [34:58<00:09,  1.01s/it, loss=0.381]
Epoch 4/4:  99%|█████████▉| 1242/1250 [34:58<00:07,  1.01it/s, loss=0.381]
Epoch 4/4:  99%|█████████▉| 1242/1250 [34:58<00:07,  1.01it/s, loss=0.364]
Epoch 4/4:  99%|█████████▉| 1243/1250 [34:58<00:06,  1.04it/s, loss=0.364]
Epoch 4/4:  99%|█████████▉| 1243/1250 [34:59<00:06,  1.04it/s, loss=0.414]
Epoch 4/4: 100%|█████████▉| 1244/1250 [34:59<00:05,  1.07it/s, loss=0.414]
Epoch 4/4: 100%|█████████▉| 1244/1250 [35:00<00:05,  1.07it/s, loss=0.554]
Epoch 4/4: 100%|█████████▉| 1245/1250 [35:00<00:04,  1.05it/s, loss=0.554]
Epoch 4/4: 100%|█████████▉| 1245/1250 [35:02<00:04,  1.05it/s, loss=0.443]
Epoch 4/4: 100%|█████████▉| 1246/1250 [35:02<00:04,  1.07s/it, loss=0.443]
Epoch 4/4: 100%|█████████▉| 1246/1250 [35:03<00:04,  1.07s/it, loss=0.492]
Epoch 4/4: 100%|█████████▉| 1247/1250 [35:03<00:03,  1.06s/it, loss=0.492]
Epoch 4/4: 100%|█████████▉| 1247/1250 [35:04<00:03,  1.06s/it, loss=0.249]
Epoch 4/4: 100%|█████████▉| 1248/1250 [35:04<00:02,  1.02s/it, loss=0.249]
Epoch 4/4: 100%|█████████▉| 1248/1250 [35:05<00:02,  1.02s/it, loss=0.495]
Epoch 4/4: 100%|█████████▉| 1249/1250 [35:05<00:01,  1.01s/it, loss=0.495]
Epoch 4/4: 100%|█████████▉| 1249/1250 [35:06<00:01,  1.01s/it, loss=0.562]
Epoch 4/4: 100%|██████████| 1250/1250 [35:06<00:00,  1.02s/it, loss=0.562]
Epoch 4/4: 100%|██████████| 1250/1250 [35:06<00:00,  1.68s/it, loss=0.562]
Epoch 4 Avg Loss: 0.4481

💾 Saving tuned VQA model to checkpoints/oculus_instruct_v1
✅ Instruction Tuning Complete!