diff --git "a/ncr_cclue/without_knowledge_random/train.log" "b/ncr_cclue/without_knowledge_random/train.log"
new file mode 100644--- /dev/null
+++ "b/ncr_cclue/without_knowledge_random/train.log"
@@ -0,0 +1,259 @@
+model training desc:  不做知识选择，使用NCR+CCLUE数据集，随机选择的关键句训练
+2023-12-10 12:22:55.340 | INFO     | __main__:init_components:108 - Initializing components...
+You are using the default legacy behaviour of the <class 'transformers.models.llama.tokenization_llama.LlamaTokenizer'>. If you see this, DO NOT PANIC! This is expected, and simply means that the `legacy` (previous) behavior will be used so nothing changes for you. If you want to use the new behaviour, set `legacy=False`. This should only be set if you understand what it means, and thouroughly read the reason why this was added as explained in https://github.com/huggingface/transformers/pull/24565
+2023-12-10 12:24:20.449 | INFO     | __main__:init_components:143 - 
+
+2023-12-10 12:24:20.449 | INFO     | __main__:init_components:144 - ********************
+2023-12-10 12:24:20.449 | INFO     | __main__:init_components:145 - using TechGPT-7B
+2023-12-10 12:24:20.449 | INFO     | __main__:init_components:146 - ********************
+2023-12-10 12:24:20.449 | INFO     | __main__:init_components:147 - 
+
+memory footprint of model: 5.472740173339844 GB
+trainable params: 319,815,680 || all params: 7,447,007,232 || trainable%: 4.294553100818044
+2023-12-10 12:24:23.862 | INFO     | component.dataset:__init__:14 - Loading data: /data0/maqi/KGLQA-data/datasets/merge/random_select/without_knowledge_random_instruction/train.jsonl
+2023-12-10 12:24:24.188 | INFO     | component.dataset:__init__:19 - there are 18892 data in dataset
+2023-12-10 12:24:24.205 | INFO     | __main__:main:231 - *** starting training ***
+
  0%|          | 0/4722 [00:00<?, ?it/s]
  0%|          | 1/4722 [00:47<61:41:12, 47.04s/it]
  0%|          | 2/4722 [01:27<56:13:20, 42.88s/it]
  0%|          | 3/4722 [02:08<55:05:23, 42.03s/it]
  0%|          | 4/4722 [02:47<53:30:14, 40.83s/it]
  0%|          | 5/4722 [03:22<50:50:29, 38.80s/it]
  0%|          | 6/4722 [04:01<51:04:18, 38.99s/it]
  0%|          | 7/4722 [04:42<52:06:30, 39.79s/it]
  0%|          | 8/4722 [05:19<50:31:28, 38.58s/it]
  0%|          | 9/4722 [05:59<51:25:14, 39.28s/it]
  0%|          | 10/4722 [06:40<52:08:49, 39.84s/it]
  0%|          | 11/4722 [07:18<51:15:07, 39.17s/it]
  0%|          | 12/4722 [07:54<50:04:11, 38.27s/it]
  0%|          | 13/4722 [08:32<49:39:50, 37.97s/it]
  0%|          | 14/4722 [09:12<50:42:52, 38.78s/it]
  0%|          | 15/4722 [09:46<48:44:47, 37.28s/it]
  0%|          | 16/4722 [10:25<49:15:29, 37.68s/it]
  0%|          | 17/4722 [11:05<50:23:47, 38.56s/it]
  0%|          | 18/4722 [11:49<52:27:54, 40.15s/it]
  0%|          | 19/4722 [12:24<50:22:47, 38.56s/it]
  0%|          | 20/4722 [13:01<49:55:48, 38.23s/it]
                                                    
{'loss': 6.0331, 'learning_rate': 4.016913319238901e-06, 'global_step': 20, 'epoch': 0.01}
+
  0%|          | 20/4722 [13:01<49:55:48, 38.23s/it]
  0%|          | 21/4722 [13:36<48:19:21, 37.01s/it]
  0%|          | 22/4722 [14:15<49:16:51, 37.75s/it]
  0%|          | 23/4722 [14:50<48:21:23, 37.05s/it]
  1%|          | 24/4722 [15:29<49:06:04, 37.63s/it]
  1%|          | 25/4722 [16:08<49:25:57, 37.89s/it]
  1%|          | 26/4722 [16:47<49:53:42, 38.25s/it]
  1%|          | 27/4722 [17:32<52:24:17, 40.18s/it]
  1%|          | 28/4722 [18:09<51:22:24, 39.40s/it]
  1%|          | 29/4722 [18:50<51:43:38, 39.68s/it]
  1%|          | 30/4722 [19:35<53:52:05, 41.33s/it]
  1%|          | 31/4722 [20:18<54:36:16, 41.91s/it]
  1%|          | 32/4722 [21:01<54:53:21, 42.13s/it]
  1%|          | 33/4722 [21:38<52:48:32, 40.54s/it]
  1%|          | 34/4722 [22:16<51:47:58, 39.78s/it]
  1%|          | 35/4722 [22:52<50:33:49, 38.84s/it]
  1%|          | 36/4722 [23:28<49:26:14, 37.98s/it]
  1%|          | 37/4722 [24:04<48:32:23, 37.30s/it]
  1%|          | 38/4722 [24:51<52:33:51, 40.40s/it]
  1%|          | 39/4722 [25:31<52:09:31, 40.10s/it]
  1%|          | 40/4722 [26:05<49:48:03, 38.29s/it]
                                                    
{'loss': 5.4374, 'learning_rate': 8.24524312896406e-06, 'global_step': 40, 'epoch': 0.03}
+
  1%|          | 40/4722 [26:05<49:48:03, 38.29s/it]
  1%|          | 41/4722 [26:51<52:40:21, 40.51s/it]
  1%|          | 42/4722 [27:25<50:21:43, 38.74s/it]
  1%|          | 43/4722 [28:04<50:23:57, 38.78s/it]
  1%|          | 44/4722 [28:41<49:41:53, 38.25s/it]
  1%|          | 45/4722 [29:20<49:51:33, 38.38s/it]
  1%|          | 46/4722 [29:50<46:50:28, 36.06s/it]
  1%|          | 47/4722 [30:29<47:50:08, 36.84s/it]
  1%|          | 48/4722 [31:06<47:46:23, 36.80s/it]
  1%|          | 49/4722 [31:47<49:20:13, 38.01s/it]
  1%|          | 50/4722 [32:34<52:49:41, 40.71s/it]
  1%|          | 51/4722 [33:16<53:17:14, 41.07s/it]
  1%|          | 52/4722 [33:57<53:18:33, 41.09s/it]
  1%|          | 53/4722 [34:36<52:30:36, 40.49s/it]
  1%|          | 54/4722 [35:19<53:44:34, 41.45s/it]
  1%|          | 55/4722 [35:56<51:50:39, 39.99s/it]
  1%|          | 56/4722 [36:37<52:14:02, 40.30s/it]
  1%|          | 57/4722 [37:11<49:39:11, 38.32s/it]
  1%|          | 58/4722 [37:50<49:50:04, 38.47s/it]
  1%|          | 59/4722 [38:36<52:54:45, 40.85s/it]
  1%|▏         | 60/4722 [39:16<52:24:40, 40.47s/it]
                                                    
{'loss': 4.6579, 'learning_rate': 1.2473572938689217e-05, 'global_step': 60, 'epoch': 0.04}
+
  1%|▏         | 60/4722 [39:16<52:24:40, 40.47s/it]
  1%|▏         | 61/4722 [39:50<50:07:32, 38.72s/it]
  1%|▏         | 62/4722 [40:35<52:33:58, 40.61s/it]
  1%|▏         | 63/4722 [41:15<52:10:55, 40.32s/it]
  1%|▏         | 64/4722 [41:54<51:31:06, 39.82s/it]
  1%|▏         | 65/4722 [42:26<48:47:19, 37.72s/it]
  1%|▏         | 66/4722 [43:04<48:39:50, 37.63s/it]
  1%|▏         | 67/4722 [43:42<48:59:37, 37.89s/it]
  1%|▏         | 68/4722 [44:20<48:57:03, 37.86s/it]
  1%|▏         | 69/4722 [44:56<48:18:00, 37.37s/it]
  1%|▏         | 70/4722 [45:33<48:07:26, 37.24s/it]
  2%|▏         | 71/4722 [46:12<48:50:45, 37.81s/it]
  2%|▏         | 72/4722 [46:51<49:10:00, 38.06s/it]
  2%|▏         | 73/4722 [47:28<48:40:19, 37.69s/it]
  2%|▏         | 74/4722 [48:12<51:09:56, 39.63s/it]
  2%|▏         | 75/4722 [48:47<49:17:11, 38.18s/it]
  2%|▏         | 76/4722 [49:26<49:47:16, 38.58s/it]
  2%|▏         | 77/4722 [50:08<50:53:34, 39.44s/it]
  2%|▏         | 78/4722 [50:52<52:53:37, 41.00s/it]
  2%|▏         | 79/4722 [51:32<52:21:02, 40.59s/it]
  2%|▏         | 80/4722 [52:13<52:36:20, 40.80s/it]
                                                    
{'loss': 1.4229, 'learning_rate': 1.6701902748414377e-05, 'global_step': 80, 'epoch': 0.05}
+
  2%|▏         | 80/4722 [52:13<52:36:20, 40.80s/it]
  2%|▏         | 81/4722 [52:50<51:02:33, 39.59s/it]
  2%|▏         | 82/4722 [53:23<48:31:03, 37.64s/it]
  2%|▏         | 83/4722 [54:08<51:23:58, 39.89s/it]
  2%|▏         | 84/4722 [54:47<51:01:27, 39.60s/it]
  2%|▏         | 85/4722 [55:25<50:07:53, 38.92s/it]
  2%|▏         | 86/4722 [55:57<47:44:27, 37.07s/it]
  2%|▏         | 87/4722 [56:37<48:38:42, 37.78s/it]
  2%|▏         | 88/4722 [57:16<49:01:50, 38.09s/it]
  2%|▏         | 89/4722 [57:53<48:41:01, 37.83s/it]
  2%|▏         | 90/4722 [58:29<48:10:50, 37.45s/it]
  2%|▏         | 91/4722 [59:12<50:07:30, 38.97s/it]
  2%|▏         | 92/4722 [59:59<53:11:51, 41.36s/it]
  2%|▏         | 93/4722 [1:00:43<54:24:29, 42.31s/it]
  2%|▏         | 94/4722 [1:01:17<51:02:33, 39.70s/it]
  2%|▏         | 95/4722 [1:02:00<52:08:36, 40.57s/it]
  2%|▏         | 96/4722 [1:02:37<50:49:13, 39.55s/it]
  2%|▏         | 97/4722 [1:03:10<48:31:40, 37.77s/it]
  2%|▏         | 98/4722 [1:03:50<49:24:41, 38.47s/it]
  2%|▏         | 99/4722 [1:04:27<48:41:42, 37.92s/it]
  2%|▏         | 100/4722 [1:05:05<48:42:37, 37.94s/it]
                                                       
{'loss': 0.7216, 'learning_rate': 2.0718816067653276e-05, 'global_step': 100, 'epoch': 0.06}
+
  2%|▏         | 100/4722 [1:05:05<48:42:37, 37.94s/it]
  2%|▏         | 101/4722 [1:05:43<48:41:47, 37.94s/it]
  2%|▏         | 102/4722 [1:06:21<48:47:13, 38.02s/it]
  2%|▏         | 103/4722 [1:06:59<48:42:43, 37.97s/it]
  2%|▏         | 104/4722 [1:07:36<48:28:12, 37.79s/it]
  2%|▏         | 105/4722 [1:08:17<49:24:24, 38.52s/it]
  2%|▏         | 106/4722 [1:08:59<50:44:11, 39.57s/it]
  2%|▏         | 107/4722 [1:09:36<49:50:42, 38.88s/it]
  2%|▏         | 108/4722 [1:10:19<51:23:41, 40.10s/it]
  2%|▏         | 109/4722 [1:10:59<51:17:15, 40.02s/it]
  2%|▏         | 110/4722 [1:11:33<49:07:44, 38.35s/it]
  2%|▏         | 111/4722 [1:12:08<47:43:50, 37.27s/it]
  2%|▏         | 112/4722 [1:12:48<48:38:33, 37.99s/it]
  2%|▏         | 113/4722 [1:13:29<49:48:27, 38.90s/it]
  2%|▏         | 114/4722 [1:14:10<50:38:27, 39.56s/it]
  2%|▏         | 115/4722 [1:14:49<50:24:44, 39.39s/it]
  2%|▏         | 116/4722 [1:15:31<51:36:06, 40.33s/it]
  2%|▏         | 117/4722 [1:16:14<52:39:29, 41.17s/it]
  2%|▏         | 118/4722 [1:16:54<52:05:18, 40.73s/it]
  3%|▎         | 119/4722 [1:17:38<53:15:02, 41.65s/it]
  3%|▎         | 120/4722 [1:18:16<51:43:58, 40.47s/it]
                                                       
{'loss': 0.7585, 'learning_rate': 2.473572938689218e-05, 'global_step': 120, 'epoch': 0.08}
+
  3%|▎         | 120/4722 [1:18:16<51:43:58, 40.47s/it]
  3%|▎         | 121/4722 [1:18:53<50:44:09, 39.70s/it]
  3%|▎         | 122/4722 [1:19:34<51:10:55, 40.06s/it]
  3%|▎         | 123/4722 [1:20:17<52:07:42, 40.81s/it]
  3%|▎         | 124/4722 [1:20:56<51:20:25, 40.20s/it]
  3%|▎         | 125/4722 [1:21:34<50:27:49, 39.52s/it]
  3%|▎         | 126/4722 [1:22:10<49:11:48, 38.54s/it]
  3%|▎         | 127/4722 [1:22:51<50:10:26, 39.31s/it]
  3%|▎         | 128/4722 [1:23:30<50:13:49, 39.36s/it]
  3%|▎         | 129/4722 [1:24:07<49:18:38, 38.65s/it]
  3%|▎         | 130/4722 [1:24:41<47:20:57, 37.12s/it]
  3%|▎         | 131/4722 [1:25:20<47:54:41, 37.57s/it]
  3%|▎         | 132/4722 [1:25:55<47:08:34, 36.97s/it]
  3%|▎         | 133/4722 [1:26:35<48:05:58, 37.73s/it]
  3%|▎         | 134/4722 [1:27:19<50:32:19, 39.66s/it]
  3%|▎         | 135/4722 [1:27:58<50:29:09, 39.62s/it]
  3%|▎         | 136/4722 [1:28:36<49:34:05, 38.91s/it]
  3%|▎         | 137/4722 [1:29:18<50:50:35, 39.92s/it]
  3%|▎         | 138/4722 [1:29:57<50:26:55, 39.62s/it]
  3%|▎         | 139/4722 [1:30:34<49:23:14, 38.79s/it]
  3%|▎         | 140/4722 [1:31:17<51:10:09, 40.20s/it]
                                                       
{'loss': 0.7468, 'learning_rate': 2.8964059196617337e-05, 'global_step': 140, 'epoch': 0.09}
+
  3%|▎         | 140/4722 [1:31:17<51:10:09, 40.20s/it]
  3%|▎         | 141/4722 [1:32:05<53:55:00, 42.37s/it]
  3%|▎         | 142/4722 [1:32:40<51:03:58, 40.14s/it]
  3%|▎         | 143/4722 [1:33:15<49:08:13, 38.63s/it]
  3%|▎         | 144/4722 [1:34:00<51:41:07, 40.64s/it]
  3%|▎         | 145/4722 [1:34:38<50:48:32, 39.96s/it]
  3%|▎         | 146/4722 [1:35:21<51:44:50, 40.71s/it]
  3%|▎         | 147/4722 [1:35:58<50:25:55, 39.68s/it]
  3%|▎         | 148/4722 [1:36:38<50:36:24, 39.83s/it]
  3%|▎         | 149/4722 [1:37:14<49:02:50, 38.61s/it]
  3%|▎         | 150/4722 [1:37:55<49:54:37, 39.30s/it]
  3%|▎         | 151/4722 [1:38:32<48:55:33, 38.53s/it]
  3%|▎         | 152/4722 [1:39:12<49:29:58, 38.99s/it]
  3%|▎         | 153/4722 [1:40:03<54:10:00, 42.68s/it]
  3%|▎         | 154/4722 [1:40:40<51:58:44, 40.96s/it]
  3%|▎         | 155/4722 [1:41:18<50:41:49, 39.96s/it]
  3%|▎         | 156/4722 [1:42:02<52:14:20, 41.19s/it]
  3%|▎         | 157/4722 [1:42:44<52:30:27, 41.41s/it]
  3%|▎         | 158/4722 [1:43:22<51:18:20, 40.47s/it]
  3%|▎         | 159/4722 [1:43:59<50:10:04, 39.58s/it]
  3%|▎         | 160/4722 [1:44:38<49:51:21, 39.34s/it]
                                                       
{'loss': 0.8176, 'learning_rate': 3.2346723044397465e-05, 'global_step': 160, 'epoch': 0.1}
+
  3%|▎         | 160/4722 [1:44:38<49:51:21, 39.34s/it]
  3%|▎         | 161/4722 [1:45:26<52:56:22, 41.79s/it]
  3%|▎         | 162/4722 [1:46:08<53:10:40, 41.98s/it]
  3%|▎         | 163/4722 [1:46:46<51:31:35, 40.69s/it]
  3%|▎         | 164/4722 [1:47:31<53:03:32, 41.91s/it]
  3%|▎         | 165/4722 [1:48:16<54:25:26, 42.99s/it]
  4%|▎         | 166/4722 [1:48:56<53:13:19, 42.05s/it]
  4%|▎         | 167/4722 [1:49:34<51:44:23, 40.89s/it]
  4%|▎         | 168/4722 [1:50:13<51:02:35, 40.35s/it]
  4%|▎         | 169/4722 [1:50:51<50:13:03, 39.71s/it]
  4%|▎         | 170/4722 [1:51:34<51:13:03, 40.51s/it]
  4%|▎         | 171/4722 [1:52:12<50:26:32, 39.90s/it]
  4%|▎         | 172/4722 [1:52:51<49:50:17, 39.43s/it]
  4%|▎         | 173/4722 [1:53:26<48:26:23, 38.33s/it]
  4%|▎         | 174/4722 [1:54:08<49:33:53, 39.23s/it]
  4%|▎         | 175/4722 [1:54:45<48:37:16, 38.49s/it]
  4%|▎         | 176/4722 [1:55:22<48:05:59, 38.09s/it]
  4%|▎         | 177/4722 [1:56:04<49:48:37, 39.45s/it]
  4%|▍         | 178/4722 [1:56:37<47:19:44, 37.50s/it]
  4%|▍         | 179/4722 [1:57:15<47:21:54, 37.53s/it]
  4%|▍         | 180/4722 [1:57:55<48:25:21, 38.38s/it]
                                                       
{'loss': 0.7115, 'learning_rate': 3.6575052854122623e-05, 'global_step': 180, 'epoch': 0.11}
+
  4%|▍         | 180/4722 [1:57:55<48:25:21, 38.38s/it]
  4%|▍         | 181/4722 [1:58:34<48:33:30, 38.50s/it]
  4%|▍         | 182/4722 [1:59:12<48:13:48, 38.24s/it]
  4%|▍         | 183/4722 [1:59:54<49:51:21, 39.54s/it]
  4%|▍         | 184/4722 [2:00:32<49:16:56, 39.10s/it]
  4%|▍         | 185/4722 [2:01:15<50:34:11, 40.13s/it]
  4%|▍         | 186/4722 [2:01:48<48:07:45, 38.20s/it]
  4%|▍         | 187/4722 [2:02:26<47:44:09, 37.89s/it]
  4%|▍         | 188/4722 [2:03:02<47:12:25, 37.48s/it]
  4%|▍         | 189/4722 [2:03:45<49:22:17, 39.21s/it]
  4%|▍         | 190/4722 [2:04:29<51:06:49, 40.60s/it]
  4%|▍         | 191/4722 [2:05:00<47:23:28, 37.65s/it]
  4%|▍         | 192/4722 [2:05:49<51:44:40, 41.12s/it]
  4%|▍         | 193/4722 [2:06:29<51:18:12, 40.78s/it]
  4%|▍         | 194/4722 [2:07:05<49:24:02, 39.28s/it]
  4%|▍         | 195/4722 [2:07:41<48:13:33, 38.35s/it]
  4%|▍         | 196/4722 [2:08:21<48:39:56, 38.71s/it]
  4%|▍         | 197/4722 [2:09:05<50:35:46, 40.25s/it]
  4%|▍         | 198/4722 [2:09:39<48:27:57, 38.57s/it]
  4%|▍         | 199/4722 [2:10:14<47:09:34, 37.54s/it]
  4%|▍         | 200/4722 [2:11:03<51:25:08, 40.94s/it]
                                                       
{'loss': 0.742, 'learning_rate': 4.080338266384778e-05, 'global_step': 200, 'epoch': 0.13}
+
  4%|▍         | 200/4722 [2:11:03<51:25:08, 40.94s/it]
  4%|▍         | 201/4722 [2:11:37<48:39:59, 38.75s/it]
  4%|▍         | 202/4722 [2:12:21<50:33:04, 40.26s/it]
  4%|▍         | 203/4722 [2:13:03<51:29:30, 41.02s/it]
  4%|▍         | 204/4722 [2:13:46<51:53:30, 41.35s/it]
  4%|▍         | 205/4722 [2:14:22<50:03:15, 39.89s/it]
  4%|▍         | 206/4722 [2:15:00<49:24:46, 39.39s/it]
  4%|▍         | 207/4722 [2:15:33<46:42:59, 37.25s/it]
  4%|▍         | 208/4722 [2:16:04<44:32:05, 35.52s/it]
  4%|▍         | 209/4722 [2:16:40<44:33:15, 35.54s/it]
  4%|▍         | 210/4722 [2:17:36<52:14:02, 41.68s/it]
  4%|▍         | 211/4722 [2:18:19<52:47:32, 42.13s/it]
  4%|▍         | 212/4722 [2:18:56<51:06:54, 40.80s/it]
  5%|▍         | 213/4722 [2:19:40<52:08:00, 41.62s/it]
  5%|▍         | 214/4722 [2:20:22<52:11:38, 41.68s/it]
  5%|▍         | 215/4722 [2:21:06<53:00:32, 42.34s/it]
  5%|▍         | 216/4722 [2:21:45<51:44:20, 41.34s/it]
  5%|▍         | 217/4722 [2:22:23<50:24:26, 40.28s/it]
  5%|▍         | 218/4722 [2:23:04<50:41:43, 40.52s/it]
  5%|▍         | 219/4722 [2:23:47<51:52:46, 41.48s/it]
  5%|▍         | 220/4722 [2:24:30<52:07:59, 41.69s/it]
                                                       
{'loss': 0.7049, 'learning_rate': 4.503171247357294e-05, 'global_step': 220, 'epoch': 0.14}
+
  5%|▍         | 220/4722 [2:24:30<52:07:59, 41.69s/it]
  5%|▍         | 221/4722 [2:25:05<49:44:44, 39.79s/it]
  5%|▍         | 222/4722 [2:25:41<48:20:47, 38.68s/it]
  5%|▍         | 223/4722 [2:26:24<49:57:25, 39.97s/it]
  5%|▍         | 224/4722 [2:27:00<48:21:01, 38.70s/it]
  5%|▍         | 225/4722 [2:27:40<49:02:52, 39.26s/it]
  5%|▍         | 226/4722 [2:28:16<47:46:14, 38.25s/it]
  5%|▍         | 227/4722 [2:29:03<51:08:34, 40.96s/it]
  5%|▍         | 228/4722 [2:29:43<50:39:13, 40.58s/it]
  5%|▍         | 229/4722 [2:30:23<50:15:14, 40.27s/it]
  5%|▍         | 230/4722 [2:31:06<51:31:05, 41.29s/it]
  5%|▍         | 231/4722 [2:31:44<50:18:11, 40.32s/it]
  5%|▍         | 232/4722 [2:32:34<53:41:02, 43.04s/it]
  5%|▍         | 233/4722 [2:33:11<51:36:44, 41.39s/it]
  5%|▍         | 234/4722 [2:33:45<48:36:56, 39.00s/it]
  5%|▍         | 235/4722 [2:34:24<48:46:18, 39.13s/it]
  5%|▍         | 236/4722 [2:35:05<49:22:05, 39.62s/it]
  5%|▌         | 237/4722 [2:35:37<46:42:53, 37.50s/it]
  5%|▌         | 238/4722 [2:36:12<45:26:06, 36.48s/it]
  5%|▌         | 239/4722 [2:36:52<47:04:57, 37.81s/it]
  5%|▌         | 240/4722 [2:37:29<46:25:34, 37.29s/it]
                                                       
{'loss': 0.7075, 'learning_rate': 4.9260042283298105e-05, 'global_step': 240, 'epoch': 0.15}
+
  5%|▌         | 240/4722 [2:37:29<46:25:34, 37.29s/it]
  5%|▌         | 241/4722 [2:38:20<51:51:25, 41.66s/it]
  5%|▌         | 242/4722 [2:39:02<51:39:02, 41.50s/it]
  5%|▌         | 243/4722 [2:39:37<49:23:03, 39.69s/it]
  5%|▌         | 244/4722 [2:40:14<48:31:00, 39.00s/it]
  5%|▌         | 245/4722 [2:40:57<49:45:25, 40.01s/it]
  5%|▌         | 246/4722 [2:41:40<51:06:44, 41.11s/it]
  5%|▌         | 247/4722 [2:42:13<47:43:49, 38.40s/it]
  5%|▌         | 248/4722 [2:42:54<48:53:36, 39.34s/it]
  5%|▌         | 249/4722 [2:43:39<51:03:51, 41.10s/it]
  5%|▌         | 250/4722 [2:44:15<48:55:22, 39.38s/it]
  5%|▌         | 251/4722 [2:44:58<50:19:28, 40.52s/it]
  5%|▌         | 252/4722 [2:45:47<53:26:41, 43.04s/it]
  5%|▌         | 253/4722 [2:46:22<50:35:19, 40.75s/it]
  5%|▌         | 254/4722 [2:46:55<47:37:40, 38.38s/it]
  5%|▌         | 255/4722 [2:47:35<48:23:12, 39.00s/it]
  5%|▌         | 256/4722 [2:48:14<48:10:24, 38.83s/it]
  5%|▌         | 257/4722 [2:48:54<48:46:46, 39.33s/it]
  5%|▌         | 258/4722 [2:49:34<49:00:50, 39.53s/it]
  5%|▌         | 259/4722 [2:50:07<46:17:32, 37.34s/it]
  6%|▌         | 260/4722 [2:50:52<49:17:08, 39.76s/it]
                                                       
{'loss': 0.7125, 'learning_rate': 5.348837209302326e-05, 'global_step': 260, 'epoch': 0.17}
+
  6%|▌         | 260/4722 [2:50:52<49:17:08, 39.76s/it]
  6%|▌         | 261/4722 [2:51:30<48:37:04, 39.23s/it]
  6%|▌         | 262/4722 [2:52:06<47:28:34, 38.32s/it]
  6%|▌         | 263/4722 [2:52:40<45:53:55, 37.06s/it]
  6%|▌         | 264/4722 [2:53:19<46:25:33, 37.49s/it]
  6%|▌         | 265/4722 [2:53:53<45:20:08, 36.62s/it]
  6%|▌         | 266/4722 [2:54:40<49:07:46, 39.69s/it]
  6%|▌         | 267/4722 [2:55:18<48:18:27, 39.04s/it]
  6%|▌         | 268/4722 [2:56:08<52:24:19, 42.36s/it]
  6%|▌         | 269/4722 [2:56:46<50:38:59, 40.95s/it]
  6%|▌         | 270/4722 [2:57:25<49:59:22, 40.42s/it]
  6%|▌         | 271/4722 [2:58:01<48:28:57, 39.21s/it]
  6%|▌         | 272/4722 [2:58:33<45:51:17, 37.10s/it]
  6%|▌         | 273/4722 [2:59:13<46:52:23, 37.93s/it]
  6%|▌         | 274/4722 [2:59:46<44:57:51, 36.39s/it]
  6%|▌         | 275/4722 [3:00:23<45:17:20, 36.66s/it]
  6%|▌         | 276/4722 [3:01:04<46:36:42, 37.74s/it]
  6%|▌         | 277/4722 [3:01:45<48:02:18, 38.91s/it]
  6%|▌         | 278/4722 [3:02:25<48:25:39, 39.23s/it]
  6%|▌         | 279/4722 [3:03:07<49:30:50, 40.12s/it]
  6%|▌         | 280/4722 [3:03:43<47:53:50, 38.82s/it]
                                                       
{'loss': 0.7052, 'learning_rate': 5.7716701902748415e-05, 'global_step': 280, 'epoch': 0.18}
+
  6%|▌         | 280/4722 [3:03:43<47:53:50, 38.82s/it]
  6%|▌         | 281/4722 [3:04:21<47:40:08, 38.64s/it]
  6%|▌         | 282/4722 [3:04:54<45:25:54, 36.84s/it]
  6%|▌         | 283/4722 [3:05:35<47:00:34, 38.12s/it]
  6%|▌         | 284/4722 [3:06:11<46:18:27, 37.56s/it]
  6%|▌         | 285/4722 [3:07:00<50:24:08, 40.89s/it]
  6%|▌         | 286/4722 [3:07:38<49:24:45, 40.10s/it]
  6%|▌         | 287/4722 [3:08:17<48:58:20, 39.75s/it]
  6%|▌         | 288/4722 [3:08:55<48:06:45, 39.06s/it]
  6%|▌         | 289/4722 [3:09:30<46:44:51, 37.96s/it]
  6%|▌         | 290/4722 [3:10:07<46:12:33, 37.53s/it]
  6%|▌         | 291/4722 [3:10:42<45:18:07, 36.81s/it]
  6%|▌         | 292/4722 [3:11:20<45:44:14, 37.17s/it]
  6%|▌         | 293/4722 [3:11:56<45:30:45, 36.99s/it]
  6%|▌         | 294/4722 [3:12:35<46:10:54, 37.55s/it]
  6%|▌         | 295/4722 [3:13:13<46:25:41, 37.75s/it]
  6%|▋         | 296/4722 [3:13:53<47:16:12, 38.45s/it]
  6%|▋         | 297/4722 [3:14:36<48:41:32, 39.61s/it]
  6%|▋         | 298/4722 [3:15:21<50:35:22, 41.17s/it]
  6%|▋         | 299/4722 [3:16:07<52:34:02, 42.79s/it]
  6%|▋         | 300/4722 [3:16:46<51:14:37, 41.72s/it]
                                                       
{'loss': 0.7236, 'learning_rate': 6.194503171247358e-05, 'global_step': 300, 'epoch': 0.19}
+
  6%|▋         | 300/4722 [3:16:46<51:14:37, 41.72s/it]
  6%|▋         | 301/4722 [3:17:27<50:58:20, 41.51s/it]
  6%|▋         | 302/4722 [3:18:02<48:20:10, 39.37s/it]
  6%|▋         | 303/4722 [3:18:34<45:41:39, 37.23s/it]
  6%|▋         | 304/4722 [3:19:12<45:56:06, 37.43s/it]
  6%|▋         | 305/4722 [3:19:51<46:26:44, 37.85s/it]
  6%|▋         | 306/4722 [3:20:27<45:46:38, 37.32s/it]
  7%|▋         | 307/4722 [3:21:07<46:59:47, 38.32s/it]
  7%|▋         | 308/4722 [3:21:49<48:12:32, 39.32s/it]
  7%|▋         | 309/4722 [3:22:21<45:23:33, 37.03s/it]
  7%|▋         | 310/4722 [3:22:57<44:57:46, 36.69s/it]
  7%|▋         | 311/4722 [3:23:37<46:14:50, 37.74s/it]
  7%|▋         | 312/4722 [3:24:12<45:06:14, 36.82s/it]
  7%|▋         | 313/4722 [3:24:43<43:04:35, 35.17s/it]
  7%|▋         | 314/4722 [3:25:21<44:02:56, 35.97s/it]
  7%|▋         | 315/4722 [3:26:00<45:17:40, 37.00s/it]
  7%|▋         | 316/4722 [3:26:38<45:36:41, 37.27s/it]
  7%|▋         | 317/4722 [3:27:17<46:11:39, 37.75s/it]
  7%|▋         | 318/4722 [3:27:57<46:58:03, 38.39s/it]
  7%|▋         | 319/4722 [3:28:37<47:44:46, 39.04s/it]
  7%|▋         | 320/4722 [3:29:14<47:01:14, 38.45s/it]
                                                       
{'loss': 0.7166, 'learning_rate': 6.617336152219874e-05, 'global_step': 320, 'epoch': 0.2}
+
  7%|▋         | 320/4722 [3:29:14<47:01:14, 38.45s/it]
  7%|▋         | 321/4722 [3:29:58<48:44:14, 39.87s/it]
  7%|▋         | 322/4722 [3:30:36<48:15:44, 39.49s/it]
  7%|▋         | 323/4722 [3:31:14<47:42:57, 39.05s/it]
  7%|▋         | 324/4722 [3:31:48<45:57:10, 37.61s/it]
  7%|▋         | 325/4722 [3:32:35<49:04:46, 40.18s/it]
  7%|▋         | 326/4722 [3:33:09<47:01:24, 38.51s/it]
  7%|▋         | 327/4722 [3:33:47<46:54:43, 38.43s/it]
  7%|▋         | 328/4722 [3:34:29<48:04:50, 39.39s/it]
  7%|▋         | 329/4722 [3:35:03<46:08:29, 37.81s/it]
  7%|▋         | 330/4722 [3:35:37<44:29:47, 36.47s/it]
  7%|▋         | 331/4722 [3:36:20<47:00:10, 38.54s/it]
  7%|▋         | 332/4722 [3:36:58<46:48:48, 38.39s/it]
  7%|▋         | 333/4722 [3:37:43<49:03:38, 40.24s/it]
  7%|▋         | 334/4722 [3:38:25<50:01:05, 41.04s/it]
  7%|▋         | 335/4722 [3:39:04<49:04:01, 40.26s/it]
  7%|▋         | 336/4722 [3:39:48<50:33:22, 41.50s/it]
  7%|▋         | 337/4722 [3:40:36<52:40:41, 43.25s/it]
  7%|▋         | 338/4722 [3:41:13<50:32:08, 41.50s/it]
  7%|▋         | 339/4722 [3:41:48<48:00:31, 39.43s/it]
  7%|▋         | 340/4722 [3:42:27<47:49:51, 39.30s/it]
                                                       
{'loss': 0.7356, 'learning_rate': 7.04016913319239e-05, 'global_step': 340, 'epoch': 0.22}
+
  7%|▋         | 340/4722 [3:42:27<47:49:51, 39.30s/it]
  7%|▋         | 341/4722 [3:42:59<45:16:58, 37.21s/it]
  7%|▋         | 342/4722 [3:43:36<45:10:40, 37.13s/it]
  7%|▋         | 343/4722 [3:44:12<44:43:45, 36.77s/it]
  7%|▋         | 344/4722 [3:44:48<44:29:48, 36.59s/it]
  7%|▋         | 345/4722 [3:45:23<43:53:46, 36.10s/it]
  7%|▋         | 346/4722 [3:46:01<44:44:15, 36.80s/it]
  7%|▋         | 347/4722 [3:46:38<44:38:28, 36.73s/it]
  7%|▋         | 348/4722 [3:47:20<46:26:49, 38.23s/it]
  7%|▋         | 349/4722 [3:48:01<47:34:02, 39.16s/it]
  7%|▋         | 350/4722 [3:48:41<47:50:31, 39.39s/it]
  7%|▋         | 351/4722 [3:49:21<48:12:25, 39.70s/it]
  7%|▋         | 352/4722 [3:49:59<47:25:48, 39.07s/it]
  7%|▋         | 353/4722 [3:50:34<45:55:27, 37.84s/it]
  7%|▋         | 354/4722 [3:51:17<47:55:37, 39.50s/it]
  8%|▊         | 355/4722 [3:51:50<45:25:31, 37.45s/it]
  8%|▊         | 356/4722 [3:52:28<45:33:27, 37.56s/it]
  8%|▊         | 357/4722 [3:53:02<44:12:21, 36.46s/it]
  8%|▊         | 358/4722 [3:53:37<43:48:07, 36.13s/it]
  8%|▊         | 359/4722 [3:54:11<43:04:58, 35.55s/it]
  8%|▊         | 360/4722 [3:54:53<45:26:27, 37.50s/it]
                                                       
{'loss': 0.766, 'learning_rate': 7.463002114164905e-05, 'global_step': 360, 'epoch': 0.23}
+
  8%|▊         | 360/4722 [3:54:53<45:26:27, 37.50s/it]
  8%|▊         | 361/4722 [3:55:31<45:20:55, 37.44s/it]
  8%|▊         | 362/4722 [3:56:06<44:34:32, 36.81s/it]
  8%|▊         | 363/4722 [3:56:51<47:25:43, 39.17s/it]
  8%|▊         | 364/4722 [3:57:30<47:38:51, 39.36s/it]
  8%|▊         | 365/4722 [3:58:03<45:20:09, 37.46s/it]
  8%|▊         | 366/4722 [3:58:43<46:05:38, 38.09s/it]
  8%|▊         | 367/4722 [3:59:22<46:13:36, 38.21s/it]
  8%|▊         | 368/4722 [3:59:58<45:38:16, 37.73s/it]
  8%|▊         | 369/4722 [4:00:41<47:25:30, 39.22s/it]
  8%|▊         | 370/4722 [4:01:16<46:00:06, 38.05s/it]
  8%|▊         | 371/4722 [4:01:54<45:49:20, 37.91s/it]
  8%|▊         | 372/4722 [4:02:30<45:15:55, 37.46s/it]
  8%|▊         | 373/4722 [4:03:14<47:39:58, 39.46s/it]
  8%|▊         | 374/4722 [4:03:49<45:51:12, 37.97s/it]
  8%|▊         | 375/4722 [4:04:29<46:38:42, 38.63s/it]
  8%|▊         | 376/4722 [4:05:06<46:06:56, 38.20s/it]
  8%|▊         | 377/4722 [4:05:43<45:35:51, 37.78s/it]
  8%|▊         | 378/4722 [4:06:18<44:42:56, 37.06s/it]
  8%|▊         | 379/4722 [4:06:56<44:57:23, 37.27s/it]
  8%|▊         | 380/4722 [4:07:35<45:26:24, 37.67s/it]
                                                       
{'loss': 0.7219, 'learning_rate': 7.885835095137421e-05, 'global_step': 380, 'epoch': 0.24}
+
  8%|▊         | 380/4722 [4:07:35<45:26:24, 37.67s/it]
  8%|▊         | 381/4722 [4:08:20<48:01:39, 39.83s/it]
  8%|▊         | 382/4722 [4:08:58<47:28:15, 39.38s/it]
  8%|▊         | 383/4722 [4:09:35<46:39:23, 38.71s/it]
  8%|▊         | 384/4722 [4:10:15<47:00:25, 39.01s/it]
  8%|▊         | 385/4722 [4:10:51<46:07:15, 38.28s/it]
  8%|▊         | 386/4722 [4:11:35<48:09:00, 39.98s/it]
  8%|▊         | 387/4722 [4:12:27<52:29:13, 43.59s/it]
  8%|▊         | 388/4722 [4:13:07<50:55:31, 42.30s/it]
  8%|▊         | 389/4722 [4:13:48<50:30:16, 41.96s/it]
  8%|▊         | 390/4722 [4:14:32<51:19:41, 42.65s/it]
  8%|▊         | 391/4722 [4:15:11<49:52:06, 41.45s/it]
  8%|▊         | 392/4722 [4:15:53<50:17:00, 41.81s/it]
  8%|▊         | 393/4722 [4:16:32<49:17:08, 40.99s/it]
  8%|▊         | 394/4722 [4:17:13<49:16:49, 40.99s/it]
  8%|▊         | 395/4722 [4:17:47<46:43:55, 38.88s/it]
  8%|▊         | 396/4722 [4:18:28<47:18:11, 39.36s/it]
  8%|▊         | 397/4722 [4:19:09<47:55:50, 39.90s/it]
  8%|▊         | 398/4722 [4:19:54<49:50:24, 41.50s/it]
  8%|▊         | 399/4722 [4:20:32<48:30:35, 40.40s/it]
  8%|▊         | 400/4722 [4:21:09<47:08:58, 39.27s/it]
                                                       
{'loss': 0.7014, 'learning_rate': 8.308668076109937e-05, 'global_step': 400, 'epoch': 0.25}
+
  8%|▊         | 400/4722 [4:21:09<47:08:58, 39.27s/it]
  8%|▊         | 401/4722 [4:21:48<47:06:55, 39.25s/it]
  9%|▊         | 402/4722 [4:22:22<45:12:32, 37.67s/it]
  9%|▊         | 403/4722 [4:22:58<44:43:54, 37.29s/it]
  9%|▊         | 404/4722 [4:23:36<44:46:10, 37.33s/it]
  9%|▊         | 405/4722 [4:24:09<43:12:10, 36.03s/it]
  9%|▊         | 406/4722 [4:24:50<44:55:42, 37.48s/it]
  9%|▊         | 407/4722 [4:25:28<45:24:32, 37.88s/it]
  9%|▊         | 408/4722 [4:26:08<45:59:38, 38.38s/it]
  9%|▊         | 409/4722 [4:26:47<46:18:30, 38.65s/it]
  9%|▊         | 410/4722 [4:27:26<46:20:03, 38.68s/it]
  9%|▊         | 411/4722 [4:28:05<46:26:50, 38.79s/it]
  9%|▊         | 412/4722 [4:28:46<47:06:52, 39.35s/it]
  9%|▊         | 413/4722 [4:29:27<48:00:10, 40.10s/it]
  9%|▉         | 414/4722 [4:30:09<48:30:55, 40.54s/it]
  9%|▉         | 415/4722 [4:30:49<48:10:45, 40.27s/it]
  9%|▉         | 416/4722 [4:31:27<47:36:53, 39.81s/it]
  9%|▉         | 417/4722 [4:32:04<46:36:34, 38.98s/it]
  9%|▉         | 418/4722 [4:32:42<46:02:09, 38.51s/it]
  9%|▉         | 419/4722 [4:33:31<49:52:42, 41.73s/it]
  9%|▉         | 420/4722 [4:34:07<47:40:01, 39.89s/it]
                                                       
{'loss': 0.7175, 'learning_rate': 8.731501057082453e-05, 'global_step': 420, 'epoch': 0.27}
+
  9%|▉         | 420/4722 [4:34:07<47:40:01, 39.89s/it]
  9%|▉         | 421/4722 [4:34:36<43:46:30, 36.64s/it]
  9%|▉         | 422/4722 [4:35:13<44:06:46, 36.93s/it]
  9%|▉         | 423/4722 [4:36:01<48:01:13, 40.21s/it]
  9%|▉         | 424/4722 [4:36:34<45:21:52, 38.00s/it]
  9%|▉         | 425/4722 [4:37:05<42:39:04, 35.73s/it]
  9%|▉         | 426/4722 [4:37:43<43:34:08, 36.51s/it]
  9%|▉         | 427/4722 [4:38:22<44:37:04, 37.40s/it]
  9%|▉         | 428/4722 [4:39:00<44:50:14, 37.59s/it]
  9%|▉         | 429/4722 [4:39:37<44:25:42, 37.26s/it]
  9%|▉         | 430/4722 [4:40:13<44:08:18, 37.02s/it]
  9%|▉         | 431/4722 [4:40:51<44:26:36, 37.29s/it]
  9%|▉         | 432/4722 [4:41:27<43:51:17, 36.80s/it]
  9%|▉         | 433/4722 [4:42:05<44:14:21, 37.13s/it]
  9%|▉         | 434/4722 [4:42:42<44:23:47, 37.27s/it]
  9%|▉         | 435/4722 [4:43:20<44:30:27, 37.38s/it]
  9%|▉         | 436/4722 [4:43:53<42:45:54, 35.92s/it]
  9%|▉         | 437/4722 [4:44:30<43:10:05, 36.27s/it]
  9%|▉         | 438/4722 [4:45:10<44:32:23, 37.43s/it]
  9%|▉         | 439/4722 [4:45:54<46:57:46, 39.47s/it]
  9%|▉         | 440/4722 [4:46:27<44:28:06, 37.39s/it]
                                                       
{'loss': 0.7089, 'learning_rate': 9.154334038054969e-05, 'global_step': 440, 'epoch': 0.28}
+
  9%|▉         | 440/4722 [4:46:27<44:28:06, 37.39s/it]
  9%|▉         | 441/4722 [4:47:04<44:31:11, 37.44s/it]
  9%|▉         | 442/4722 [4:47:48<46:51:26, 39.41s/it]
  9%|▉         | 443/4722 [4:48:31<48:12:28, 40.56s/it]
  9%|▉         | 444/4722 [4:49:16<49:38:26, 41.77s/it]
  9%|▉         | 445/4722 [4:49:54<48:14:25, 40.60s/it]
  9%|▉         | 446/4722 [4:50:33<47:47:51, 40.24s/it]
  9%|▉         | 447/4722 [4:51:18<49:22:01, 41.57s/it]
  9%|▉         | 448/4722 [4:51:56<48:04:30, 40.49s/it]
 10%|▉         | 449/4722 [4:52:33<46:50:16, 39.46s/it]
 10%|▉         | 450/4722 [4:53:10<46:01:07, 38.78s/it]
 10%|▉         | 451/4722 [4:53:45<44:33:42, 37.56s/it]
 10%|▉         | 452/4722 [4:54:20<43:40:20, 36.82s/it]
 10%|▉         | 453/4722 [4:54:59<44:30:57, 37.54s/it]
 10%|▉         | 454/4722 [4:55:38<44:54:22, 37.88s/it]
 10%|▉         | 455/4722 [4:56:22<47:11:06, 39.81s/it]
 10%|▉         | 456/4722 [4:57:08<49:10:43, 41.50s/it]
 10%|▉         | 457/4722 [4:57:45<47:51:03, 40.39s/it]
 10%|▉         | 458/4722 [4:58:18<45:12:04, 38.16s/it]
 10%|▉         | 459/4722 [4:59:03<47:24:19, 40.03s/it]
 10%|▉         | 460/4722 [4:59:40<46:28:27, 39.26s/it]
                                                       
{'loss': 0.7039, 'learning_rate': 9.577167019027485e-05, 'global_step': 460, 'epoch': 0.29}
+
 10%|▉         | 460/4722 [4:59:40<46:28:27, 39.26s/it]
 10%|▉         | 461/4722 [5:00:13<44:20:11, 37.46s/it]
 10%|▉         | 462/4722 [5:00:51<44:26:57, 37.56s/it]
 10%|▉         | 463/4722 [5:01:27<43:52:47, 37.09s/it]
 10%|▉         | 464/4722 [5:02:06<44:37:29, 37.73s/it]
 10%|▉         | 465/4722 [5:02:45<44:48:48, 37.90s/it]
 10%|▉         | 466/4722 [5:03:24<45:27:45, 38.46s/it]
 10%|▉         | 467/4722 [5:04:07<46:52:50, 39.66s/it]
 10%|▉         | 468/4722 [5:04:39<44:05:15, 37.31s/it]
 10%|▉         | 469/4722 [5:05:21<45:51:23, 38.82s/it]
 10%|▉         | 470/4722 [5:06:04<47:08:04, 39.91s/it]
 10%|▉         | 471/4722 [5:06:43<46:53:43, 39.71s/it]
 10%|▉         | 472/4722 [5:07:26<48:08:10, 40.77s/it]
 10%|█         | 473/4722 [5:08:04<47:10:24, 39.97s/it]
 10%|█         | 474/4722 [5:08:40<45:42:10, 38.73s/it]
 10%|█         | 475/4722 [5:09:27<48:32:38, 41.15s/it]
 10%|█         | 476/4722 [5:10:09<49:02:57, 41.59s/it]
 10%|█         | 477/4722 [5:10:51<49:01:25, 41.57s/it]
 10%|█         | 478/4722 [5:11:33<49:12:48, 41.75s/it]
 10%|█         | 479/4722 [5:12:06<46:13:44, 39.22s/it]
 10%|█         | 480/4722 [5:12:44<45:28:50, 38.60s/it]
                                                       
{'loss': 0.702, 'learning_rate': 0.0001, 'global_step': 480, 'epoch': 0.3}
+
 10%|█         | 480/4722 [5:12:44<45:28:50, 38.60s/it]
 10%|█         | 481/4722 [5:13:16<43:07:03, 36.60s/it]
 10%|█         | 482/4722 [5:13:50<42:32:09, 36.12s/it]
 10%|█         | 483/4722 [5:14:31<44:07:30, 37.47s/it]
 10%|█         | 484/4722 [5:15:11<44:55:49, 38.17s/it]
 10%|█         | 485/4722 [5:15:49<44:53:11, 38.14s/it]
 10%|█         | 486/4722 [5:16:25<44:13:56, 37.59s/it]
 10%|█         | 487/4722 [5:17:11<47:07:58, 40.07s/it]
 10%|█         | 488/4722 [5:17:53<47:41:06, 40.54s/it]
 10%|█         | 489/4722 [5:18:29<46:02:08, 39.15s/it]
 10%|█         | 490/4722 [5:19:07<45:44:01, 38.90s/it]
 10%|█         | 491/4722 [5:19:42<44:18:01, 37.69s/it]
 10%|█         | 492/4722 [5:20:22<45:05:25, 38.37s/it]
 10%|█         | 493/4722 [5:20:58<44:25:21, 37.82s/it]
 10%|█         | 494/4722 [5:21:42<46:28:14, 39.57s/it]
 10%|█         | 495/4722 [5:22:26<47:52:44, 40.78s/it]
 11%|█         | 496/4722 [5:23:04<46:53:53, 39.95s/it]
 11%|█         | 497/4722 [5:23:40<45:44:18, 38.97s/it]
 11%|█         | 498/4722 [5:24:15<44:03:26, 37.55s/it]
 11%|█         | 499/4722 [5:24:55<45:13:18, 38.55s/it]
 11%|█         | 500/4722 [5:25:35<45:25:36, 38.73s/it]
                                                       
{'loss': 0.7014, 'learning_rate': 0.0001, 'global_step': 500, 'epoch': 0.32}
+
 11%|█         | 500/4722 [5:25:35<45:25:36, 38.73s/it]
 11%|█         | 501/4722 [5:26:23<48:48:32, 41.63s/it]
 11%|█         | 502/4722 [5:27:02<47:46:56, 40.76s/it]
 11%|█         | 503/4722 [5:27:47<49:13:27, 42.00s/it]
 11%|█         | 504/4722 [5:28:29<49:25:33, 42.18s/it]
 11%|█         | 505/4722 [5:29:10<48:59:24, 41.82s/it]
 11%|█         | 506/4722 [5:29:44<46:08:47, 39.40s/it]
 11%|█         | 507/4722 [5:30:18<44:22:06, 37.89s/it]
 11%|█         | 508/4722 [5:30:49<41:40:22, 35.60s/it]
 11%|█         | 509/4722 [5:31:22<40:56:27, 34.98s/it]
 11%|█         | 510/4722 [5:32:03<42:56:14, 36.70s/it]
 11%|█         | 511/4722 [5:32:51<46:52:40, 40.08s/it]
 11%|█         | 512/4722 [5:33:28<46:00:44, 39.35s/it]
 11%|█         | 513/4722 [5:34:01<43:45:45, 37.43s/it]
 11%|█         | 514/4722 [5:34:41<44:31:49, 38.10s/it]
 11%|█         | 515/4722 [5:35:18<44:12:49, 37.83s/it]
 11%|█         | 516/4722 [5:36:00<45:32:57, 38.99s/it]
 11%|█         | 517/4722 [5:36:33<43:23:47, 37.15s/it]
 11%|█         | 518/4722 [5:37:08<42:37:51, 36.51s/it]
 11%|█         | 519/4722 [5:37:46<43:08:40, 36.95s/it]
 11%|█         | 520/4722 [5:38:34<46:59:47, 40.26s/it]
                                                       
{'loss': 0.6886, 'learning_rate': 0.0001, 'global_step': 520, 'epoch': 0.33}
+
 11%|█         | 520/4722 [5:38:34<46:59:47, 40.26s/it]
 11%|█         | 521/4722 [5:39:13<46:38:10, 39.96s/it]
 11%|█         | 522/4722 [5:39:47<44:35:53, 38.23s/it]
 11%|█         | 523/4722 [5:40:22<43:17:07, 37.11s/it]
 11%|█         | 524/4722 [5:41:00<43:47:42, 37.56s/it]
 11%|█         | 525/4722 [5:41:36<42:56:35, 36.83s/it]
 11%|█         | 526/4722 [5:42:16<44:14:37, 37.96s/it]
 11%|█         | 527/4722 [5:42:51<43:10:14, 37.05s/it]
 11%|█         | 528/4722 [5:43:36<45:51:23, 39.36s/it]
 11%|█         | 529/4722 [5:44:14<45:34:26, 39.13s/it]
 11%|█         | 530/4722 [5:44:55<45:55:46, 39.44s/it]
 11%|█         | 531/4722 [5:45:28<43:41:59, 37.54s/it]
 11%|█▏        | 532/4722 [5:46:07<44:29:26, 38.23s/it]
 11%|█▏        | 533/4722 [5:46:45<44:04:55, 37.88s/it]
 11%|█▏        | 534/4722 [5:47:21<43:42:22, 37.57s/it]
 11%|█▏        | 535/4722 [5:48:01<44:27:16, 38.22s/it]
 11%|█▏        | 536/4722 [5:48:40<44:44:54, 38.48s/it]
 11%|█▏        | 537/4722 [5:49:23<46:15:48, 39.80s/it]
 11%|█▏        | 538/4722 [5:49:58<44:26:21, 38.24s/it]
 11%|█▏        | 539/4722 [5:50:42<46:33:34, 40.07s/it]
 11%|█▏        | 540/4722 [5:51:28<48:26:48, 41.70s/it]
                                                       
{'loss': 0.7174, 'learning_rate': 0.0001, 'global_step': 540, 'epoch': 0.34}
+
 11%|█▏        | 540/4722 [5:51:28<48:26:48, 41.70s/it]
 11%|█▏        | 541/4722 [5:52:13<49:46:29, 42.86s/it]
 11%|█▏        | 542/4722 [5:52:55<49:16:15, 42.43s/it]
 11%|█▏        | 543/4722 [5:53:39<49:57:04, 43.03s/it]
 12%|█▏        | 544/4722 [5:54:19<48:48:51, 42.06s/it]
 12%|█▏        | 545/4722 [5:55:00<48:35:18, 41.88s/it]
 12%|█▏        | 546/4722 [5:55:41<48:15:01, 41.60s/it]
 12%|█▏        | 547/4722 [5:56:18<46:36:48, 40.19s/it]
 12%|█▏        | 548/4722 [5:56:55<45:24:58, 39.17s/it]
 12%|█▏        | 549/4722 [5:57:35<45:41:34, 39.42s/it]
 12%|█▏        | 550/4722 [5:58:12<44:46:55, 38.64s/it]
 12%|█▏        | 551/4722 [5:58:46<43:09:16, 37.25s/it]
 12%|█▏        | 552/4722 [5:59:24<43:25:15, 37.49s/it]
 12%|█▏        | 553/4722 [6:00:03<44:08:24, 38.12s/it]
 12%|█▏        | 554/4722 [6:00:47<46:00:09, 39.73s/it]
 12%|█▏        | 555/4722 [6:01:27<46:06:34, 39.84s/it]
 12%|█▏        | 556/4722 [6:02:06<45:48:44, 39.59s/it]
 12%|█▏        | 557/4722 [6:02:43<44:50:16, 38.76s/it]
 12%|█▏        | 558/4722 [6:03:21<44:42:37, 38.65s/it]
 12%|█▏        | 559/4722 [6:04:01<45:07:23, 39.02s/it]
 12%|█▏        | 560/4722 [6:04:44<46:37:49, 40.33s/it]
                                                       
{'loss': 0.7239, 'learning_rate': 0.0001, 'global_step': 560, 'epoch': 0.36}
+
 12%|█▏        | 560/4722 [6:04:44<46:37:49, 40.33s/it]
 12%|█▏        | 561/4722 [6:05:26<47:04:38, 40.73s/it]
 12%|█▏        | 562/4722 [6:06:12<48:59:21, 42.39s/it]
 12%|█▏        | 563/4722 [6:06:51<47:34:45, 41.18s/it]
 12%|█▏        | 564/4722 [6:07:35<48:35:19, 42.07s/it]
 12%|█▏        | 565/4722 [6:08:12<46:52:23, 40.59s/it]
 12%|█▏        | 566/4722 [6:08:44<43:58:15, 38.09s/it]
 12%|█▏        | 567/4722 [6:09:28<45:54:46, 39.78s/it]
 12%|█▏        | 568/4722 [6:10:16<48:49:11, 42.31s/it]
 12%|█▏        | 569/4722 [6:11:00<49:23:17, 42.81s/it]
 12%|█▏        | 570/4722 [6:11:40<48:17:35, 41.87s/it]
 12%|█▏        | 571/4722 [6:12:17<46:36:59, 40.43s/it]
 12%|█▏        | 572/4722 [6:13:01<48:03:05, 41.68s/it]
 12%|█▏        | 573/4722 [6:13:38<46:24:30, 40.27s/it]
 12%|█▏        | 574/4722 [6:14:21<47:01:04, 40.81s/it]
 12%|█▏        | 575/4722 [6:14:57<45:38:44, 39.62s/it]
 12%|█▏        | 576/4722 [6:15:36<45:18:18, 39.34s/it]
 12%|█▏        | 577/4722 [6:16:19<46:25:59, 40.33s/it]
 12%|█▏        | 578/4722 [6:16:54<44:31:41, 38.68s/it]
 12%|█▏        | 579/4722 [6:17:42<47:58:50, 41.69s/it]
 12%|█▏        | 580/4722 [6:18:26<48:34:45, 42.22s/it]
                                                       
{'loss': 0.6954, 'learning_rate': 0.0001, 'global_step': 580, 'epoch': 0.37}
+
 12%|█▏        | 580/4722 [6:18:26<48:34:45, 42.22s/it]
 12%|█▏        | 581/4722 [6:19:08<48:38:28, 42.29s/it]
 12%|█▏        | 582/4722 [6:19:47<47:35:44, 41.39s/it]
 12%|█▏        | 583/4722 [6:20:25<46:22:15, 40.33s/it]
 12%|█▏        | 584/4722 [6:20:59<43:57:43, 38.25s/it]
 12%|█▏        | 585/4722 [6:21:36<43:43:40, 38.05s/it]
 12%|█▏        | 586/4722 [6:22:28<48:20:33, 42.08s/it]
 12%|█▏        | 587/4722 [6:23:11<48:38:01, 42.34s/it]
 12%|█▏        | 588/4722 [6:23:51<47:55:59, 41.74s/it]
 12%|█▏        | 589/4722 [6:24:28<46:19:35, 40.35s/it]
 12%|█▏        | 590/4722 [6:25:15<48:33:09, 42.30s/it]
 13%|█▎        | 591/4722 [6:25:54<47:24:30, 41.31s/it]
 13%|█▎        | 592/4722 [6:26:37<47:51:38, 41.72s/it]
 13%|█▎        | 593/4722 [6:27:22<48:58:51, 42.71s/it]
 13%|█▎        | 594/4722 [6:28:03<48:23:00, 42.19s/it]
 13%|█▎        | 595/4722 [6:28:43<47:40:11, 41.58s/it]
 13%|█▎        | 596/4722 [6:29:23<47:19:06, 41.29s/it]
 13%|█▎        | 597/4722 [6:30:02<46:31:57, 40.61s/it]
 13%|█▎        | 598/4722 [6:30:43<46:33:29, 40.64s/it]
 13%|█▎        | 599/4722 [6:31:20<45:16:57, 39.54s/it]
 13%|█▎        | 600/4722 [6:31:58<44:45:26, 39.09s/it]
                                                       
{'loss': 0.6941, 'learning_rate': 0.0001, 'global_step': 600, 'epoch': 0.38}
+
 13%|█▎        | 600/4722 [6:31:58<44:45:26, 39.09s/it]
 13%|█▎        | 601/4722 [6:32:42<46:20:18, 40.48s/it]
 13%|█▎        | 602/4722 [6:33:20<45:22:19, 39.65s/it]
 13%|█▎        | 603/4722 [6:33:59<45:07:03, 39.43s/it]
 13%|█▎        | 604/4722 [6:34:34<43:53:00, 38.36s/it]
 13%|█▎        | 605/4722 [6:35:08<42:22:34, 37.05s/it]
 13%|█▎        | 606/4722 [6:35:45<42:14:20, 36.94s/it]
 13%|█▎        | 607/4722 [6:36:19<41:18:17, 36.14s/it]
 13%|█▎        | 608/4722 [6:36:57<41:51:32, 36.63s/it]
 13%|█▎        | 609/4722 [6:37:39<43:35:15, 38.15s/it]
 13%|█▎        | 610/4722 [6:38:21<45:05:06, 39.47s/it]
 13%|█▎        | 611/4722 [6:39:02<45:36:28, 39.94s/it]
 13%|█▎        | 612/4722 [6:39:41<45:02:43, 39.46s/it]
 13%|█▎        | 613/4722 [6:40:21<45:27:51, 39.83s/it]
 13%|█▎        | 614/4722 [6:40:56<43:36:35, 38.22s/it]
 13%|█▎        | 615/4722 [6:41:36<44:16:15, 38.81s/it]
 13%|█▎        | 616/4722 [6:42:21<46:24:28, 40.69s/it]
 13%|█▎        | 617/4722 [6:43:01<46:06:13, 40.43s/it]
 13%|█▎        | 618/4722 [6:43:46<47:48:12, 41.93s/it]
 13%|█▎        | 619/4722 [6:44:20<44:48:12, 39.31s/it]
 13%|█▎        | 620/4722 [6:44:54<43:09:13, 37.87s/it]
                                                       
{'loss': 0.737, 'learning_rate': 0.0001, 'global_step': 620, 'epoch': 0.39}
+
 13%|█▎        | 620/4722 [6:44:54<43:09:13, 37.87s/it]
 13%|█▎        | 621/4722 [6:45:34<43:53:00, 38.52s/it]
 13%|█▎        | 622/4722 [6:46:11<43:17:22, 38.01s/it]
 13%|█▎        | 623/4722 [6:46:49<43:14:57, 37.98s/it]
 13%|█▎        | 624/4722 [6:47:33<45:15:22, 39.76s/it]
 13%|█▎        | 625/4722 [6:48:10<44:28:35, 39.08s/it]
 13%|█▎        | 626/4722 [6:48:48<43:51:10, 38.54s/it]
 13%|█▎        | 627/4722 [6:49:35<46:42:30, 41.06s/it]
 13%|█▎        | 628/4722 [6:50:13<45:37:47, 40.12s/it]
 13%|█▎        | 629/4722 [6:50:54<46:05:49, 40.54s/it]
 13%|█▎        | 630/4722 [6:51:33<45:23:16, 39.93s/it]
 13%|█▎        | 631/4722 [6:52:16<46:28:36, 40.90s/it]
 13%|█▎        | 632/4722 [6:52:58<47:01:08, 41.39s/it]
 13%|█▎        | 633/4722 [6:53:36<45:51:09, 40.37s/it]
 13%|█▎        | 634/4722 [6:54:20<46:59:01, 41.38s/it]
 13%|█▎        | 635/4722 [6:54:57<45:30:00, 40.08s/it]
 13%|█▎        | 636/4722 [6:55:34<44:31:33, 39.23s/it]
 13%|█▎        | 637/4722 [6:56:18<46:06:05, 40.63s/it]
 14%|█▎        | 638/4722 [6:56:58<45:53:56, 40.46s/it]
 14%|█▎        | 639/4722 [6:57:35<44:47:56, 39.50s/it]
 14%|█▎        | 640/4722 [6:58:16<45:16:41, 39.93s/it]
                                                       
{'loss': 0.7092, 'learning_rate': 0.0001, 'global_step': 640, 'epoch': 0.41}
+
 14%|█▎        | 640/4722 [6:58:16<45:16:41, 39.93s/it]
 14%|█▎        | 641/4722 [6:59:00<46:24:25, 40.94s/it]
 14%|█▎        | 642/4722 [6:59:39<45:41:20, 40.31s/it]
 14%|█▎        | 643/4722 [7:00:18<45:17:44, 39.98s/it]
 14%|█▎        | 644/4722 [7:00:55<44:15:21, 39.07s/it]
 14%|█▎        | 645/4722 [7:01:30<43:05:50, 38.05s/it]
 14%|█▎        | 646/4722 [7:02:05<41:49:14, 36.94s/it]
 14%|█▎        | 647/4722 [7:02:50<44:34:56, 39.39s/it]
 14%|█▎        | 648/4722 [7:03:27<43:41:45, 38.61s/it]
 14%|█▎        | 649/4722 [7:04:12<46:02:46, 40.70s/it]
 14%|█▍        | 650/4722 [7:04:57<47:26:18, 41.94s/it]
 14%|█▍        | 651/4722 [7:05:37<46:47:12, 41.37s/it]
 14%|█▍        | 652/4722 [7:06:17<46:20:05, 40.98s/it]
 14%|█▍        | 653/4722 [7:06:54<45:05:46, 39.90s/it]
 14%|█▍        | 654/4722 [7:07:35<45:10:44, 39.98s/it]
 14%|█▍        | 655/4722 [7:08:06<42:17:42, 37.44s/it]
 14%|█▍        | 656/4722 [7:08:44<42:29:42, 37.62s/it]
 14%|���▍        | 657/4722 [7:09:24<43:07:59, 38.20s/it]
 14%|█▍        | 658/4722 [7:09:59<42:12:11, 37.38s/it]
 14%|█▍        | 659/4722 [7:10:40<43:20:42, 38.41s/it]
 14%|█▍        | 660/4722 [7:11:19<43:23:35, 38.46s/it]
                                                       
{'loss': 0.6924, 'learning_rate': 0.0001, 'global_step': 660, 'epoch': 0.42}
+
 14%|█▍        | 660/4722 [7:11:19<43:23:35, 38.46s/it]
 14%|█▍        | 661/4722 [7:11:53<42:10:14, 37.38s/it]
 14%|█▍        | 662/4722 [7:12:32<42:32:41, 37.72s/it]
 14%|█▍        | 663/4722 [7:13:10<42:33:11, 37.74s/it]
 14%|█▍        | 664/4722 [7:13:46<42:04:48, 37.33s/it]
 14%|█▍        | 665/4722 [7:14:32<45:00:30, 39.94s/it]
 14%|█▍        | 666/4722 [7:15:09<44:04:29, 39.12s/it]
 14%|█▍        | 667/4722 [7:15:55<46:17:00, 41.09s/it]
 14%|█▍        | 668/4722 [7:16:38<46:47:21, 41.55s/it]
 14%|█▍        | 669/4722 [7:17:12<44:15:40, 39.31s/it]
 14%|█▍        | 670/4722 [7:17:55<45:39:43, 40.57s/it]
 14%|█▍        | 671/4722 [7:18:31<43:59:05, 39.09s/it]
 14%|█▍        | 672/4722 [7:19:04<41:56:27, 37.28s/it]
 14%|█▍        | 673/4722 [7:19:42<42:07:11, 37.45s/it]
 14%|█▍        | 674/4722 [7:20:21<42:46:07, 38.04s/it]
 14%|█▍        | 675/4722 [7:21:03<44:06:59, 39.24s/it]
 14%|█▍        | 676/4722 [7:21:40<43:11:08, 38.43s/it]
 14%|█▍        | 677/4722 [7:22:11<40:38:12, 36.17s/it]
 14%|█▍        | 678/4722 [7:22:50<41:37:09, 37.05s/it]
 14%|█▍        | 679/4722 [7:23:26<41:08:43, 36.64s/it]
 14%|█▍        | 680/4722 [7:24:11<44:02:31, 39.23s/it]
                                                       
{'loss': 0.685, 'learning_rate': 0.0001, 'global_step': 680, 'epoch': 0.43}
+
 14%|█▍        | 680/4722 [7:24:11<44:02:31, 39.23s/it]
 14%|█▍        | 681/4722 [7:24:57<46:27:07, 41.38s/it]
 14%|█▍        | 682/4722 [7:25:40<46:57:36, 41.85s/it]
 14%|█▍        | 683/4722 [7:26:27<48:36:53, 43.33s/it]
 14%|█▍        | 684/4722 [7:27:06<47:07:10, 42.01s/it]
 15%|█▍        | 685/4722 [7:27:45<46:07:01, 41.12s/it]
 15%|█▍        | 686/4722 [7:28:22<44:40:13, 39.84s/it]
 15%|█▍        | 687/4722 [7:29:08<46:47:41, 41.75s/it]
 15%|█▍        | 688/4722 [7:29:47<45:52:00, 40.93s/it]
 15%|█▍        | 689/4722 [7:30:28<45:56:16, 41.01s/it]
 15%|█▍        | 690/4722 [7:31:01<43:07:58, 38.51s/it]
 15%|█▍        | 691/4722 [7:31:37<42:26:43, 37.91s/it]
 15%|█▍        | 692/4722 [7:32:14<42:05:35, 37.60s/it]
 15%|█▍        | 693/4722 [7:32:51<41:48:34, 37.36s/it]
 15%|█▍        | 694/4722 [7:33:28<41:31:15, 37.11s/it]
 15%|█▍        | 695/4722 [7:34:08<42:48:00, 38.26s/it]
 15%|█▍        | 696/4722 [7:34:45<42:13:37, 37.76s/it]
 15%|█▍        | 697/4722 [7:35:22<42:04:39, 37.63s/it]
 15%|█▍        | 698/4722 [7:36:02<42:52:59, 38.36s/it]
 15%|█▍        | 699/4722 [7:36:37<41:41:02, 37.30s/it]
 15%|█▍        | 700/4722 [7:37:15<41:43:38, 37.35s/it]
                                                       
{'loss': 0.6935, 'learning_rate': 0.0001, 'global_step': 700, 'epoch': 0.44}
+
 15%|█▍        | 700/4722 [7:37:15<41:43:38, 37.35s/it]
 15%|█▍        | 701/4722 [7:37:55<42:43:29, 38.25s/it]
 15%|█▍        | 702/4722 [7:38:36<43:29:55, 38.95s/it]
 15%|█▍        | 703/4722 [7:39:13<42:58:41, 38.50s/it]
 15%|█▍        | 704/4722 [7:39:47<41:29:42, 37.18s/it]
 15%|█▍        | 705/4722 [7:40:22<40:45:20, 36.52s/it]
 15%|█▍        | 706/4722 [7:41:03<41:59:48, 37.65s/it]
 15%|█▍        | 707/4722 [7:41:38<41:07:34, 36.88s/it]
 15%|█▍        | 708/4722 [7:42:19<42:34:19, 38.18s/it]
 15%|█▌        | 709/4722 [7:43:00<43:43:20, 39.22s/it]
 15%|█▌        | 710/4722 [7:43:43<44:56:44, 40.33s/it]
 15%|█▌        | 711/4722 [7:44:26<45:43:34, 41.04s/it]
 15%|█▌        | 712/4722 [7:45:05<45:04:04, 40.46s/it]
 15%|█▌        | 713/4722 [7:45:51<46:54:21, 42.12s/it]
 15%|█▌        | 714/4722 [7:46:21<42:45:04, 38.40s/it]
 15%|█▌        | 715/4722 [7:47:07<45:19:06, 40.72s/it]
 15%|█▌        | 716/4722 [7:47:49<45:45:38, 41.12s/it]
 15%|█▌        | 717/4722 [7:48:28<45:05:15, 40.53s/it]
 15%|█▌        | 718/4722 [7:49:07<44:20:39, 39.87s/it]
 15%|█▌        | 719/4722 [7:49:40<42:18:32, 38.05s/it]
 15%|█▌        | 720/4722 [7:50:18<42:01:08, 37.80s/it]
                                                       
{'loss': 0.7088, 'learning_rate': 0.0001, 'global_step': 720, 'epoch': 0.46}
+
 15%|█▌        | 720/4722 [7:50:18<42:01:08, 37.80s/it]
 15%|█▌        | 721/4722 [7:50:52<40:46:49, 36.69s/it]
 15%|█▌        | 722/4722 [7:51:34<42:31:04, 38.27s/it]
 15%|█▌        | 723/4722 [7:52:12<42:22:56, 38.15s/it]
 15%|█▌        | 724/4722 [7:52:41<39:35:51, 35.66s/it]
 15%|█▌        | 725/4722 [7:53:20<40:31:27, 36.50s/it]
 15%|█▌        | 726/4722 [7:53:56<40:17:01, 36.29s/it]
 15%|█▌        | 727/4722 [7:54:36<41:28:27, 37.37s/it]
 15%|█▌        | 728/4722 [7:55:15<42:01:28, 37.88s/it]
 15%|█▌        | 729/4722 [7:55:50<41:14:22, 37.18s/it]
 15%|█▌        | 730/4722 [7:56:26<40:46:29, 36.77s/it]
 15%|█▌        | 731/4722 [7:57:02<40:31:39, 36.56s/it]
 16%|█▌        | 732/4722 [7:57:45<42:40:51, 38.51s/it]
 16%|█▌        | 733/4722 [7:58:34<46:10:23, 41.67s/it]
 16%|█▌        | 734/4722 [7:59:13<45:17:59, 40.89s/it]
 16%|█▌        | 735/4722 [7:59:53<45:04:37, 40.70s/it]
 16%|█▌        | 736/4722 [8:00:26<42:12:37, 38.12s/it]
 16%|█▌        | 737/4722 [8:01:10<44:07:53, 39.87s/it]
 16%|█▌        | 738/4722 [8:01:47<43:26:38, 39.26s/it]
 16%|█▌        | 739/4722 [8:02:22<42:02:52, 38.00s/it]
 16%|█▌        | 740/4722 [8:03:02<42:39:29, 38.57s/it]
                                                       
{'loss': 0.6861, 'learning_rate': 0.0001, 'global_step': 740, 'epoch': 0.47}
+
 16%|█▌        | 740/4722 [8:03:02<42:39:29, 38.57s/it]
 16%|█▌        | 741/4722 [8:03:42<43:07:47, 39.00s/it]
 16%|█▌        | 742/4722 [8:04:22<43:22:33, 39.23s/it]
 16%|█▌        | 743/4722 [8:04:53<40:40:45, 36.80s/it]
 16%|█▌        | 744/4722 [8:05:42<44:42:47, 40.46s/it]
 16%|█▌        | 745/4722 [8:06:21<44:00:31, 39.84s/it]
 16%|█▌        | 746/4722 [8:07:01<44:04:19, 39.90s/it]
 16%|█▌        | 747/4722 [8:07:39<43:35:47, 39.48s/it]
 16%|█▌        | 748/4722 [8:08:19<43:44:01, 39.62s/it]
 16%|█▌        | 749/4722 [8:09:06<46:01:36, 41.71s/it]
 16%|█▌        | 750/4722 [8:09:47<45:58:17, 41.67s/it]
 16%|█▌        | 751/4722 [8:10:23<44:07:36, 40.00s/it]
 16%|█▌        | 752/4722 [8:11:04<44:09:23, 40.04s/it]
 16%|█▌        | 753/4722 [8:11:44<44:07:56, 40.03s/it]
 16%|█▌        | 754/4722 [8:12:19<42:32:27, 38.60s/it]
 16%|█▌        | 755/4722 [8:13:01<43:51:54, 39.81s/it]
 16%|█▌        | 756/4722 [8:13:50<46:46:12, 42.45s/it]
 16%|█▌        | 757/4722 [8:14:31<46:21:55, 42.10s/it]
 16%|█▌        | 758/4722 [8:15:10<45:05:58, 40.96s/it]
 16%|█▌        | 759/4722 [8:15:49<44:38:20, 40.55s/it]
 16%|█▌        | 760/4722 [8:16:38<47:14:18, 42.92s/it]
                                                       
{'loss': 0.691, 'learning_rate': 0.0001, 'global_step': 760, 'epoch': 0.48}
+
 16%|█▌        | 760/4722 [8:16:38<47:14:18, 42.92s/it]
 16%|█▌        | 761/4722 [8:17:21<47:22:51, 43.06s/it]
 16%|█▌        | 762/4722 [8:18:03<47:03:30, 42.78s/it]
 16%|█▌        | 763/4722 [8:18:46<47:01:12, 42.76s/it]
 16%|█▌        | 764/4722 [8:19:25<45:51:37, 41.71s/it]
 16%|█▌        | 765/4722 [8:20:04<45:02:27, 40.98s/it]
 16%|█▌        | 766/4722 [8:20:43<44:08:44, 40.17s/it]
 16%|█▌        | 767/4722 [8:21:28<45:47:45, 41.69s/it]
 16%|█▋        | 768/4722 [8:22:05<44:21:47, 40.39s/it]
 16%|█▋        | 769/4722 [8:22:44<43:55:15, 40.00s/it]
 16%|█▋        | 770/4722 [8:23:18<41:53:16, 38.16s/it]
 16%|█▋        | 771/4722 [8:23:57<42:13:28, 38.47s/it]
 16%|█▋        | 772/4722 [8:24:34<41:34:43, 37.89s/it]
 16%|█▋        | 773/4722 [8:25:20<44:06:42, 40.21s/it]
 16%|█▋        | 774/4722 [8:25:54<42:09:25, 38.44s/it]
 16%|█▋        | 775/4722 [8:26:37<43:36:55, 39.78s/it]
 16%|█▋        | 776/4722 [8:27:08<40:51:38, 37.28s/it]
 16%|█▋        | 777/4722 [8:27:43<40:09:43, 36.65s/it]
 16%|█▋        | 778/4722 [8:28:13<37:48:30, 34.51s/it]
 16%|█▋        | 779/4722 [8:28:49<38:12:08, 34.88s/it]
 17%|█▋        | 780/4722 [8:29:28<39:42:55, 36.27s/it]
                                                       
{'loss': 0.6685, 'learning_rate': 0.0001, 'global_step': 780, 'epoch': 0.5}
+
 17%|█▋        | 780/4722 [8:29:28<39:42:55, 36.27s/it]
 17%|█▋        | 781/4722 [8:30:04<39:38:28, 36.21s/it]
 17%|█▋        | 782/4722 [8:30:49<42:19:14, 38.67s/it]
 17%|█▋        | 783/4722 [8:31:24<41:21:53, 37.80s/it]
 17%|█▋        | 784/4722 [8:32:07<43:00:09, 39.31s/it]
 17%|█▋        | 785/4722 [8:32:52<44:55:20, 41.08s/it]
 17%|█▋        | 786/4722 [8:33:35<45:27:15, 41.57s/it]
 17%|█▋        | 787/4722 [8:34:15<44:54:37, 41.09s/it]
 17%|█▋        | 788/4722 [8:34:53<43:57:11, 40.22s/it]
 17%|█▋        | 789/4722 [8:35:30<42:51:56, 39.24s/it]
 17%|█▋        | 790/4722 [8:36:09<42:45:59, 39.16s/it]
 17%|█▋        | 791/4722 [8:36:51<43:31:49, 39.86s/it]
 17%|█▋        | 792/4722 [8:37:28<42:39:54, 39.08s/it]
 17%|█▋        | 793/4722 [8:38:10<43:26:48, 39.81s/it]
 17%|█▋        | 794/4722 [8:38:51<44:06:57, 40.43s/it]
 17%|█▋        | 795/4722 [8:39:26<42:07:44, 38.62s/it]
 17%|█▋        | 796/4722 [8:40:02<41:14:35, 37.82s/it]
 17%|█▋        | 797/4722 [8:40:35<39:39:13, 36.37s/it]
 17%|█▋        | 798/4722 [8:41:15<40:59:11, 37.60s/it]
 17%|█▋        | 799/4722 [8:41:47<39:03:43, 35.85s/it]
 17%|█▋        | 800/4722 [8:42:25<39:40:00, 36.41s/it]
                                                       
{'loss': 0.6819, 'learning_rate': 0.0001, 'global_step': 800, 'epoch': 0.51}
+
 17%|█▋        | 800/4722 [8:42:25<39:40:00, 36.41s/it]
 17%|█▋        | 801/4722 [8:43:03<40:12:39, 36.92s/it]
 17%|█▋        | 802/4722 [8:43:39<40:05:27, 36.82s/it]
 17%|█▋        | 803/4722 [8:44:15<39:41:18, 36.46s/it]
 17%|█▋        | 804/4722 [8:44:48<38:41:33, 35.55s/it]
 17%|█▋        | 805/4722 [8:45:28<40:04:57, 36.84s/it]
 17%|█▋        | 806/4722 [8:46:10<41:31:26, 38.17s/it]
 17%|█▋        | 807/4722 [8:46:50<42:05:37, 38.71s/it]
 17%|█▋        | 808/4722 [8:47:28<41:58:09, 38.60s/it]
 17%|█▋        | 809/4722 [8:48:01<40:18:14, 37.08s/it]
 17%|█▋        | 810/4722 [8:48:36<39:27:00, 36.30s/it]
 17%|█▋        | 811/4722 [8:49:14<40:09:47, 36.97s/it]
 17%|█▋        | 812/4722 [8:50:01<43:08:25, 39.72s/it]
 17%|█▋        | 813/4722 [8:50:44<44:22:54, 40.87s/it]
 17%|█▋        | 814/4722 [8:51:26<44:48:02, 41.27s/it]
 17%|█▋        | 815/4722 [8:52:00<42:20:01, 39.01s/it]
 17%|█▋        | 816/4722 [8:52:47<45:01:11, 41.49s/it]
 17%|█▋        | 817/4722 [8:53:27<44:17:48, 40.84s/it]
 17%|█▋        | 818/4722 [8:54:04<43:04:58, 39.73s/it]
 17%|█▋        | 819/4722 [8:54:40<41:55:43, 38.67s/it]
 17%|█▋        | 820/4722 [8:55:18<41:46:54, 38.55s/it]
                                                       
{'loss': 0.6847, 'learning_rate': 0.0001, 'global_step': 820, 'epoch': 0.52}
+
 17%|█▋        | 820/4722 [8:55:18<41:46:54, 38.55s/it]
 17%|█▋        | 821/4722 [8:56:03<43:38:04, 40.27s/it]
 17%|█▋        | 822/4722 [8:56:41<43:00:03, 39.69s/it]
 17%|█▋        | 823/4722 [8:57:18<42:10:11, 38.94s/it]
 17%|█▋        | 824/4722 [8:57:58<42:24:22, 39.16s/it]
 17%|█▋        | 825/4722 [8:58:43<44:29:37, 41.10s/it]
 17%|█▋        | 826/4722 [8:59:23<43:51:22, 40.52s/it]
 18%|█▊        | 827/4722 [9:00:06<44:54:18, 41.50s/it]
 18%|█▊        | 828/4722 [9:00:50<45:41:34, 42.24s/it]
 18%|█▊        | 829/4722 [9:01:30<44:59:38, 41.61s/it]
 18%|█▊        | 830/4722 [9:02:13<45:24:33, 42.00s/it]
 18%|█▊        | 831/4722 [9:03:01<47:14:52, 43.71s/it]
 18%|█▊        | 832/4722 [9:03:44<46:50:34, 43.35s/it]
 18%|█▊        | 833/4722 [9:04:33<48:42:24, 45.09s/it]
 18%|█▊        | 834/4722 [9:05:06<44:54:15, 41.58s/it]
 18%|█▊        | 835/4722 [9:05:38<41:47:33, 38.71s/it]
 18%|█▊        | 836/4722 [9:06:25<44:34:33, 41.30s/it]
 18%|█▊        | 837/4722 [9:07:02<42:54:10, 39.76s/it]
 18%|█▊        | 838/4722 [9:07:44<43:49:03, 40.61s/it]
 18%|█▊        | 839/4722 [9:08:21<42:38:08, 39.53s/it]
 18%|█▊        | 840/4722 [9:09:05<43:55:18, 40.73s/it]
                                                       
{'loss': 0.6785, 'learning_rate': 0.0001, 'global_step': 840, 'epoch': 0.53}
+
 18%|█▊        | 840/4722 [9:09:05<43:55:18, 40.73s/it]
 18%|█▊        | 841/4722 [9:09:39<41:51:32, 38.83s/it]
 18%|█▊        | 842/4722 [9:10:23<43:18:56, 40.19s/it]
 18%|█▊        | 843/4722 [9:10:59<42:11:48, 39.16s/it]
 18%|█▊        | 844/4722 [9:11:37<41:40:16, 38.68s/it]
 18%|█▊        | 845/4722 [9:12:09<39:33:08, 36.73s/it]
 18%|█▊        | 846/4722 [9:12:48<40:22:05, 37.49s/it]
 18%|█▊        | 847/4722 [9:13:30<41:38:27, 38.69s/it]
 18%|█▊        | 848/4722 [9:14:04<40:10:40, 37.34s/it]
 18%|█▊        | 849/4722 [9:14:46<41:35:16, 38.66s/it]
 18%|█▊        | 850/4722 [9:15:19<39:56:05, 37.13s/it]
 18%|█▊        | 851/4722 [9:15:59<40:37:25, 37.78s/it]
 18%|█▊        | 852/4722 [9:16:36<40:37:18, 37.79s/it]
 18%|█▊        | 853/4722 [9:17:19<42:17:38, 39.35s/it]
 18%|█▊        | 854/4722 [9:17:54<40:53:02, 38.05s/it]
 18%|█▊        | 855/4722 [9:18:36<42:09:14, 39.24s/it]
 18%|█▊        | 856/4722 [9:19:15<41:51:10, 38.97s/it]
 18%|█▊        | 857/4722 [9:19:58<43:03:42, 40.11s/it]
 18%|█▊        | 858/4722 [9:20:33<41:33:36, 38.72s/it]
 18%|█▊        | 859/4722 [9:21:16<42:51:51, 39.95s/it]
 18%|█▊        | 860/4722 [9:21:56<42:55:31, 40.01s/it]
                                                       
{'loss': 0.6951, 'learning_rate': 0.0001, 'global_step': 860, 'epoch': 0.55}
+
 18%|█▊        | 860/4722 [9:21:56<42:55:31, 40.01s/it]
 18%|█▊        | 861/4722 [9:22:38<43:26:32, 40.51s/it]
 18%|█▊        | 862/4722 [9:23:15<42:29:01, 39.62s/it]
 18%|█▊        | 863/4722 [9:23:53<41:47:11, 38.98s/it]
 18%|█▊        | 864/4722 [9:24:33<42:11:02, 39.36s/it]
 18%|█▊        | 865/4722 [9:25:13<42:23:29, 39.57s/it]
 18%|█▊        | 866/4722 [9:25:50<41:41:13, 38.92s/it]
 18%|█▊        | 867/4722 [9:26:34<43:10:59, 40.33s/it]
 18%|█▊        | 868/4722 [9:27:08<41:02:26, 38.34s/it]
 18%|█▊        | 869/4722 [9:27:43<40:10:15, 37.53s/it]
 18%|█▊        | 870/4722 [9:28:24<41:06:51, 38.42s/it]
 18%|█▊        | 871/4722 [9:29:01<40:39:31, 38.01s/it]
 18%|█▊        | 872/4722 [9:29:46<42:46:40, 40.00s/it]
 18%|█▊        | 873/4722 [9:30:30<44:16:54, 41.42s/it]
 19%|█▊        | 874/4722 [9:31:19<46:40:30, 43.67s/it]
 19%|█▊        | 875/4722 [9:32:02<46:21:32, 43.38s/it]
 19%|█▊        | 876/4722 [9:32:43<45:31:40, 42.62s/it]
 19%|█▊        | 877/4722 [9:33:25<45:28:14, 42.57s/it]
 19%|█▊        | 878/4722 [9:34:07<45:05:33, 42.23s/it]
 19%|█▊        | 879/4722 [9:34:41<42:28:47, 39.79s/it]
 19%|█▊        | 880/4722 [9:35:20<42:15:51, 39.60s/it]
                                                       
{'loss': 0.6787, 'learning_rate': 0.0001, 'global_step': 880, 'epoch': 0.56}
+
 19%|█▊        | 880/4722 [9:35:20<42:15:51, 39.60s/it]
 19%|█▊        | 881/4722 [9:35:52<39:44:18, 37.24s/it]
 19%|█▊        | 882/4722 [9:36:37<42:13:16, 39.58s/it]
 19%|█▊        | 883/4722 [9:37:18<42:39:40, 40.01s/it]
 19%|█▊        | 884/4722 [9:37:57<42:30:37, 39.87s/it]
 19%|█▊        | 885/4722 [9:38:38<42:46:13, 40.13s/it]
 19%|█▉        | 886/4722 [9:39:13<41:16:02, 38.73s/it]
 19%|█▉        | 887/4722 [9:39:52<41:15:47, 38.73s/it]
 19%|█▉        | 888/4722 [9:40:28<40:15:07, 37.80s/it]
 19%|█▉        | 889/4722 [9:41:19<44:40:08, 41.95s/it]
 19%|█▉        | 890/4722 [9:42:01<44:26:54, 41.76s/it]
 19%|█▉        | 891/4722 [9:42:36<42:20:54, 39.79s/it]
 19%|█▉        | 892/4722 [9:43:14<41:47:41, 39.28s/it]
 19%|█▉        | 893/4722 [9:43:55<42:27:25, 39.92s/it]
 19%|█▉        | 894/4722 [9:44:35<42:19:23, 39.80s/it]
 19%|█▉        | 895/4722 [9:45:15<42:15:20, 39.75s/it]
 19%|█▉        | 896/4722 [9:46:00<43:55:49, 41.34s/it]
 19%|█▉        | 897/4722 [9:46:33<41:13:40, 38.80s/it]
 19%|█▉        | 898/4722 [9:47:13<41:35:15, 39.15s/it]
 19%|█▉        | 899/4722 [9:47:53<41:59:44, 39.55s/it]
 19%|█▉        | 900/4722 [9:48:28<40:38:28, 38.28s/it]
                                                       
{'loss': 0.6704, 'learning_rate': 0.0001, 'global_step': 900, 'epoch': 0.57}
+
 19%|█▉        | 900/4722 [9:48:28<40:38:28, 38.28s/it]
 19%|█▉        | 901/4722 [9:49:11<41:59:42, 39.57s/it]
 19%|█▉        | 902/4722 [9:49:42<39:16:23, 37.01s/it]
 19%|█▉        | 903/4722 [9:50:22<40:10:46, 37.88s/it]
 19%|█▉        | 904/4722 [9:51:03<41:16:27, 38.92s/it]
 19%|█▉        | 905/4722 [9:51:38<39:57:18, 37.68s/it]
 19%|█▉        | 906/4722 [9:52:17<40:16:49, 38.00s/it]
 19%|█▉        | 907/4722 [9:53:01<42:25:08, 40.03s/it]
 19%|█▉        | 908/4722 [9:53:37<41:03:55, 38.76s/it]
 19%|█▉        | 909/4722 [9:54:12<39:37:44, 37.42s/it]
 19%|█▉        | 910/4722 [9:54:47<38:52:04, 36.71s/it]
 19%|█▉        | 911/4722 [9:55:23<38:42:40, 36.57s/it]
 19%|█▉        | 912/4722 [9:55:57<37:59:08, 35.89s/it]
 19%|█▉        | 913/4722 [9:56:32<37:47:45, 35.72s/it]
 19%|█▉        | 914/4722 [9:57:04<36:29:39, 34.50s/it]
 19%|█▉        | 915/4722 [9:57:39<36:26:44, 34.46s/it]
 19%|█▉        | 916/4722 [9:58:16<37:22:48, 35.36s/it]
 19%|█▉        | 917/4722 [9:58:50<36:58:28, 34.98s/it]
 19%|█▉        | 918/4722 [9:59:29<38:09:35, 36.11s/it]
 19%|█▉        | 919/4722 [10:00:05<38:16:56, 36.24s/it]
 19%|█▉        | 920/4722 [10:00:52<41:34:14, 39.36s/it]
                                                        
{'loss': 0.6836, 'learning_rate': 0.0001, 'global_step': 920, 'epoch': 0.58}
+
 19%|█▉        | 920/4722 [10:00:52<41:34:14, 39.36s/it]
 20%|█▉        | 921/4722 [10:01:35<42:45:58, 40.50s/it]
 20%|█▉        | 922/4722 [10:02:18<43:33:35, 41.27s/it]
 20%|█▉        | 923/4722 [10:02:55<42:14:29, 40.03s/it]
 20%|█▉        | 924/4722 [10:03:34<41:51:23, 39.67s/it]
 20%|█▉        | 925/4722 [10:04:13<41:42:14, 39.54s/it]
 20%|█▉        | 926/4722 [10:04:53<41:35:53, 39.45s/it]
 20%|█▉        | 927/4722 [10:05:27<39:50:47, 37.80s/it]
 20%|█▉        | 928/4722 [10:05:57<37:34:26, 35.65s/it]
 20%|█▉        | 929/4722 [10:06:35<38:04:54, 36.14s/it]
 20%|█▉        | 930/4722 [10:07:19<40:51:23, 38.79s/it]
 20%|█▉        | 931/4722 [10:07:58<40:52:46, 38.82s/it]
 20%|█▉        | 932/4722 [10:08:35<40:13:37, 38.21s/it]
 20%|█▉        | 933/4722 [10:09:14<40:31:45, 38.51s/it]
 20%|█▉        | 934/4722 [10:09:53<40:30:45, 38.50s/it]
 20%|█▉        | 935/4722 [10:10:29<39:51:34, 37.89s/it]
 20%|█▉        | 936/4722 [10:11:03<38:27:04, 36.56s/it]
 20%|█▉        | 937/4722 [10:11:36<37:28:03, 35.64s/it]
 20%|█▉        | 938/4722 [10:12:14<38:05:23, 36.24s/it]
 20%|█▉        | 939/4722 [10:13:00<41:16:04, 39.27s/it]
 20%|█▉        | 940/4722 [10:13:41<41:38:10, 39.63s/it]
                                                        
{'loss': 0.6727, 'learning_rate': 0.0001, 'global_step': 940, 'epoch': 0.6}
+
 20%|█▉        | 940/4722 [10:13:41<41:38:10, 39.63s/it]
 20%|█▉        | 941/4722 [10:14:21<41:54:54, 39.91s/it]
 20%|█▉        | 942/4722 [10:15:02<42:13:07, 40.21s/it]
 20%|█▉        | 943/4722 [10:15:41<41:53:32, 39.91s/it]
 20%|█▉        | 944/4722 [10:16:30<44:36:50, 42.51s/it]
 20%|██        | 945/4722 [10:17:11<44:16:38, 42.20s/it]
 20%|██        | 946/4722 [10:17:48<42:19:43, 40.36s/it]
 20%|██        | 947/4722 [10:18:26<41:48:24, 39.87s/it]
 20%|██        | 948/4722 [10:19:09<42:45:41, 40.79s/it]
 20%|██        | 949/4722 [10:19:46<41:28:48, 39.58s/it]
 20%|██        | 950/4722 [10:20:21<40:05:47, 38.27s/it]
 20%|██        | 951/4722 [10:21:01<40:41:10, 38.84s/it]
 20%|██        | 952/4722 [10:21:42<41:21:09, 39.49s/it]
 20%|██        | 953/4722 [10:22:25<42:11:23, 40.30s/it]
 20%|██        | 954/4722 [10:23:13<44:43:13, 42.73s/it]
 20%|██        | 955/4722 [10:23:49<42:41:45, 40.80s/it]
 20%|██        | 956/4722 [10:24:33<43:43:26, 41.80s/it]
 20%|██        | 957/4722 [10:25:11<42:34:02, 40.70s/it]
 20%|██        | 958/4722 [10:25:50<41:58:14, 40.14s/it]
 20%|██        | 959/4722 [10:26:30<41:40:55, 39.88s/it]
 20%|██        | 960/4722 [10:27:06<40:44:24, 38.99s/it]
                                                        
{'loss': 0.6839, 'learning_rate': 0.0001, 'global_step': 960, 'epoch': 0.61}
+
 20%|██        | 960/4722 [10:27:06<40:44:24, 38.99s/it]
 20%|██        | 961/4722 [10:27:47<41:17:44, 39.53s/it]
 20%|██        | 962/4722 [10:28:28<41:32:59, 39.78s/it]
 20%|██        | 963/4722 [10:29:09<42:00:19, 40.23s/it]
 20%|██        | 964/4722 [10:29:45<40:36:11, 38.90s/it]
 20%|██        | 965/4722 [10:30:19<39:02:54, 37.42s/it]
 20%|██        | 966/4722 [10:30:52<37:42:43, 36.15s/it]
 20%|██        | 967/4722 [10:31:38<40:41:44, 39.02s/it]
 20%|██        | 968/4722 [10:32:14<39:49:26, 38.19s/it]
 21%|██        | 969/4722 [10:32:52<39:45:29, 38.14s/it]
 21%|██        | 970/4722 [10:33:30<39:44:44, 38.14s/it]
 21%|██        | 971/4722 [10:34:20<43:25:48, 41.68s/it]
 21%|██        | 972/4722 [10:34:57<41:57:34, 40.28s/it]
 21%|██        | 973/4722 [10:35:38<42:11:12, 40.51s/it]
 21%|██        | 974/4722 [10:36:13<40:35:14, 38.98s/it]
 21%|██        | 975/4722 [10:36:59<42:34:21, 40.90s/it]
 21%|██        | 976/4722 [10:37:38<42:02:38, 40.41s/it]
 21%|██        | 977/4722 [10:38:15<40:58:01, 39.38s/it]
 21%|██        | 978/4722 [10:38:55<41:11:40, 39.61s/it]
 21%|██        | 979/4722 [10:39:31<40:02:33, 38.51s/it]
 21%|██        | 980/4722 [10:40:14<41:28:13, 39.90s/it]
                                                        
{'loss': 0.6351, 'learning_rate': 0.0001, 'global_step': 980, 'epoch': 0.62}
+
 21%|██        | 980/4722 [10:40:14<41:28:13, 39.90s/it]
 21%|██        | 981/4722 [10:40:46<39:03:21, 37.58s/it]
 21%|██        | 982/4722 [10:41:23<38:43:34, 37.28s/it]
 21%|██        | 983/4722 [10:42:02<39:16:18, 37.81s/it]
 21%|██        | 984/4722 [10:42:43<40:21:30, 38.87s/it]
 21%|██        | 985/4722 [10:43:20<39:31:06, 38.07s/it]
 21%|██        | 986/4722 [10:44:02<40:54:25, 39.42s/it]
 21%|██        | 987/4722 [10:44:42<41:05:38, 39.61s/it]
 21%|██        | 988/4722 [10:45:21<40:50:26, 39.38s/it]
 21%|██        | 989/4722 [10:46:07<42:43:32, 41.20s/it]
 21%|██        | 990/4722 [10:46:52<43:53:34, 42.34s/it]
 21%|██        | 991/4722 [10:47:28<42:09:53, 40.68s/it]
 21%|██        | 992/4722 [10:48:15<44:03:29, 42.52s/it]
 21%|██        | 993/4722 [10:49:01<44:58:03, 43.41s/it]
 21%|██        | 994/4722 [10:49:40<43:32:56, 42.05s/it]
 21%|██        | 995/4722 [10:50:18<42:24:14, 40.96s/it]
 21%|██        | 996/4722 [10:50:56<41:32:52, 40.14s/it]
 21%|██        | 997/4722 [10:51:34<40:52:08, 39.50s/it]
 21%|██        | 998/4722 [10:52:08<39:05:44, 37.79s/it]
 21%|██        | 999/4722 [10:52:53<41:13:23, 39.86s/it]
 21%|██        | 1000/4722 [10:53:29<40:02:07, 38.72s/it]
                                                         
{'loss': 0.6772, 'learning_rate': 0.0001, 'global_step': 1000, 'epoch': 0.64}
+
 21%|██        | 1000/4722 [10:53:29<40:02:07, 38.72s/it]
 21%|██        | 1001/4722 [10:54:16<42:43:46, 41.34s/it]
 21%|██        | 1002/4722 [10:54:49<40:05:28, 38.80s/it]
 21%|██        | 1003/4722 [10:55:27<39:49:40, 38.55s/it]
 21%|██▏       | 1004/4722 [10:56:06<39:54:48, 38.65s/it]
 21%|██▏       | 1005/4722 [10:56:42<39:00:14, 37.78s/it]
 21%|██▏       | 1006/4722 [10:57:21<39:20:35, 38.12s/it]
 21%|██▏       | 1007/4722 [10:57:59<39:35:24, 38.36s/it]
 21%|██▏       | 1008/4722 [10:58:38<39:44:10, 38.52s/it]
 21%|██▏       | 1009/4722 [10:59:23<41:39:39, 40.39s/it]
 21%|██▏       | 1010/4722 [11:00:03<41:20:15, 40.09s/it]
 21%|██▏       | 1011/4722 [11:00:41<40:55:01, 39.69s/it]
 21%|██▏       | 1012/4722 [11:01:25<42:02:49, 40.80s/it]
 21%|██▏       | 1013/4722 [11:02:07<42:35:10, 41.33s/it]
 21%|██▏       | 1014/4722 [11:02:56<44:50:40, 43.54s/it]
 21%|██▏       | 1015/4722 [11:03:43<45:47:15, 44.47s/it]
 22%|██▏       | 1016/4722 [11:04:25<45:01:07, 43.73s/it]
 22%|██▏       | 1017/4722 [11:05:09<45:04:17, 43.79s/it]
 22%|██▏       | 1018/4722 [11:05:45<42:49:37, 41.62s/it]
 22%|██▏       | 1019/4722 [11:06:27<42:49:40, 41.64s/it]
 22%|██▏       | 1020/4722 [11:07:03<41:16:34, 40.14s/it]
                                                         
{'loss': 0.6626, 'learning_rate': 0.0001, 'global_step': 1020, 'epoch': 0.65}
+
 22%|██▏       | 1020/4722 [11:07:03<41:16:34, 40.14s/it]
 22%|██▏       | 1021/4722 [11:07:48<42:35:57, 41.44s/it]
 22%|██▏       | 1022/4722 [11:08:29<42:27:05, 41.30s/it]
 22%|██▏       | 1023/4722 [11:09:04<40:31:25, 39.44s/it]
 22%|██▏       | 1024/4722 [11:09:43<40:20:50, 39.28s/it]
 22%|██▏       | 1025/4722 [11:10:24<40:52:07, 39.80s/it]
 22%|██▏       | 1026/4722 [11:11:02<40:24:58, 39.37s/it]
 22%|██▏       | 1027/4722 [11:11:44<41:01:15, 39.97s/it]
 22%|██▏       | 1028/4722 [11:12:26<41:53:05, 40.82s/it]
 22%|██▏       | 1029/4722 [11:13:09<42:16:32, 41.21s/it]
 22%|██▏       | 1030/4722 [11:13:47<41:20:02, 40.30s/it]
 22%|██▏       | 1031/4722 [11:14:25<40:51:15, 39.85s/it]
 22%|██▏       | 1032/4722 [11:15:05<40:45:15, 39.76s/it]
 22%|██▏       | 1033/4722 [11:15:48<41:48:07, 40.79s/it]
 22%|██▏       | 1034/4722 [11:16:23<39:48:20, 38.86s/it]
 22%|██▏       | 1035/4722 [11:17:00<39:23:25, 38.46s/it]
 22%|██▏       | 1036/4722 [11:17:32<37:18:48, 36.44s/it]
 22%|██▏       | 1037/4722 [11:18:13<38:36:48, 37.72s/it]
 22%|██▏       | 1038/4722 [11:18:53<39:23:23, 38.49s/it]
 22%|██▏       | 1039/4722 [11:19:33<39:54:19, 39.01s/it]
 22%|██▏       | 1040/4722 [11:20:15<40:49:29, 39.92s/it]
                                                         
{'loss': 0.6675, 'learning_rate': 0.0001, 'global_step': 1040, 'epoch': 0.66}
+
 22%|██▏       | 1040/4722 [11:20:15<40:49:29, 39.92s/it]
 22%|██▏       | 1041/4722 [11:20:50<39:17:59, 38.44s/it]
 22%|██▏       | 1042/4722 [11:21:37<41:59:22, 41.08s/it]
 22%|██▏       | 1043/4722 [11:22:17<41:29:17, 40.60s/it]
 22%|██▏       | 1044/4722 [11:22:55<40:37:15, 39.76s/it]
 22%|██▏       | 1045/4722 [11:23:30<39:11:19, 38.37s/it]
 22%|██▏       | 1046/4722 [11:24:15<41:11:00, 40.33s/it]
 22%|██▏       | 1047/4722 [11:24:54<40:54:50, 40.08s/it]
 22%|██▏       | 1048/4722 [11:25:21<36:59:15, 36.24s/it]
 22%|██▏       | 1049/4722 [11:25:54<35:53:29, 35.18s/it]
 22%|██▏       | 1050/4722 [11:26:31<36:32:02, 35.82s/it]
 22%|██▏       | 1051/4722 [11:27:09<36:58:05, 36.25s/it]
 22%|██▏       | 1052/4722 [11:27:46<37:23:00, 36.67s/it]
 22%|██▏       | 1053/4722 [11:28:21<36:53:49, 36.20s/it]
 22%|██▏       | 1054/4722 [11:29:07<39:43:17, 38.99s/it]
 22%|██▏       | 1055/4722 [11:29:42<38:27:27, 37.75s/it]
 22%|██▏       | 1056/4722 [11:30:16<37:29:40, 36.82s/it]
 22%|██▏       | 1057/4722 [11:30:52<37:09:17, 36.50s/it]
 22%|██▏       | 1058/4722 [11:31:36<39:27:05, 38.76s/it]
 22%|██▏       | 1059/4722 [11:32:16<39:40:57, 39.00s/it]
 22%|██▏       | 1060/4722 [11:32:49<37:59:26, 37.35s/it]
                                                         
{'loss': 0.6329, 'learning_rate': 0.0001, 'global_step': 1060, 'epoch': 0.67}
+
 22%|██▏       | 1060/4722 [11:32:49<37:59:26, 37.35s/it]
 22%|██▏       | 1061/4722 [11:33:29<38:49:37, 38.18s/it]
 22%|██▏       | 1062/4722 [11:34:11<39:50:28, 39.19s/it]
 23%|██▎       | 1063/4722 [11:34:54<40:54:04, 40.24s/it]
 23%|██▎       | 1064/4722 [11:35:37<41:59:06, 41.32s/it]
 23%|██▎       | 1065/4722 [11:36:09<39:00:25, 38.40s/it]
 23%|██▎       | 1066/4722 [11:36:43<37:41:33, 37.12s/it]
 23%|██▎       | 1067/4722 [11:37:23<38:28:55, 37.90s/it]
 23%|██▎       | 1068/4722 [11:37:58<37:40:29, 37.12s/it]
 23%|██▎       | 1069/4722 [11:38:36<37:47:34, 37.24s/it]
 23%|██▎       | 1070/4722 [11:39:29<42:47:23, 42.18s/it]
 23%|██▎       | 1071/4722 [11:40:04<40:20:52, 39.78s/it]
 23%|██▎       | 1072/4722 [11:40:44<40:26:23, 39.89s/it]
 23%|██▎       | 1073/4722 [11:41:22<39:57:10, 39.42s/it]
 23%|██▎       | 1074/4722 [11:41:58<38:50:15, 38.33s/it]
 23%|██▎       | 1075/4722 [11:42:34<38:02:02, 37.54s/it]
 23%|██▎       | 1076/4722 [11:43:11<37:55:16, 37.44s/it]
 23%|██▎       | 1077/4722 [11:43:46<37:09:08, 36.69s/it]
 23%|██▎       | 1078/4722 [11:44:32<39:57:39, 39.48s/it]
 23%|██▎       | 1079/4722 [11:45:09<39:15:13, 38.79s/it]
 23%|██▎       | 1080/4722 [11:45:47<39:06:07, 38.65s/it]
                                                         
{'loss': 0.6878, 'learning_rate': 0.0001, 'global_step': 1080, 'epoch': 0.69}
+
 23%|██▎       | 1080/4722 [11:45:47<39:06:07, 38.65s/it]
 23%|██▎       | 1081/4722 [11:46:25<38:52:27, 38.44s/it]
 23%|██▎       | 1082/4722 [11:47:05<39:11:33, 38.76s/it]
 23%|██▎       | 1083/4722 [11:47:42<38:45:12, 38.34s/it]
 23%|██▎       | 1084/4722 [11:48:20<38:38:46, 38.24s/it]
 23%|██▎       | 1085/4722 [11:48:56<37:59:27, 37.60s/it]
 23%|██▎       | 1086/4722 [11:49:38<39:07:38, 38.74s/it]
 23%|██▎       | 1087/4722 [11:50:15<38:44:19, 38.37s/it]
 23%|██▎       | 1088/4722 [11:50:50<37:36:18, 37.25s/it]
 23%|██▎       | 1089/4722 [11:51:33<39:29:46, 39.14s/it]
 23%|██▎       | 1090/4722 [11:52:14<39:51:47, 39.51s/it]
 23%|██▎       | 1091/4722 [11:52:51<39:04:53, 38.75s/it]
 23%|██▎       | 1092/4722 [11:53:30<39:15:30, 38.93s/it]
 23%|██▎       | 1093/4722 [11:54:09<39:12:41, 38.90s/it]
 23%|██▎       | 1094/4722 [11:54:50<39:51:54, 39.56s/it]
 23%|██▎       | 1095/4722 [11:55:31<40:14:38, 39.94s/it]
 23%|██▎       | 1096/4722 [11:56:06<38:44:13, 38.46s/it]
 23%|██▎       | 1097/4722 [11:56:41<37:43:57, 37.47s/it]
 23%|██▎       | 1098/4722 [11:57:25<39:44:25, 39.48s/it]
 23%|██▎       | 1099/4722 [11:58:01<38:42:29, 38.46s/it]
 23%|██▎       | 1100/4722 [11:58:44<39:58:21, 39.73s/it]
                                                         
{'loss': 0.6659, 'learning_rate': 0.0001, 'global_step': 1100, 'epoch': 0.7}
+
 23%|██▎       | 1100/4722 [11:58:44<39:58:21, 39.73s/it]
 23%|██▎       | 1101/4722 [11:59:29<41:38:45, 41.40s/it]
 23%|██▎       | 1102/4722 [12:00:07<40:41:22, 40.46s/it]
 23%|██▎       | 1103/4722 [12:00:50<41:16:32, 41.06s/it]
 23%|██▎       | 1104/4722 [12:01:29<40:47:55, 40.60s/it]
 23%|██▎       | 1105/4722 [12:02:10<40:52:49, 40.69s/it]
 23%|██▎       | 1106/4722 [12:02:47<39:35:54, 39.42s/it]
 23%|██▎       | 1107/4722 [12:03:26<39:27:25, 39.29s/it]
 23%|██▎       | 1108/4722 [12:04:02<38:28:35, 38.33s/it]
 23%|██▎       | 1109/4722 [12:04:39<38:04:40, 37.94s/it]
 24%|██▎       | 1110/4722 [12:05:17<38:04:36, 37.95s/it]
 24%|██▎       | 1111/4722 [12:05:58<38:54:24, 38.79s/it]
 24%|██▎       | 1112/4722 [12:06:36<38:45:04, 38.64s/it]
 24%|██▎       | 1113/4722 [12:07:13<38:20:10, 38.24s/it]
 24%|██▎       | 1114/4722 [12:07:48<37:18:16, 37.22s/it]
 24%|██▎       | 1115/4722 [12:08:38<41:05:46, 41.02s/it]
 24%|██▎       | 1116/4722 [12:09:26<43:14:21, 43.17s/it]
 24%|██▎       | 1117/4722 [12:10:03<41:22:17, 41.31s/it]
 24%|██▎       | 1118/4722 [12:10:46<41:43:25, 41.68s/it]
 24%|██▎       | 1119/4722 [12:11:26<41:14:03, 41.20s/it]
 24%|██▎       | 1120/4722 [12:12:07<41:23:54, 41.38s/it]
                                                         
{'loss': 0.6665, 'learning_rate': 0.0001, 'global_step': 1120, 'epoch': 0.71}
+
 24%|██▎       | 1120/4722 [12:12:07<41:23:54, 41.38s/it]
 24%|██▎       | 1121/4722 [12:12:43<39:35:22, 39.58s/it]
 24%|██▍       | 1122/4722 [12:13:21<39:04:14, 39.07s/it]
 24%|██▍       | 1123/4722 [12:13:57<38:13:00, 38.23s/it]
 24%|██▍       | 1124/4722 [12:14:40<39:31:18, 39.54s/it]
 24%|██▍       | 1125/4722 [12:15:17<38:47:23, 38.82s/it]
 24%|██▍       | 1126/4722 [12:15:49<36:43:00, 36.76s/it]
 24%|██▍       | 1127/4722 [12:16:30<38:00:06, 38.05s/it]
 24%|██▍       | 1128/4722 [12:17:10<38:40:56, 38.75s/it]
 24%|██▍       | 1129/4722 [12:17:54<40:07:21, 40.20s/it]
 24%|██▍       | 1130/4722 [12:18:36<40:38:57, 40.74s/it]
 24%|██▍       | 1131/4722 [12:19:12<39:11:44, 39.29s/it]
 24%|██▍       | 1132/4722 [12:19:51<39:20:07, 39.44s/it]
 24%|██▍       | 1133/4722 [12:20:27<38:05:00, 38.20s/it]
 24%|██▍       | 1134/4722 [12:21:07<38:36:23, 38.74s/it]
 24%|██▍       | 1135/4722 [12:21:51<40:17:58, 40.45s/it]
 24%|██▍       | 1136/4722 [12:22:33<40:38:00, 40.79s/it]
 24%|██▍       | 1137/4722 [12:23:08<38:59:56, 39.16s/it]
 24%|██▍       | 1138/4722 [12:23:45<38:15:15, 38.43s/it]
 24%|██▍       | 1139/4722 [12:24:23<38:13:08, 38.40s/it]
 24%|██▍       | 1140/4722 [12:25:06<39:32:21, 39.74s/it]
                                                         
{'loss': 0.6391, 'learning_rate': 0.0001, 'global_step': 1140, 'epoch': 0.72}
+
 24%|██▍       | 1140/4722 [12:25:06<39:32:21, 39.74s/it]
 24%|██▍       | 1141/4722 [12:25:43<38:42:51, 38.92s/it]
 24%|██▍       | 1142/4722 [12:26:19<37:48:10, 38.01s/it]
 24%|██▍       | 1143/4722 [12:26:54<37:02:59, 37.27s/it]
 24%|██▍       | 1144/4722 [12:27:31<36:47:48, 37.02s/it]
 24%|██▍       | 1145/4722 [12:28:11<37:42:10, 37.95s/it]
 24%|██▍       | 1146/4722 [12:28:51<38:27:26, 38.72s/it]
 24%|██▍       | 1147/4722 [12:29:29<38:10:21, 38.44s/it]
 24%|██▍       | 1148/4722 [12:30:04<37:06:29, 37.38s/it]
 24%|██▍       | 1149/4722 [12:30:45<38:05:21, 38.38s/it]
 24%|██▍       | 1150/4722 [12:31:22<37:46:05, 38.06s/it]
 24%|██▍       | 1151/4722 [12:32:02<38:13:41, 38.54s/it]
 24%|██▍       | 1152/4722 [12:32:40<38:10:41, 38.50s/it]
 24%|██▍       | 1153/4722 [12:33:13<36:34:17, 36.89s/it]
 24%|██▍       | 1154/4722 [12:33:52<37:05:34, 37.43s/it]
 24%|██▍       | 1155/4722 [12:34:32<37:44:50, 38.10s/it]
 24%|██▍       | 1156/4722 [12:35:22<41:19:21, 41.72s/it]
 25%|██▍       | 1157/4722 [12:35:57<39:23:30, 39.78s/it]
 25%|██▍       | 1158/4722 [12:36:31<37:35:22, 37.97s/it]
 25%|██▍       | 1159/4722 [12:37:11<38:05:18, 38.48s/it]
 25%|██▍       | 1160/4722 [12:37:50<38:18:00, 38.71s/it]
                                                         
{'loss': 0.64, 'learning_rate': 0.0001, 'global_step': 1160, 'epoch': 0.74}
+
 25%|██▍       | 1160/4722 [12:37:50<38:18:00, 38.71s/it]
 25%|██▍       | 1161/4722 [12:38:31<38:53:12, 39.31s/it]
 25%|██▍       | 1162/4722 [12:39:13<39:48:01, 40.25s/it]
 25%|██▍       | 1163/4722 [12:39:53<39:35:02, 40.04s/it]
 25%|██▍       | 1164/4722 [12:40:27<37:57:30, 38.41s/it]
 25%|██▍       | 1165/4722 [12:41:13<40:12:51, 40.70s/it]
 25%|██▍       | 1166/4722 [12:41:49<38:39:40, 39.14s/it]
 25%|██▍       | 1167/4722 [12:42:26<38:02:19, 38.52s/it]
 25%|██▍       | 1168/4722 [12:43:09<39:24:15, 39.91s/it]
 25%|██▍       | 1169/4722 [12:43:55<41:13:58, 41.78s/it]
 25%|██▍       | 1170/4722 [12:44:36<40:54:46, 41.47s/it]
 25%|██▍       | 1171/4722 [12:45:11<39:01:32, 39.56s/it]
 25%|██▍       | 1172/4722 [12:45:50<38:55:44, 39.48s/it]
 25%|██▍       | 1173/4722 [12:46:28<38:30:12, 39.06s/it]
 25%|██▍       | 1174/4722 [12:47:09<38:56:27, 39.51s/it]
 25%|██▍       | 1175/4722 [12:47:46<38:18:53, 38.89s/it]
 25%|██▍       | 1176/4722 [12:48:17<35:55:25, 36.47s/it]
 25%|██▍       | 1177/4722 [12:48:56<36:30:49, 37.08s/it]
 25%|██▍       | 1178/4722 [12:49:33<36:36:40, 37.19s/it]
 25%|██▍       | 1179/4722 [12:50:22<40:12:00, 40.85s/it]
 25%|██▍       | 1180/4722 [12:50:54<37:34:27, 38.19s/it]
                                                         
{'loss': 0.6322, 'learning_rate': 0.0001, 'global_step': 1180, 'epoch': 0.75}
+
 25%|██▍       | 1180/4722 [12:50:54<37:34:27, 38.19s/it]
 25%|██▌       | 1181/4722 [12:51:31<37:05:09, 37.70s/it]
 25%|██▌       | 1182/4722 [12:52:15<38:52:05, 39.53s/it]
 25%|██▌       | 1183/4722 [12:52:58<39:53:41, 40.58s/it]
 25%|██▌       | 1184/4722 [12:53:32<37:52:59, 38.55s/it]
 25%|██▌       | 1185/4722 [12:54:15<39:13:59, 39.93s/it]
 25%|██▌       | 1186/4722 [12:54:48<37:07:49, 37.80s/it]
 25%|██▌       | 1187/4722 [12:55:25<36:51:47, 37.54s/it]
 25%|██▌       | 1188/4722 [12:56:05<37:44:45, 38.45s/it]
 25%|██▌       | 1189/4722 [12:56:45<38:11:52, 38.92s/it]
 25%|██▌       | 1190/4722 [12:57:30<39:51:38, 40.63s/it]
 25%|██▌       | 1191/4722 [12:58:07<38:50:29, 39.60s/it]
 25%|██▌       | 1192/4722 [12:58:46<38:48:28, 39.58s/it]
 25%|██▌       | 1193/4722 [12:59:23<37:59:44, 38.76s/it]
 25%|██▌       | 1194/4722 [13:00:01<37:41:07, 38.45s/it]
 25%|██▌       | 1195/4722 [13:00:37<36:51:33, 37.62s/it]
 25%|██▌       | 1196/4722 [13:01:18<38:00:06, 38.80s/it]
 25%|██▌       | 1197/4722 [13:01:54<37:03:39, 37.85s/it]
 25%|██▌       | 1198/4722 [13:02:38<38:44:48, 39.58s/it]
 25%|██▌       | 1199/4722 [13:03:16<38:19:26, 39.16s/it]
 25%|██▌       | 1200/4722 [13:03:52<37:21:05, 38.18s/it]
                                                         
{'loss': 0.671, 'learning_rate': 0.0001, 'global_step': 1200, 'epoch': 0.76}
+
 25%|██▌       | 1200/4722 [13:03:52<37:21:05, 38.18s/it]
 25%|██▌       | 1201/4722 [13:04:38<39:48:43, 40.71s/it]
 25%|██▌       | 1202/4722 [13:05:18<39:37:23, 40.52s/it]
 25%|██▌       | 1203/4722 [13:06:04<41:10:45, 42.13s/it]
 25%|██▌       | 1204/4722 [13:06:43<40:16:47, 41.22s/it]
 26%|██▌       | 1205/4722 [13:07:23<39:48:35, 40.75s/it]
 26%|██▌       | 1206/4722 [13:08:13<42:30:11, 43.52s/it]
 26%|██▌       | 1207/4722 [13:08:51<40:52:56, 41.87s/it]
 26%|██▌       | 1208/4722 [13:09:32<40:29:25, 41.48s/it]
 26%|██▌       | 1209/4722 [13:10:11<39:55:58, 40.92s/it]
 26%|██▌       | 1210/4722 [13:10:59<42:00:05, 43.05s/it]
 26%|██▌       | 1211/4722 [13:11:33<39:18:49, 40.31s/it]
 26%|██▌       | 1212/4722 [13:12:21<41:37:32, 42.69s/it]
 26%|██▌       | 1213/4722 [13:13:01<40:39:52, 41.72s/it]
 26%|██▌       | 1214/4722 [13:13:38<39:26:44, 40.48s/it]
 26%|██▌       | 1215/4722 [13:14:15<38:20:14, 39.35s/it]
 26%|██▌       | 1216/4722 [13:14:51<37:21:36, 38.36s/it]
 26%|██▌       | 1217/4722 [13:15:22<35:11:25, 36.14s/it]
 26%|██▌       | 1218/4722 [13:16:06<37:23:39, 38.42s/it]
 26%|██▌       | 1219/4722 [13:16:42<36:46:42, 37.80s/it]
 26%|██▌       | 1220/4722 [13:17:19<36:24:37, 37.43s/it]
                                                         
{'loss': 0.6406, 'learning_rate': 0.0001, 'global_step': 1220, 'epoch': 0.77}
+
 26%|██▌       | 1220/4722 [13:17:19<36:24:37, 37.43s/it]
 26%|██▌       | 1221/4722 [13:17:59<37:19:43, 38.38s/it]
 26%|██▌       | 1222/4722 [13:18:33<35:47:25, 36.81s/it]
 26%|██▌       | 1223/4722 [13:19:07<35:07:13, 36.13s/it]
 26%|██▌       | 1224/4722 [13:19:44<35:26:42, 36.48s/it]
 26%|██▌       | 1225/4722 [13:20:20<35:14:43, 36.28s/it]
 26%|██▌       | 1226/4722 [13:21:01<36:34:41, 37.67s/it]
 26%|██▌       | 1227/4722 [13:21:39<36:39:36, 37.76s/it]
 26%|██▌       | 1228/4722 [13:22:13<35:34:33, 36.66s/it]
 26%|██▌       | 1229/4722 [13:22:54<36:56:08, 38.07s/it]
 26%|██▌       | 1230/4722 [13:23:30<36:17:29, 37.41s/it]
 26%|██▌       | 1231/4722 [13:24:11<37:12:56, 38.38s/it]
 26%|██▌       | 1232/4722 [13:24:53<38:07:54, 39.33s/it]
 26%|██▌       | 1233/4722 [13:25:40<40:36:40, 41.90s/it]
 26%|██▌       | 1234/4722 [13:26:25<41:20:12, 42.66s/it]
 26%|██▌       | 1235/4722 [13:27:09<41:39:16, 43.00s/it]
 26%|██▌       | 1236/4722 [13:27:44<39:21:44, 40.65s/it]
 26%|██▌       | 1237/4722 [13:28:24<39:06:06, 40.39s/it]
 26%|██▌       | 1238/4722 [13:29:02<38:24:38, 39.69s/it]
 26%|██▌       | 1239/4722 [13:29:45<39:30:58, 40.84s/it]
 26%|██▋       | 1240/4722 [13:30:28<40:07:01, 41.48s/it]
                                                         
{'loss': 0.6248, 'learning_rate': 0.0001, 'global_step': 1240, 'epoch': 0.79}
+
 26%|██▋       | 1240/4722 [13:30:28<40:07:01, 41.48s/it]
 26%|██▋       | 1241/4722 [13:31:04<38:28:35, 39.79s/it]
 26%|██▋       | 1242/4722 [13:31:40<37:24:50, 38.70s/it]
 26%|██▋       | 1243/4722 [13:32:24<38:53:28, 40.24s/it]
 26%|██▋       | 1244/4722 [13:33:04<38:49:35, 40.19s/it]
 26%|██▋       | 1245/4722 [13:33:45<39:04:45, 40.46s/it]
 26%|██▋       | 1246/4722 [13:34:21<37:41:08, 39.03s/it]
 26%|██▋       | 1247/4722 [13:35:00<37:33:19, 38.91s/it]
 26%|██▋       | 1248/4722 [13:35:39<37:35:07, 38.95s/it]
 26%|██▋       | 1249/4722 [13:36:18<37:50:06, 39.22s/it]
 26%|██▋       | 1250/4722 [13:37:04<39:31:40, 40.99s/it]
 26%|██▋       | 1251/4722 [13:37:43<39:10:05, 40.62s/it]
 27%|██▋       | 1252/4722 [13:38:17<37:14:58, 38.65s/it]
 27%|██▋       | 1253/4722 [13:38:55<37:05:52, 38.50s/it]
 27%|██▋       | 1254/4722 [13:39:35<37:28:01, 38.89s/it]
 27%|██▋       | 1255/4722 [13:40:12<36:57:14, 38.37s/it]
 27%|██▋       | 1256/4722 [13:41:00<39:41:12, 41.22s/it]
 27%|██▋       | 1257/4722 [13:41:39<38:52:33, 40.39s/it]
 27%|██▋       | 1258/4722 [13:42:13<37:08:37, 38.60s/it]
 27%|██▋       | 1259/4722 [13:42:54<37:48:23, 39.30s/it]
 27%|██▋       | 1260/4722 [13:43:32<37:14:21, 38.72s/it]
                                                         
{'loss': 0.6691, 'learning_rate': 0.0001, 'global_step': 1260, 'epoch': 0.8}
+
 27%|██▋       | 1260/4722 [13:43:32<37:14:21, 38.72s/it]
 27%|██▋       | 1261/4722 [13:44:09<36:50:00, 38.31s/it]
 27%|██▋       | 1262/4722 [13:44:45<36:04:30, 37.53s/it]
 27%|██▋       | 1263/4722 [13:45:23<36:20:03, 37.82s/it]
 27%|██▋       | 1264/4722 [13:46:03<36:54:21, 38.42s/it]
 27%|██▋       | 1265/4722 [13:46:40<36:23:26, 37.90s/it]
 27%|██▋       | 1266/4722 [13:47:15<35:32:38, 37.03s/it]
 27%|██▋       | 1267/4722 [13:47:52<35:42:13, 37.20s/it]
 27%|██▋       | 1268/4722 [13:48:28<35:12:02, 36.69s/it]
 27%|██▋       | 1269/4722 [13:49:11<37:06:48, 38.69s/it]
 27%|██▋       | 1270/4722 [13:49:53<38:06:38, 39.74s/it]
 27%|██▋       | 1271/4722 [13:50:32<37:57:19, 39.59s/it]
 27%|██▋       | 1272/4722 [13:51:12<37:53:17, 39.54s/it]
 27%|██▋       | 1273/4722 [13:51:50<37:35:56, 39.25s/it]
 27%|██▋       | 1274/4722 [13:52:27<36:45:01, 38.37s/it]
 27%|██▋       | 1275/4722 [13:53:03<36:04:09, 37.67s/it]
 27%|██▋       | 1276/4722 [13:53:42<36:29:43, 38.13s/it]
 27%|██▋       | 1277/4722 [13:54:22<37:02:44, 38.71s/it]
 27%|██▋       | 1278/4722 [13:54:55<35:23:22, 36.99s/it]
 27%|██▋       | 1279/4722 [13:55:31<35:08:39, 36.75s/it]
 27%|██▋       | 1280/4722 [13:56:11<36:06:43, 37.77s/it]
                                                         
{'loss': 0.6671, 'learning_rate': 0.0001, 'global_step': 1280, 'epoch': 0.81}
+
 27%|██▋       | 1280/4722 [13:56:11<36:06:43, 37.77s/it]
 27%|██▋       | 1281/4722 [13:56:51<36:34:38, 38.27s/it]
 27%|██▋       | 1282/4722 [13:57:26<35:34:16, 37.23s/it]
 27%|██▋       | 1283/4722 [13:58:02<35:13:12, 36.87s/it]
 27%|██▋       | 1284/4722 [13:58:43<36:37:14, 38.35s/it]
 27%|██▋       | 1285/4722 [13:59:16<34:57:56, 36.62s/it]
 27%|██▋       | 1286/4722 [13:59:49<33:50:02, 35.45s/it]
 27%|██▋       | 1287/4722 [14:00:31<35:45:07, 37.47s/it]
 27%|██▋       | 1288/4722 [14:01:10<36:10:31, 37.92s/it]
 27%|██▋       | 1289/4722 [14:01:46<35:34:45, 37.31s/it]
 27%|██▋       | 1290/4722 [14:02:27<36:46:49, 38.58s/it]
 27%|██▋       | 1291/4722 [14:03:05<36:33:43, 38.36s/it]
 27%|██▋       | 1292/4722 [14:03:47<37:35:02, 39.45s/it]
 27%|██▋       | 1293/4722 [14:04:22<36:10:13, 37.97s/it]
 27%|██▋       | 1294/4722 [14:04:58<35:46:01, 37.56s/it]
 27%|██▋       | 1295/4722 [14:05:36<35:39:29, 37.46s/it]
 27%|██▋       | 1296/4722 [14:06:14<35:58:50, 37.81s/it]
 27%|██▋       | 1297/4722 [14:06:53<36:17:54, 38.15s/it]
 27%|██▋       | 1298/4722 [14:07:31<36:16:32, 38.14s/it]
 28%|██▊       | 1299/4722 [14:08:05<35:05:16, 36.90s/it]
 28%|██▊       | 1300/4722 [14:08:45<35:51:14, 37.72s/it]
                                                         
{'loss': 0.6482, 'learning_rate': 0.0001, 'global_step': 1300, 'epoch': 0.83}
+
 28%|██▊       | 1300/4722 [14:08:45<35:51:14, 37.72s/it]
 28%|██▊       | 1301/4722 [14:09:26<36:55:28, 38.86s/it]
 28%|██▊       | 1302/4722 [14:10:01<35:39:23, 37.53s/it]
 28%|██▊       | 1303/4722 [14:10:39<35:51:41, 37.76s/it]
 28%|██▊       | 1304/4722 [14:11:10<33:58:09, 35.78s/it]
 28%|██▊       | 1305/4722 [14:11:45<33:34:04, 35.37s/it]
 28%|██▊       | 1306/4722 [14:12:24<34:42:00, 36.57s/it]
 28%|██▊       | 1307/4722 [14:12:57<33:38:04, 35.46s/it]
 28%|██▊       | 1308/4722 [14:13:36<34:33:40, 36.44s/it]
 28%|██▊       | 1309/4722 [14:14:14<35:02:40, 36.96s/it]
 28%|██▊       | 1310/4722 [14:14:48<34:11:09, 36.07s/it]
 28%|██▊       | 1311/4722 [14:15:32<36:23:40, 38.41s/it]
 28%|██▊       | 1312/4722 [14:16:14<37:22:06, 39.45s/it]
 28%|██▊       | 1313/4722 [14:16:59<38:55:47, 41.11s/it]
 28%|██▊       | 1314/4722 [14:17:41<39:14:22, 41.45s/it]
 28%|██▊       | 1315/4722 [14:18:26<40:17:41, 42.58s/it]
 28%|██▊       | 1316/4722 [14:19:09<40:18:45, 42.61s/it]
 28%|██▊       | 1317/4722 [14:19:51<40:11:25, 42.49s/it]
 28%|██▊       | 1318/4722 [14:20:32<39:48:41, 42.10s/it]
 28%|██▊       | 1319/4722 [14:21:03<36:35:14, 38.71s/it]
 28%|██▊       | 1320/4722 [14:21:38<35:40:35, 37.75s/it]
                                                         
{'loss': 0.615, 'learning_rate': 0.0001, 'global_step': 1320, 'epoch': 0.84}
+
 28%|██▊       | 1320/4722 [14:21:38<35:40:35, 37.75s/it]
 28%|██▊       | 1321/4722 [14:22:13<34:39:07, 36.68s/it]
 28%|██▊       | 1322/4722 [14:22:51<35:13:15, 37.29s/it]
 28%|██▊       | 1323/4722 [14:23:30<35:39:59, 37.78s/it]
 28%|██▊       | 1324/4722 [14:24:09<35:49:04, 37.95s/it]
 28%|██▊       | 1325/4722 [14:24:43<34:46:40, 36.86s/it]
 28%|██▊       | 1326/4722 [14:25:15<33:29:39, 35.51s/it]
 28%|██▊       | 1327/4722 [14:25:49<33:02:47, 35.04s/it]
 28%|██▊       | 1328/4722 [14:26:29<34:25:31, 36.51s/it]
 28%|██▊       | 1329/4722 [14:27:08<35:10:49, 37.33s/it]
 28%|██▊       | 1330/4722 [14:27:42<34:01:03, 36.10s/it]
 28%|██▊       | 1331/4722 [14:28:21<34:49:16, 36.97s/it]
 28%|██▊       | 1332/4722 [14:28:58<34:58:53, 37.15s/it]
 28%|██▊       | 1333/4722 [14:29:34<34:43:22, 36.88s/it]
 28%|██▊       | 1334/4722 [14:30:17<36:16:18, 38.54s/it]
 28%|██▊       | 1335/4722 [14:30:54<35:47:16, 38.04s/it]
 28%|██▊       | 1336/4722 [14:31:31<35:39:39, 37.91s/it]
 28%|██▊       | 1337/4722 [14:32:09<35:31:29, 37.78s/it]
 28%|██▊       | 1338/4722 [14:32:48<35:59:06, 38.28s/it]
 28%|██▊       | 1339/4722 [14:33:25<35:29:18, 37.76s/it]
 28%|██▊       | 1340/4722 [14:34:08<36:56:13, 39.32s/it]
                                                         
{'loss': 0.6038, 'learning_rate': 0.0001, 'global_step': 1340, 'epoch': 0.85}
+
 28%|██▊       | 1340/4722 [14:34:08<36:56:13, 39.32s/it]
 28%|██▊       | 1341/4722 [14:34:46<36:41:57, 39.08s/it]
 28%|██▊       | 1342/4722 [14:35:23<35:53:59, 38.24s/it]
 28%|██▊       | 1343/4722 [14:36:08<37:58:50, 40.46s/it]
 28%|██▊       | 1344/4722 [14:36:46<37:07:55, 39.57s/it]
 28%|██▊       | 1345/4722 [14:37:29<38:03:37, 40.57s/it]
 29%|██▊       | 1346/4722 [14:38:06<37:03:00, 39.51s/it]
 29%|██▊       | 1347/4722 [14:38:48<37:57:36, 40.49s/it]
 29%|██▊       | 1348/4722 [14:39:25<36:46:28, 39.24s/it]
 29%|██▊       | 1349/4722 [14:40:06<37:20:31, 39.86s/it]
 29%|██▊       | 1350/4722 [14:40:43<36:29:52, 38.97s/it]
 29%|██▊       | 1351/4722 [14:41:19<35:48:26, 38.24s/it]
 29%|██▊       | 1352/4722 [14:42:04<37:27:48, 40.02s/it]
 29%|██▊       | 1353/4722 [14:42:38<35:55:06, 38.38s/it]
 29%|██▊       | 1354/4722 [14:43:18<36:24:47, 38.92s/it]
 29%|██▊       | 1355/4722 [14:43:52<34:51:25, 37.27s/it]
 29%|██▊       | 1356/4722 [14:44:25<33:37:53, 35.97s/it]
 29%|██▊       | 1357/4722 [14:45:00<33:18:58, 35.64s/it]
 29%|██▉       | 1358/4722 [14:45:31<32:01:34, 34.27s/it]
 29%|██▉       | 1359/4722 [14:46:18<35:45:02, 38.27s/it]
 29%|██▉       | 1360/4722 [14:46:57<35:59:33, 38.54s/it]
                                                         
{'loss': 0.6402, 'learning_rate': 0.0001, 'global_step': 1360, 'epoch': 0.86}
+
 29%|██▉       | 1360/4722 [14:46:57<35:59:33, 38.54s/it]
 29%|██▉       | 1361/4722 [14:47:37<36:19:20, 38.91s/it]
 29%|██▉       | 1362/4722 [14:48:18<36:48:30, 39.44s/it]
 29%|██▉       | 1363/4722 [14:48:52<35:25:40, 37.97s/it]
 29%|██▉       | 1364/4722 [14:49:28<34:52:08, 37.38s/it]
 29%|██▉       | 1365/4722 [14:50:10<35:55:27, 38.52s/it]
 29%|██▉       | 1366/4722 [14:50:46<35:18:32, 37.88s/it]
 29%|██▉       | 1367/4722 [14:51:25<35:38:20, 38.24s/it]
 29%|██▉       | 1368/4722 [14:52:04<35:55:31, 38.56s/it]
 29%|██▉       | 1369/4722 [14:52:50<38:00:28, 40.81s/it]
 29%|██▉       | 1370/4722 [14:53:32<38:13:25, 41.05s/it]
 29%|██▉       | 1371/4722 [14:54:14<38:28:03, 41.33s/it]
 29%|██▉       | 1372/4722 [14:54:48<36:21:16, 39.07s/it]
 29%|██▉       | 1373/4722 [14:55:19<34:10:28, 36.74s/it]
 29%|██▉       | 1374/4722 [14:55:55<34:02:27, 36.60s/it]
 29%|██▉       | 1375/4722 [14:56:30<33:23:33, 35.92s/it]
 29%|██▉       | 1376/4722 [14:57:05<33:08:29, 35.66s/it]
 29%|██▉       | 1377/4722 [14:57:42<33:41:58, 36.27s/it]
 29%|██▉       | 1378/4722 [14:58:21<34:24:11, 37.04s/it]
 29%|██▉       | 1379/4722 [14:58:56<33:51:58, 36.47s/it]
 29%|██▉       | 1380/4722 [14:59:33<33:47:17, 36.40s/it]
                                                         
{'loss': 0.6416, 'learning_rate': 0.0001, 'global_step': 1380, 'epoch': 0.88}
+
 29%|██▉       | 1380/4722 [14:59:33<33:47:17, 36.40s/it]
 29%|██▉       | 1381/4722 [15:00:07<33:19:30, 35.91s/it]
 29%|██▉       | 1382/4722 [15:00:48<34:42:40, 37.41s/it]
 29%|██▉       | 1383/4722 [15:01:25<34:31:39, 37.23s/it]
 29%|██▉       | 1384/4722 [15:02:07<35:43:26, 38.53s/it]
 29%|██▉       | 1385/4722 [15:02:47<36:08:04, 38.98s/it]
 29%|██▉       | 1386/4722 [15:03:21<34:48:31, 37.56s/it]
 29%|██▉       | 1387/4722 [15:03:53<33:11:05, 35.82s/it]
 29%|██▉       | 1388/4722 [15:04:28<32:56:05, 35.56s/it]
 29%|██▉       | 1389/4722 [15:05:02<32:41:16, 35.31s/it]
 29%|██▉       | 1390/4722 [15:05:44<34:17:55, 37.06s/it]
 29%|██▉       | 1391/4722 [15:06:20<34:12:21, 36.97s/it]
 29%|██▉       | 1392/4722 [15:06:57<34:11:34, 36.97s/it]
 30%|██▉       | 1393/4722 [15:07:34<33:58:24, 36.74s/it]
 30%|██▉       | 1394/4722 [15:08:15<35:17:05, 38.17s/it]
 30%|██▉       | 1395/4722 [15:08:56<36:00:43, 38.97s/it]
 30%|██▉       | 1396/4722 [15:09:29<34:23:14, 37.22s/it]
 30%|██▉       | 1397/4722 [15:10:09<35:03:27, 37.96s/it]
 30%|██▉       | 1398/4722 [15:10:47<35:04:31, 37.99s/it]
 30%|██▉       | 1399/4722 [15:11:22<34:19:23, 37.18s/it]
 30%|██▉       | 1400/4722 [15:12:00<34:27:32, 37.34s/it]
                                                         
{'loss': 0.664, 'learning_rate': 0.0001, 'global_step': 1400, 'epoch': 0.89}
+
 30%|██▉       | 1400/4722 [15:12:00<34:27:32, 37.34s/it]
 30%|██▉       | 1401/4722 [15:12:43<36:12:52, 39.26s/it]
 30%|██▉       | 1402/4722 [15:13:23<36:09:44, 39.21s/it]
 30%|██▉       | 1403/4722 [15:14:04<36:52:28, 40.00s/it]
 30%|██▉       | 1404/4722 [15:14:48<37:58:06, 41.20s/it]
 30%|██▉       | 1405/4722 [15:15:27<37:19:27, 40.51s/it]
 30%|██▉       | 1406/4722 [15:16:15<39:13:33, 42.59s/it]
 30%|██▉       | 1407/4722 [15:16:57<39:01:21, 42.38s/it]
 30%|██▉       | 1408/4722 [15:17:31<36:44:13, 39.91s/it]
 30%|██▉       | 1409/4722 [15:18:11<36:43:58, 39.92s/it]
 30%|██▉       | 1410/4722 [15:18:59<39:07:49, 42.53s/it]
 30%|██▉       | 1411/4722 [15:19:39<38:20:00, 41.68s/it]
 30%|██▉       | 1412/4722 [15:20:15<36:48:14, 40.03s/it]
 30%|██▉       | 1413/4722 [15:20:49<35:11:57, 38.29s/it]
 30%|██▉       | 1414/4722 [15:21:23<33:49:36, 36.81s/it]
 30%|██▉       | 1415/4722 [15:21:57<33:12:17, 36.15s/it]
 30%|██▉       | 1416/4722 [15:22:37<34:04:04, 37.10s/it]
 30%|███       | 1417/4722 [15:23:18<35:08:20, 38.28s/it]
 30%|███       | 1418/4722 [15:23:57<35:26:08, 38.61s/it]
 30%|███       | 1419/4722 [15:24:42<37:07:01, 40.45s/it]
 30%|███       | 1420/4722 [15:25:20<36:25:41, 39.72s/it]
                                                         
{'loss': 0.6518, 'learning_rate': 0.0001, 'global_step': 1420, 'epoch': 0.9}
+
 30%|███       | 1420/4722 [15:25:20<36:25:41, 39.72s/it]
 30%|███       | 1421/4722 [15:25:55<35:03:03, 38.23s/it]
 30%|███       | 1422/4722 [15:26:33<35:01:23, 38.21s/it]
 30%|███       | 1423/4722 [15:27:06<33:32:58, 36.61s/it]
 30%|███       | 1424/4722 [15:27:46<34:26:00, 37.59s/it]
 30%|███       | 1425/4722 [15:28:27<35:23:10, 38.64s/it]
 30%|███       | 1426/4722 [15:29:17<38:31:57, 42.09s/it]
 30%|███       | 1427/4722 [15:29:58<38:11:14, 41.72s/it]
 30%|███       | 1428/4722 [15:30:33<36:31:28, 39.92s/it]
 30%|███       | 1429/4722 [15:31:18<37:48:20, 41.33s/it]
 30%|███       | 1430/4722 [15:32:01<38:10:15, 41.74s/it]
 30%|███       | 1431/4722 [15:32:48<39:38:14, 43.36s/it]
 30%|███       | 1432/4722 [15:33:26<38:10:36, 41.77s/it]
 30%|███       | 1433/4722 [15:34:09<38:36:26, 42.26s/it]
 30%|███       | 1434/4722 [15:34:46<37:09:07, 40.68s/it]
 30%|███       | 1435/4722 [15:35:30<37:50:59, 41.45s/it]
 30%|███       | 1436/4722 [15:36:19<39:56:09, 43.75s/it]
 30%|███       | 1437/4722 [15:36:50<36:31:18, 40.02s/it]
 30%|███       | 1438/4722 [15:37:25<35:14:17, 38.63s/it]
 30%|███       | 1439/4722 [15:38:03<34:58:26, 38.35s/it]
 30%|███       | 1440/4722 [15:38:49<36:56:25, 40.52s/it]
                                                         
{'loss': 0.6877, 'learning_rate': 0.0001, 'global_step': 1440, 'epoch': 0.91}
+
 30%|███       | 1440/4722 [15:38:49<36:56:25, 40.52s/it]
 31%|███       | 1441/4722 [15:39:26<36:11:00, 39.70s/it]
 31%|███       | 1442/4722 [15:40:06<36:12:29, 39.74s/it]
 31%|███       | 1443/4722 [15:40:54<38:22:03, 42.12s/it]
 31%|███       | 1444/4722 [15:41:41<39:49:14, 43.73s/it]
 31%|███       | 1445/4722 [15:42:19<37:59:36, 41.74s/it]
 31%|███       | 1446/4722 [15:43:00<37:53:44, 41.64s/it]
 31%|███       | 1447/4722 [15:43:34<35:50:24, 39.40s/it]
 31%|███       | 1448/4722 [15:44:10<34:57:45, 38.44s/it]
 31%|███       | 1449/4722 [15:44:48<34:47:26, 38.27s/it]
 31%|███       | 1450/4722 [15:45:29<35:30:08, 39.06s/it]
 31%|███       | 1451/4722 [15:46:05<34:38:25, 38.12s/it]
 31%|███       | 1452/4722 [15:46:47<35:37:27, 39.22s/it]
 31%|███       | 1453/4722 [15:47:22<34:37:37, 38.13s/it]
 31%|███       | 1454/4722 [15:47:58<33:53:10, 37.33s/it]
 31%|███       | 1455/4722 [15:48:37<34:14:54, 37.74s/it]
 31%|███       | 1456/4722 [15:49:11<33:25:15, 36.84s/it]
 31%|███       | 1457/4722 [15:49:46<32:46:03, 36.13s/it]
 31%|███       | 1458/4722 [15:50:23<33:02:03, 36.43s/it]
 31%|███       | 1459/4722 [15:51:05<34:31:04, 38.08s/it]
 31%|███       | 1460/4722 [15:51:46<35:27:54, 39.14s/it]
                                                         
{'loss': 0.6685, 'learning_rate': 0.0001, 'global_step': 1460, 'epoch': 0.93}
+
 31%|███       | 1460/4722 [15:51:46<35:27:54, 39.14s/it]
 31%|███       | 1461/4722 [15:52:26<35:38:27, 39.35s/it]
 31%|███       | 1462/4722 [15:52:59<33:44:54, 37.27s/it]
 31%|███       | 1463/4722 [15:53:32<32:39:39, 36.08s/it]
 31%|███       | 1464/4722 [15:54:06<32:12:35, 35.59s/it]
 31%|███       | 1465/4722 [15:54:49<34:07:37, 37.72s/it]
 31%|███       | 1466/4722 [15:55:35<36:23:15, 40.23s/it]
 31%|███       | 1467/4722 [15:56:23<38:30:34, 42.59s/it]
 31%|███       | 1468/4722 [15:57:00<36:55:40, 40.85s/it]
 31%|███       | 1469/4722 [15:57:39<36:18:36, 40.18s/it]
 31%|███       | 1470/4722 [15:58:12<34:31:32, 38.22s/it]
 31%|███       | 1471/4722 [15:58:47<33:36:34, 37.22s/it]
 31%|███       | 1472/4722 [15:59:29<34:50:25, 38.59s/it]
 31%|███       | 1473/4722 [16:00:07<34:37:49, 38.37s/it]
 31%|███       | 1474/4722 [16:00:51<36:09:52, 40.08s/it]
 31%|███       | 1475/4722 [16:01:27<35:01:06, 38.83s/it]
 31%|███▏      | 1476/4722 [16:01:59<33:17:38, 36.92s/it]
 31%|███▏      | 1477/4722 [16:02:43<35:01:14, 38.85s/it]
 31%|███▏      | 1478/4722 [16:03:21<34:43:48, 38.54s/it]
 31%|███▏      | 1479/4722 [16:04:05<36:19:00, 40.31s/it]
 31%|███▏      | 1480/4722 [16:04:52<38:12:00, 42.42s/it]
                                                         
{'loss': 0.6522, 'learning_rate': 0.0001, 'global_step': 1480, 'epoch': 0.94}
+
 31%|███▏      | 1480/4722 [16:04:52<38:12:00, 42.42s/it]
 31%|███▏      | 1481/4722 [16:05:33<37:51:02, 42.04s/it]
 31%|███▏      | 1482/4722 [16:06:06<35:13:50, 39.15s/it]
 31%|███▏      | 1483/4722 [16:06:45<35:09:59, 39.09s/it]
 31%|███▏      | 1484/4722 [16:07:20<34:11:53, 38.02s/it]
 31%|███▏      | 1485/4722 [16:07:55<33:23:18, 37.13s/it]
 31%|███▏      | 1486/4722 [16:08:35<34:08:11, 37.98s/it]
 31%|███▏      | 1487/4722 [16:09:10<33:18:24, 37.06s/it]
 32%|███▏      | 1488/4722 [16:09:53<34:42:57, 38.64s/it]
 32%|███▏      | 1489/4722 [16:10:42<37:34:09, 41.83s/it]
 32%|███▏      | 1490/4722 [16:11:20<36:25:36, 40.57s/it]
 32%|███▏      | 1491/4722 [16:12:06<38:03:06, 42.40s/it]
 32%|███▏      | 1492/4722 [16:12:41<36:03:39, 40.19s/it]
 32%|███▏      | 1493/4722 [16:13:23<36:31:30, 40.72s/it]
 32%|███▏      | 1494/4722 [16:14:03<36:09:27, 40.32s/it]
 32%|███▏      | 1495/4722 [16:14:36<34:09:16, 38.10s/it]
 32%|███▏      | 1496/4722 [16:15:11<33:34:10, 37.46s/it]
 32%|███▏      | 1497/4722 [16:15:48<33:24:57, 37.30s/it]
 32%|███▏      | 1498/4722 [16:16:24<33:01:41, 36.88s/it]
 32%|███▏      | 1499/4722 [16:17:05<33:58:16, 37.95s/it]
 32%|███▏      | 1500/4722 [16:17:50<36:00:59, 40.24s/it]
                                                         
{'loss': 0.6292, 'learning_rate': 0.0001, 'global_step': 1500, 'epoch': 0.95}
+
 32%|███▏      | 1500/4722 [16:17:50<36:00:59, 40.24s/it]
 32%|███▏      | 1501/4722 [16:18:40<38:28:13, 43.00s/it]
 32%|███▏      | 1502/4722 [16:19:22<38:11:46, 42.70s/it]
 32%|███▏      | 1503/4722 [16:20:01<37:10:33, 41.58s/it]
 32%|███▏      | 1504/4722 [16:20:39<36:13:30, 40.53s/it]
 32%|███▏      | 1505/4722 [16:21:14<34:46:12, 38.91s/it]
 32%|███▏      | 1506/4722 [16:21:47<33:05:56, 37.05s/it]
 32%|███▏      | 1507/4722 [16:22:22<32:33:38, 36.46s/it]
 32%|███▏      | 1508/4722 [16:22:57<32:11:13, 36.05s/it]
 32%|███▏      | 1509/4722 [16:23:42<34:35:17, 38.75s/it]
 32%|███▏      | 1510/4722 [16:24:23<35:07:17, 39.36s/it]
 32%|███▏      | 1511/4722 [16:25:07<36:33:30, 40.99s/it]
 32%|███▏      | 1512/4722 [16:25:39<34:04:39, 38.22s/it]
 32%|███▏      | 1513/4722 [16:26:24<35:53:44, 40.27s/it]
 32%|███▏      | 1514/4722 [16:27:00<34:45:47, 39.01s/it]
 32%|███▏      | 1515/4722 [16:27:35<33:35:41, 37.71s/it]
 32%|███▏      | 1516/4722 [16:28:09<32:27:31, 36.45s/it]
 32%|███▏      | 1517/4722 [16:28:45<32:27:46, 36.46s/it]
 32%|███▏      | 1518/4722 [16:29:27<33:49:24, 38.00s/it]
 32%|███▏      | 1519/4722 [16:30:03<33:23:54, 37.54s/it]
 32%|███▏      | 1520/4722 [16:30:47<35:10:08, 39.54s/it]
                                                         
{'loss': 0.6319, 'learning_rate': 0.0001, 'global_step': 1520, 'epoch': 0.97}
+
 32%|███▏      | 1520/4722 [16:30:47<35:10:08, 39.54s/it]
 32%|███▏      | 1521/4722 [16:31:33<36:40:34, 41.25s/it]
 32%|███▏      | 1522/4722 [16:32:14<36:37:50, 41.21s/it]
 32%|███▏      | 1523/4722 [16:32:59<37:36:47, 42.33s/it]
 32%|███▏      | 1524/4722 [16:33:41<37:43:43, 42.47s/it]
 32%|███▏      | 1525/4722 [16:34:21<37:03:04, 41.72s/it]
 32%|███▏      | 1526/4722 [16:35:04<37:20:38, 42.06s/it]
 32%|███▏      | 1527/4722 [16:35:40<35:46:36, 40.31s/it]
 32%|███▏      | 1528/4722 [16:36:15<34:16:16, 38.63s/it]
 32%|███▏      | 1529/4722 [16:36:56<34:49:36, 39.27s/it]
 32%|███▏      | 1530/4722 [16:37:35<34:48:18, 39.25s/it]
 32%|███▏      | 1531/4722 [16:38:10<33:45:18, 38.08s/it]
 32%|███▏      | 1532/4722 [16:38:48<33:43:21, 38.06s/it]
 32%|███▏      | 1533/4722 [16:39:28<34:00:09, 38.39s/it]
 32%|███▏      | 1534/4722 [16:40:04<33:33:36, 37.90s/it]
 33%|███▎      | 1535/4722 [16:40:42<33:26:34, 37.78s/it]
 33%|███▎      | 1536/4722 [16:41:27<35:15:32, 39.84s/it]
 33%|███▎      | 1537/4722 [16:42:05<34:54:59, 39.47s/it]
 33%|███▎      | 1538/4722 [16:42:43<34:35:23, 39.11s/it]
 33%|███▎      | 1539/4722 [16:43:23<34:38:35, 39.18s/it]
 33%|███▎      | 1540/4722 [16:43:58<33:36:46, 38.03s/it]
                                                         
{'loss': 0.6283, 'learning_rate': 0.0001, 'global_step': 1540, 'epoch': 0.98}
+
 33%|███▎      | 1540/4722 [16:43:58<33:36:46, 38.03s/it]
 33%|███▎      | 1541/4722 [16:44:39<34:18:45, 38.83s/it]
 33%|███▎      | 1542/4722 [16:45:20<35:02:38, 39.67s/it]
 33%|███▎      | 1543/4722 [16:46:04<36:09:09, 40.94s/it]
 33%|███▎      | 1544/4722 [16:46:48<36:44:04, 41.61s/it]
 33%|███▎      | 1545/4722 [16:47:32<37:25:34, 42.41s/it]
 33%|███▎      | 1546/4722 [16:48:08<35:40:42, 40.44s/it]
 33%|███▎      | 1547/4722 [16:48:48<35:41:32, 40.47s/it]
 33%|███▎      | 1548/4722 [16:49:26<35:04:22, 39.78s/it]
 33%|███▎      | 1549/4722 [16:50:05<34:41:30, 39.36s/it]
 33%|███▎      | 1550/4722 [16:50:43<34:17:31, 38.92s/it]
 33%|███▎      | 1551/4722 [16:51:23<34:39:09, 39.34s/it]
 33%|███▎      | 1552/4722 [16:52:01<34:22:41, 39.04s/it]
 33%|███▎      | 1553/4722 [16:52:40<34:23:22, 39.07s/it]
 33%|███▎      | 1554/4722 [16:53:18<34:04:41, 38.73s/it]
 33%|███▎      | 1555/4722 [16:53:57<34:02:50, 38.70s/it]
 33%|███▎      | 1556/4722 [16:54:36<34:10:48, 38.87s/it]
 33%|███▎      | 1557/4722 [16:55:07<32:03:49, 36.47s/it]
 33%|███▎      | 1558/4722 [16:55:46<32:48:10, 37.32s/it]
 33%|███▎      | 1559/4722 [16:56:30<34:21:07, 39.10s/it]
 33%|███▎      | 1560/4722 [16:57:13<35:22:18, 40.27s/it]
                                                         
{'loss': 0.5989, 'learning_rate': 0.0001, 'global_step': 1560, 'epoch': 0.99}
+
 33%|███▎      | 1560/4722 [16:57:13<35:22:18, 40.27s/it]
 33%|███▎      | 1561/4722 [16:57:56<36:07:21, 41.14s/it]
 33%|███▎      | 1562/4722 [16:58:41<37:04:16, 42.23s/it]
 33%|███▎      | 1563/4722 [16:59:22<36:48:43, 41.95s/it]
 33%|███▎      | 1564/4722 [16:59:56<34:47:17, 39.66s/it]
 33%|███▎      | 1565/4722 [17:00:36<34:52:11, 39.76s/it]
 33%|███▎      | 1566/4722 [17:01:22<36:25:22, 41.55s/it]
 33%|███▎      | 1567/4722 [17:01:55<34:16:02, 39.10s/it]
 33%|███▎      | 1568/4722 [17:02:38<35:11:08, 40.16s/it]
 33%|███▎      | 1569/4722 [17:03:15<34:18:17, 39.17s/it]
 33%|███▎      | 1570/4722 [17:03:56<34:55:27, 39.89s/it]
 33%|███▎      | 1571/4722 [17:04:37<34:59:08, 39.97s/it]
 33%|███▎      | 1572/4722 [17:05:16<34:51:53, 39.85s/it]
 33%|███▎      | 1573/4722 [17:05:52<33:50:40, 38.69s/it]
 33%|███▎      | 1574/4722 [17:06:28<33:07:41, 37.88s/it]
 33%|███▎      | 1575/4722 [17:07:01<31:45:57, 36.34s/it]
 33%|███▎      | 1576/4722 [17:07:34<30:54:25, 35.37s/it]
 33%|███▎      | 1577/4722 [17:08:18<33:06:33, 37.90s/it]
 33%|███▎      | 1578/4722 [17:09:00<34:12:46, 39.18s/it]
 33%|███▎      | 1579/4722 [17:09:39<34:16:56, 39.27s/it]
 33%|███▎      | 1580/4722 [17:10:14<32:59:19, 37.80s/it]
                                                         
{'loss': 0.6319, 'learning_rate': 0.0001, 'global_step': 1580, 'epoch': 1.0}
+
 33%|███▎      | 1580/4722 [17:10:14<32:59:19, 37.80s/it]
 33%|███▎      | 1581/4722 [17:10:54<33:31:24, 38.42s/it]
 34%|███▎      | 1582/4722 [17:11:35<34:13:15, 39.23s/it]
 34%|███▎      | 1583/4722 [17:12:17<34:59:02, 40.12s/it]
 34%|███▎      | 1584/4722 [17:12:49<32:59:37, 37.85s/it]
 34%|███▎      | 1585/4722 [17:13:27<32:49:15, 37.67s/it]
 34%|███▎      | 1586/4722 [17:14:01<31:52:00, 36.58s/it]
 34%|███▎      | 1587/4722 [17:14:42<33:03:15, 37.96s/it]
 34%|███▎      | 1588/4722 [17:15:18<32:39:00, 37.50s/it]
 34%|███▎      | 1589/4722 [17:15:58<33:17:56, 38.26s/it]
 34%|███▎      | 1590/4722 [17:16:42<34:47:31, 39.99s/it]
 34%|███▎      | 1591/4722 [17:17:18<33:38:25, 38.68s/it]
 34%|███▎      | 1592/4722 [17:17:57<33:48:12, 38.88s/it]
 34%|███▎      | 1593/4722 [17:18:33<32:48:12, 37.74s/it]
 34%|███▍      | 1594/4722 [17:19:07<31:49:51, 36.63s/it]
 34%|███▍      | 1595/4722 [17:19:48<32:59:47, 37.99s/it]
 34%|███▍      | 1596/4722 [17:20:25<32:49:22, 37.80s/it]
 34%|███▍      | 1597/4722 [17:21:00<32:11:46, 37.09s/it]
 34%|███▍      | 1598/4722 [17:21:48<34:59:51, 40.33s/it]
 34%|███▍      | 1599/4722 [17:22:29<35:07:02, 40.48s/it]
 34%|███▍      | 1600/4722 [17:23:12<35:38:05, 41.09s/it]
                                                         
{'loss': 0.6013, 'learning_rate': 0.0001, 'global_step': 1600, 'epoch': 1.02}
+
 34%|███▍      | 1600/4722 [17:23:12<35:38:05, 41.09s/it]
 34%|███▍      | 1601/4722 [17:23:57<36:38:13, 42.26s/it]
 34%|███▍      | 1602/4722 [17:24:40<37:00:18, 42.70s/it]
 34%|███▍      | 1603/4722 [17:25:15<35:00:19, 40.40s/it]
 34%|███▍      | 1604/4722 [17:25:56<35:05:06, 40.51s/it]
 34%|███▍      | 1605/4722 [17:26:33<33:59:42, 39.26s/it]
 34%|███▍      | 1606/4722 [17:27:18<35:38:44, 41.18s/it]
 34%|███▍      | 1607/4722 [17:27:50<33:10:12, 38.33s/it]
 34%|███▍      | 1608/4722 [17:28:32<34:06:48, 39.44s/it]
 34%|███▍      | 1609/4722 [17:29:05<32:34:12, 37.67s/it]
 34%|███▍      | 1610/4722 [17:29:41<32:03:53, 37.09s/it]
 34%|███▍      | 1611/4722 [17:30:22<33:07:55, 38.34s/it]
 34%|███▍      | 1612/4722 [17:30:59<32:42:55, 37.87s/it]
 34%|███▍      | 1613/4722 [17:31:38<33:03:06, 38.27s/it]
 34%|███▍      | 1614/4722 [17:32:16<32:46:29, 37.96s/it]
 34%|███▍      | 1615/4722 [17:32:51<32:11:07, 37.29s/it]
 34%|███▍      | 1616/4722 [17:33:33<33:20:08, 38.64s/it]
 34%|███▍      | 1617/4722 [17:34:27<37:07:05, 43.04s/it]
 34%|███▍      | 1618/4722 [17:35:05<35:47:56, 41.52s/it]
 34%|███▍      | 1619/4722 [17:35:50<36:47:25, 42.68s/it]
 34%|███▍      | 1620/4722 [17:36:31<36:19:26, 42.16s/it]
                                                         
{'loss': 0.523, 'learning_rate': 0.0001, 'global_step': 1620, 'epoch': 1.03}
+
 34%|███▍      | 1620/4722 [17:36:31<36:19:26, 42.16s/it]
 34%|███▍      | 1621/4722 [17:37:06<34:28:41, 40.03s/it]
 34%|███▍      | 1622/4722 [17:37:46<34:28:41, 40.04s/it]
 34%|███▍      | 1623/4722 [17:38:23<33:45:36, 39.22s/it]
 34%|███▍      | 1624/4722 [17:39:07<34:48:37, 40.45s/it]
 34%|███▍      | 1625/4722 [17:39:46<34:39:03, 40.28s/it]
 34%|███▍      | 1626/4722 [17:40:27<34:47:57, 40.46s/it]
 34%|███▍      | 1627/4722 [17:41:08<34:57:10, 40.66s/it]
 34%|███▍      | 1628/4722 [17:41:45<33:54:56, 39.46s/it]
 34%|███▍      | 1629/4722 [17:42:14<31:15:34, 36.38s/it]
 35%|███▍      | 1630/4722 [17:42:59<33:23:55, 38.89s/it]
 35%|███▍      | 1631/4722 [17:43:33<32:01:56, 37.31s/it]
 35%|███▍      | 1632/4722 [17:44:11<32:17:18, 37.62s/it]
 35%|███▍      | 1633/4722 [17:44:47<31:46:09, 37.02s/it]
 35%|███▍      | 1634/4722 [17:45:30<33:27:06, 39.00s/it]
 35%|███▍      | 1635/4722 [17:46:15<34:55:26, 40.73s/it]
 35%|███▍      | 1636/4722 [17:46:52<33:53:07, 39.53s/it]
 35%|███▍      | 1637/4722 [17:47:41<36:18:15, 42.36s/it]
 35%|███▍      | 1638/4722 [17:48:19<35:15:57, 41.17s/it]
 35%|███▍      | 1639/4722 [17:48:53<33:21:06, 38.94s/it]
 35%|███▍      | 1640/4722 [17:49:31<33:00:10, 38.55s/it]
                                                         
{'loss': 0.5803, 'learning_rate': 0.0001, 'global_step': 1640, 'epoch': 1.04}
+
 35%|███▍      | 1640/4722 [17:49:31<33:00:10, 38.55s/it]
 35%|███▍      | 1641/4722 [17:50:08<32:36:27, 38.10s/it]
 35%|███▍      | 1642/4722 [17:50:43<31:50:43, 37.22s/it]
 35%|███▍      | 1643/4722 [17:51:22<32:14:21, 37.69s/it]
 35%|███▍      | 1644/4722 [17:51:53<30:35:24, 35.78s/it]
 35%|███▍      | 1645/4722 [17:52:33<31:36:36, 36.98s/it]
 35%|███▍      | 1646/4722 [17:53:26<35:41:19, 41.77s/it]
 35%|███▍      | 1647/4722 [17:54:10<36:17:33, 42.49s/it]
 35%|███▍      | 1648/4722 [17:54:48<35:15:27, 41.29s/it]
 35%|███▍      | 1649/4722 [17:55:34<36:21:50, 42.60s/it]
 35%|███▍      | 1650/4722 [17:56:08<34:14:42, 40.13s/it]
 35%|███▍      | 1651/4722 [17:56:48<34:02:40, 39.91s/it]
 35%|███▍      | 1652/4722 [17:57:34<35:35:23, 41.73s/it]
 35%|███▌      | 1653/4722 [17:58:20<36:38:41, 42.99s/it]
 35%|███▌      | 1654/4722 [17:58:55<34:50:04, 40.88s/it]
 35%|███▌      | 1655/4722 [17:59:33<33:55:35, 39.82s/it]
 35%|███▌      | 1656/4722 [18:00:12<33:37:35, 39.48s/it]
 35%|███▌      | 1657/4722 [18:00:53<34:14:06, 40.21s/it]
 35%|███▌      | 1658/4722 [18:01:33<33:57:09, 39.89s/it]
 35%|███▌      | 1659/4722 [18:02:09<32:57:11, 38.73s/it]
 35%|███▌      | 1660/4722 [18:02:45<32:22:41, 38.07s/it]
                                                         
{'loss': 0.5543, 'learning_rate': 0.0001, 'global_step': 1660, 'epoch': 1.05}
+
 35%|███▌      | 1660/4722 [18:02:45<32:22:41, 38.07s/it]
 35%|███▌      | 1661/4722 [18:03:21<31:51:30, 37.47s/it]
 35%|███▌      | 1662/4722 [18:03:57<31:31:32, 37.09s/it]
 35%|███▌      | 1663/4722 [18:04:39<32:44:15, 38.53s/it]
 35%|███▌      | 1664/4722 [18:05:23<33:59:28, 40.02s/it]
 35%|███▌      | 1665/4722 [18:06:03<33:54:04, 39.92s/it]
 35%|███▌      | 1666/4722 [18:06:40<33:15:31, 39.18s/it]
 35%|███▌      | 1667/4722 [18:07:19<33:09:41, 39.08s/it]
 35%|███▌      | 1668/4722 [18:07:59<33:31:00, 39.51s/it]
 35%|███▌      | 1669/4722 [18:08:37<33:01:19, 38.94s/it]
 35%|███▌      | 1670/4722 [18:09:17<33:24:12, 39.40s/it]
 35%|███▌      | 1671/4722 [18:10:02<34:35:49, 40.82s/it]
 35%|███▌      | 1672/4722 [18:10:41<34:14:06, 40.41s/it]
 35%|███▌      | 1673/4722 [18:11:23<34:33:01, 40.79s/it]
 35%|███▌      | 1674/4722 [18:12:03<34:27:35, 40.70s/it]
 35%|███▌      | 1675/4722 [18:12:36<32:23:49, 38.28s/it]
 35%|███▌      | 1676/4722 [18:13:13<32:12:07, 38.06s/it]
 36%|███▌      | 1677/4722 [18:13:49<31:41:42, 37.47s/it]
 36%|███▌      | 1678/4722 [18:14:30<32:35:04, 38.54s/it]
 36%|███▌      | 1679/4722 [18:15:15<33:58:59, 40.20s/it]
 36%|███▌      | 1680/4722 [18:15:57<34:30:12, 40.83s/it]
                                                         
{'loss': 0.5372, 'learning_rate': 0.0001, 'global_step': 1680, 'epoch': 1.07}
+
 36%|███▌      | 1680/4722 [18:15:57<34:30:12, 40.83s/it]
 36%|███▌      | 1681/4722 [18:16:32<33:01:18, 39.09s/it]
 36%|███▌      | 1682/4722 [18:17:14<33:48:16, 40.03s/it]
 36%|███▌      | 1683/4722 [18:17:57<34:32:13, 40.91s/it]
 36%|███▌      | 1684/4722 [18:18:32<33:07:26, 39.25s/it]
 36%|███▌      | 1685/4722 [18:19:20<35:18:56, 41.86s/it]
 36%|███▌      | 1686/4722 [18:19:50<32:13:55, 38.22s/it]
 36%|███▌      | 1687/4722 [18:20:33<33:29:10, 39.72s/it]
 36%|███▌      | 1688/4722 [18:21:12<33:17:09, 39.50s/it]
 36%|███▌      | 1689/4722 [18:21:51<33:07:40, 39.32s/it]
 36%|███▌      | 1690/4722 [18:22:31<33:06:57, 39.32s/it]
 36%|███▌      | 1691/4722 [18:23:08<32:38:00, 38.76s/it]
 36%|███▌      | 1692/4722 [18:23:43<31:42:53, 37.68s/it]
 36%|███▌      | 1693/4722 [18:24:28<33:28:58, 39.79s/it]
 36%|███▌      | 1694/4722 [18:25:04<32:34:59, 38.74s/it]
 36%|███▌      | 1695/4722 [18:25:50<34:16:06, 40.76s/it]
 36%|███▌      | 1696/4722 [18:26:27<33:21:07, 39.68s/it]
 36%|███▌      | 1697/4722 [18:27:03<32:28:21, 38.65s/it]
 36%|███▌      | 1698/4722 [18:27:37<31:24:16, 37.39s/it]
 36%|███▌      | 1699/4722 [18:28:15<31:19:29, 37.30s/it]
 36%|███▌      | 1700/4722 [18:28:53<31:29:43, 37.52s/it]
                                                         
{'loss': 0.5814, 'learning_rate': 0.0001, 'global_step': 1700, 'epoch': 1.08}
+
 36%|███▌      | 1700/4722 [18:28:53<31:29:43, 37.52s/it]
 36%|███▌      | 1701/4722 [18:29:33<32:07:13, 38.28s/it]
 36%|███▌      | 1702/4722 [18:30:07<31:04:11, 37.04s/it]
 36%|███▌      | 1703/4722 [18:30:47<31:51:18, 37.99s/it]
 36%|███▌      | 1704/4722 [18:31:29<32:45:50, 39.08s/it]
 36%|███▌      | 1705/4722 [18:32:12<33:46:31, 40.30s/it]
 36%|███▌      | 1706/4722 [18:32:52<33:37:09, 40.13s/it]
 36%|███▌      | 1707/4722 [18:33:30<33:13:36, 39.67s/it]
 36%|███▌      | 1708/4722 [18:34:12<33:45:57, 40.33s/it]
 36%|███▌      | 1709/4722 [18:34:55<34:32:08, 41.26s/it]
 36%|███▌      | 1710/4722 [18:35:33<33:43:17, 40.30s/it]
 36%|███▌      | 1711/4722 [18:36:09<32:24:24, 38.75s/it]
 36%|███▋      | 1712/4722 [18:36:43<31:12:23, 37.32s/it]
 36%|███▋      | 1713/4722 [18:37:18<30:36:29, 36.62s/it]
 36%|███▋      | 1714/4722 [18:37:56<31:09:18, 37.29s/it]
 36%|███▋      | 1715/4722 [18:38:35<31:22:54, 37.57s/it]
 36%|███▋      | 1716/4722 [18:39:17<32:27:54, 38.88s/it]
 36%|███▋      | 1717/4722 [18:40:00<33:30:13, 40.14s/it]
 36%|███▋      | 1718/4722 [18:40:38<33:04:52, 39.64s/it]
 36%|███▋      | 1719/4722 [18:41:15<32:17:42, 38.72s/it]
 36%|███▋      | 1720/4722 [18:41:58<33:21:50, 40.01s/it]
                                                         
{'loss': 0.5101, 'learning_rate': 0.0001, 'global_step': 1720, 'epoch': 1.09}
+
 36%|███▋      | 1720/4722 [18:41:58<33:21:50, 40.01s/it]
 36%|███▋      | 1721/4722 [18:42:36<32:48:07, 39.35s/it]
 36%|███▋      | 1722/4722 [18:43:12<32:05:17, 38.51s/it]
 36%|███▋      | 1723/4722 [18:43:51<32:04:03, 38.49s/it]
 37%|███▋      | 1724/4722 [18:44:29<32:01:00, 38.45s/it]
 37%|███▋      | 1725/4722 [18:45:19<34:52:14, 41.89s/it]
 37%|███▋      | 1726/4722 [18:45:56<33:48:48, 40.63s/it]
 37%|███▋      | 1727/4722 [18:46:30<32:04:27, 38.55s/it]
 37%|███▋      | 1728/4722 [18:47:06<31:16:06, 37.60s/it]
 37%|███▋      | 1729/4722 [18:47:46<31:56:49, 38.43s/it]
 37%|███▋      | 1730/4722 [18:48:27<32:43:00, 39.37s/it]
 37%|███▋      | 1731/4722 [18:49:03<31:39:37, 38.11s/it]
 37%|███▋      | 1732/4722 [18:49:41<31:38:41, 38.10s/it]
 37%|███▋      | 1733/4722 [18:50:14<30:29:02, 36.72s/it]
 37%|███▋      | 1734/4722 [18:50:53<30:54:09, 37.23s/it]
 37%|███▋      | 1735/4722 [18:51:24<29:22:04, 35.39s/it]
 37%|███▋      | 1736/4722 [18:51:59<29:15:56, 35.28s/it]
 37%|███▋      | 1737/4722 [18:52:36<29:39:43, 35.77s/it]
 37%|███▋      | 1738/4722 [18:53:09<29:03:01, 35.05s/it]
 37%|███▋      | 1739/4722 [18:53:50<30:35:52, 36.93s/it]
 37%|███▋      | 1740/4722 [18:54:31<31:35:18, 38.13s/it]
                                                         
{'loss': 0.5555, 'learning_rate': 0.0001, 'global_step': 1740, 'epoch': 1.11}
+
 37%|███▋      | 1740/4722 [18:54:31<31:35:18, 38.13s/it]
 37%|███▋      | 1741/4722 [18:55:17<33:21:06, 40.28s/it]
 37%|███▋      | 1742/4722 [18:55:55<32:54:42, 39.76s/it]
 37%|███▋      | 1743/4722 [18:56:36<33:14:03, 40.16s/it]
 37%|███▋      | 1744/4722 [18:57:14<32:40:58, 39.51s/it]
 37%|███▋      | 1745/4722 [18:57:54<32:50:26, 39.71s/it]
 37%|███▋      | 1746/4722 [18:58:37<33:29:09, 40.51s/it]
 37%|███▋      | 1747/4722 [18:59:14<32:45:05, 39.63s/it]
 37%|███▋      | 1748/4722 [18:59:50<31:48:27, 38.50s/it]
 37%|███▋      | 1749/4722 [19:00:28<31:43:06, 38.41s/it]
 37%|███▋      | 1750/4722 [19:01:08<32:02:34, 38.81s/it]
 37%|███▋      | 1751/4722 [19:01:44<31:13:51, 37.84s/it]
 37%|███▋      | 1752/4722 [19:02:19<30:38:36, 37.14s/it]
 37%|███▋      | 1753/4722 [19:03:00<31:37:29, 38.35s/it]
 37%|███▋      | 1754/4722 [19:03:43<32:42:06, 39.67s/it]
 37%|███▋      | 1755/4722 [19:04:20<31:55:32, 38.74s/it]
 37%|███▋      | 1756/4722 [19:05:00<32:24:40, 39.34s/it]
 37%|███▋      | 1757/4722 [19:05:43<33:14:54, 40.37s/it]
 37%|███▋      | 1758/4722 [19:06:22<32:48:33, 39.85s/it]
 37%|███▋      | 1759/4722 [19:06:59<32:00:42, 38.89s/it]
 37%|███▋      | 1760/4722 [19:07:38<32:11:29, 39.13s/it]
                                                         
{'loss': 0.5373, 'learning_rate': 0.0001, 'global_step': 1760, 'epoch': 1.12}
+
 37%|███▋      | 1760/4722 [19:07:38<32:11:29, 39.13s/it]
 37%|███▋      | 1761/4722 [19:08:15<31:33:39, 38.37s/it]
 37%|███▋      | 1762/4722 [19:08:48<30:08:52, 36.67s/it]
 37%|███▋      | 1763/4722 [19:09:19<28:49:07, 35.06s/it]
 37%|███▋      | 1764/4722 [19:09:57<29:40:26, 36.11s/it]
 37%|███▋      | 1765/4722 [19:10:32<29:12:47, 35.57s/it]
 37%|███▋      | 1766/4722 [19:11:07<29:05:24, 35.43s/it]
 37%|███▋      | 1767/4722 [19:11:45<29:38:50, 36.12s/it]
 37%|███▋      | 1768/4722 [19:12:23<30:10:06, 36.77s/it]
 37%|███▋      | 1769/4722 [19:12:59<29:55:20, 36.48s/it]
 37%|███▋      | 1770/4722 [19:13:41<31:23:22, 38.28s/it]
 38%|███▊      | 1771/4722 [19:14:21<31:47:28, 38.78s/it]
 38%|███▊      | 1772/4722 [19:14:59<31:35:38, 38.56s/it]
 38%|███▊      | 1773/4722 [19:15:44<33:04:23, 40.37s/it]
 38%|███▊      | 1774/4722 [19:16:21<32:25:32, 39.60s/it]
 38%|███▊      | 1775/4722 [19:16:57<31:18:25, 38.24s/it]
 38%|███▊      | 1776/4722 [19:17:29<29:46:09, 36.38s/it]
 38%|███▊      | 1777/4722 [19:18:04<29:26:01, 35.98s/it]
 38%|███▊      | 1778/4722 [19:18:36<28:32:59, 34.91s/it]
 38%|███▊      | 1779/4722 [19:19:11<28:30:34, 34.87s/it]
 38%|███▊      | 1780/4722 [19:19:50<29:38:48, 36.28s/it]
                                                         
{'loss': 0.54, 'learning_rate': 0.0001, 'global_step': 1780, 'epoch': 1.13}
+
 38%|███▊      | 1780/4722 [19:19:50<29:38:48, 36.28s/it]
 38%|███▊      | 1781/4722 [19:20:38<32:24:54, 39.68s/it]
 38%|███▊      | 1782/4722 [19:21:15<31:50:17, 38.99s/it]
 38%|███▊      | 1783/4722 [19:21:52<31:16:00, 38.30s/it]
 38%|███▊      | 1784/4722 [19:22:30<31:14:59, 38.29s/it]
 38%|███▊      | 1785/4722 [19:23:14<32:35:03, 39.94s/it]
 38%|███▊      | 1786/4722 [19:23:51<31:41:37, 38.86s/it]
 38%|███▊      | 1787/4722 [19:24:37<33:27:16, 41.03s/it]
 38%|███▊      | 1788/4722 [19:25:18<33:27:19, 41.05s/it]
 38%|███▊      | 1789/4722 [19:25:55<32:32:31, 39.94s/it]
 38%|███▊      | 1790/4722 [19:26:38<33:18:05, 40.89s/it]
 38%|███▊      | 1791/4722 [19:27:16<32:33:32, 39.99s/it]
 38%|███▊      | 1792/4722 [19:27:56<32:27:29, 39.88s/it]
 38%|███▊      | 1793/4722 [19:28:30<31:03:53, 38.18s/it]
 38%|███▊      | 1794/4722 [19:29:06<30:31:21, 37.53s/it]
 38%|███▊      | 1795/4722 [19:29:40<29:40:03, 36.49s/it]
 38%|███▊      | 1796/4722 [19:30:16<29:36:17, 36.42s/it]
 38%|███▊      | 1797/4722 [19:30:52<29:29:49, 36.30s/it]
 38%|███▊      | 1798/4722 [19:31:32<30:13:56, 37.22s/it]
 38%|███▊      | 1799/4722 [19:32:12<30:53:53, 38.05s/it]
 38%|███▊      | 1800/4722 [19:32:55<32:06:37, 39.56s/it]
                                                         
{'loss': 0.5472, 'learning_rate': 0.0001, 'global_step': 1800, 'epoch': 1.14}
+
 38%|███▊      | 1800/4722 [19:32:55<32:06:37, 39.56s/it]
 38%|███▊      | 1801/4722 [19:33:39<33:19:44, 41.08s/it]
 38%|███▊      | 1802/4722 [19:34:15<32:04:13, 39.54s/it]
 38%|███▊      | 1803/4722 [19:34:51<31:12:10, 38.48s/it]
 38%|███▊      | 1804/4722 [19:35:31<31:33:57, 38.94s/it]
 38%|███▊      | 1805/4722 [19:36:08<30:57:59, 38.22s/it]
 38%|███▊      | 1806/4722 [19:36:44<30:34:45, 37.75s/it]
 38%|███▊      | 1807/4722 [19:37:23<30:42:40, 37.93s/it]
 38%|███▊      | 1808/4722 [19:38:01<30:49:29, 38.08s/it]
 38%|███▊      | 1809/4722 [19:38:46<32:23:05, 40.02s/it]
 38%|███▊      | 1810/4722 [19:39:23<31:47:27, 39.30s/it]
 38%|███▊      | 1811/4722 [19:39:57<30:20:36, 37.53s/it]
 38%|███▊      | 1812/4722 [19:40:42<32:05:58, 39.71s/it]
 38%|███▊      | 1813/4722 [19:41:22<32:12:44, 39.86s/it]
 38%|███▊      | 1814/4722 [19:42:03<32:25:40, 40.14s/it]
 38%|███▊      | 1815/4722 [19:42:43<32:24:54, 40.14s/it]
 38%|███▊      | 1816/4722 [19:43:29<33:55:36, 42.03s/it]
 38%|███▊      | 1817/4722 [19:44:05<32:22:18, 40.12s/it]
 39%|███▊      | 1818/4722 [19:44:47<32:50:35, 40.71s/it]
 39%|███▊      | 1819/4722 [19:45:24<31:50:54, 39.50s/it]
 39%|███▊      | 1820/4722 [19:46:00<31:04:31, 38.55s/it]
                                                         
{'loss': 0.567, 'learning_rate': 0.0001, 'global_step': 1820, 'epoch': 1.16}
+
 39%|███▊      | 1820/4722 [19:46:00<31:04:31, 38.55s/it]
 39%|███▊      | 1821/4722 [19:46:48<33:21:13, 41.39s/it]
 39%|███▊      | 1822/4722 [19:47:24<32:06:03, 39.85s/it]
 39%|███▊      | 1823/4722 [19:47:57<30:24:47, 37.77s/it]
 39%|███▊      | 1824/4722 [19:48:31<29:32:34, 36.70s/it]
 39%|███▊      | 1825/4722 [19:49:15<31:09:46, 38.73s/it]
 39%|███▊      | 1826/4722 [19:49:52<30:45:37, 38.24s/it]
 39%|███▊      | 1827/4722 [19:50:34<31:43:34, 39.45s/it]
 39%|███▊      | 1828/4722 [19:51:05<29:39:41, 36.90s/it]
 39%|███▊      | 1829/4722 [19:51:43<29:49:41, 37.12s/it]
 39%|███▉      | 1830/4722 [19:52:17<29:14:32, 36.40s/it]
 39%|███▉      | 1831/4722 [19:53:11<33:14:42, 41.40s/it]
 39%|███▉      | 1832/4722 [19:53:51<32:54:17, 40.99s/it]
 39%|███▉      | 1833/4722 [19:54:31<32:45:43, 40.83s/it]
 39%|███▉      | 1834/4722 [19:55:07<31:32:24, 39.32s/it]
 39%|███▉      | 1835/4722 [19:55:46<31:22:56, 39.13s/it]
 39%|███▉      | 1836/4722 [19:56:29<32:22:36, 40.39s/it]
 39%|███▉      | 1837/4722 [19:57:04<31:06:53, 38.83s/it]
 39%|███▉      | 1838/4722 [19:57:40<30:30:38, 38.09s/it]
 39%|███▉      | 1839/4722 [19:58:22<31:25:21, 39.24s/it]
 39%|███▉      | 1840/4722 [19:59:10<33:26:26, 41.77s/it]
                                                         
{'loss': 0.5342, 'learning_rate': 0.0001, 'global_step': 1840, 'epoch': 1.17}
+
 39%|███▉      | 1840/4722 [19:59:10<33:26:26, 41.77s/it]
 39%|███▉      | 1841/4722 [19:59:42<31:06:29, 38.87s/it]
 39%|███▉      | 1842/4722 [20:00:25<31:57:33, 39.95s/it]
 39%|███▉      | 1843/4722 [20:01:01<31:03:40, 38.84s/it]
 39%|███▉      | 1844/4722 [20:01:45<32:17:35, 40.39s/it]
 39%|███▉      | 1845/4722 [20:02:26<32:28:38, 40.64s/it]
 39%|███▉      | 1846/4722 [20:03:02<31:17:24, 39.17s/it]
 39%|███▉      | 1847/4722 [20:03:40<31:00:27, 38.83s/it]
 39%|███▉      | 1848/4722 [20:04:15<30:01:17, 37.61s/it]
 39%|███▉      | 1849/4722 [20:04:52<30:01:06, 37.61s/it]
 39%|███▉      | 1850/4722 [20:05:09<25:00:11, 31.34s/it]
 39%|███▉      | 1851/4722 [20:05:27<21:43:11, 27.24s/it]
 39%|███▉      | 1852/4722 [20:05:45<19:33:44, 24.54s/it]
 39%|███▉      | 1853/4722 [20:06:06<18:48:03, 23.59s/it]
 39%|███▉      | 1854/4722 [20:06:29<18:35:27, 23.34s/it]
 39%|███▉      | 1855/4722 [20:06:44<16:42:17, 20.98s/it]
 39%|███▉      | 1856/4722 [20:07:01<15:45:59, 19.80s/it]
 39%|███▉      | 1857/4722 [20:07:20<15:27:25, 19.42s/it]
 39%|███▉      | 1858/4722 [20:07:43<16:16:48, 20.46s/it]
 39%|███▉      | 1859/4722 [20:08:00<15:23:25, 19.35s/it]
 39%|███▉      | 1860/4722 [20:08:17<14:55:55, 18.78s/it]
                                                         
{'loss': 0.5612, 'learning_rate': 0.0001, 'global_step': 1860, 'epoch': 1.18}
+
 39%|███▉      | 1860/4722 [20:08:17<14:55:55, 18.78s/it]
 39%|███▉      | 1861/4722 [20:08:35<14:42:12, 18.50s/it]
 39%|███▉      | 1862/4722 [20:08:53<14:41:18, 18.49s/it]
 39%|███▉      | 1863/4722 [20:09:08<13:47:59, 17.38s/it]
 39%|███▉      | 1864/4722 [20:09:25<13:34:31, 17.10s/it]
 39%|███▉      | 1865/4722 [20:09:40<13:13:03, 16.66s/it]
 40%|███▉      | 1866/4722 [20:09:55<12:48:36, 16.15s/it]
 40%|███▉      | 1867/4722 [20:10:12<12:51:50, 16.22s/it]
 40%|███▉      | 1868/4722 [20:10:27<12:45:53, 16.10s/it]
 40%|███▉      | 1869/4722 [20:10:43<12:40:21, 15.99s/it]
 40%|███▉      | 1870/4722 [20:11:01<13:06:32, 16.55s/it]
 40%|███▉      | 1871/4722 [20:11:19<13:30:29, 17.06s/it]
 40%|███▉      | 1872/4722 [20:11:38<13:48:43, 17.45s/it]
 40%|███▉      | 1873/4722 [20:11:55<13:41:57, 17.31s/it]
 40%|███▉      | 1874/4722 [20:12:13<13:50:16, 17.49s/it]
 40%|███▉      | 1875/4722 [20:12:32<14:13:41, 17.99s/it]
 40%|███▉      | 1876/4722 [20:12:50<14:15:36, 18.04s/it]
 40%|███▉      | 1877/4722 [20:13:06<13:51:05, 17.53s/it]
 40%|███▉      | 1878/4722 [20:13:23<13:40:42, 17.31s/it]
 40%|███▉      | 1879/4722 [20:13:43<14:25:58, 18.28s/it]
 40%|███▉      | 1880/4722 [20:14:00<14:02:18, 17.78s/it]
                                                         
{'loss': 0.5275, 'learning_rate': 0.0001, 'global_step': 1880, 'epoch': 1.19}
+
 40%|███▉      | 1880/4722 [20:14:00<14:02:18, 17.78s/it]
 40%|███▉      | 1881/4722 [20:14:17<13:48:52, 17.51s/it]
 40%|███▉      | 1882/4722 [20:14:34<13:48:01, 17.49s/it]
 40%|███▉      | 1883/4722 [20:14:51<13:30:52, 17.14s/it]
 40%|███▉      | 1884/4722 [20:15:05<12:55:12, 16.39s/it]
 40%|███▉      | 1885/4722 [20:15:23<13:10:59, 16.73s/it]
 40%|███▉      | 1886/4722 [20:15:39<12:56:54, 16.44s/it]
 40%|███▉      | 1887/4722 [20:15:57<13:18:47, 16.91s/it]
 40%|███▉      | 1888/4722 [20:16:11<12:46:59, 16.24s/it]
 40%|████      | 1889/4722 [20:16:29<13:12:24, 16.78s/it]
 40%|████      | 1890/4722 [20:16:46<13:08:08, 16.70s/it]
 40%|████      | 1891/4722 [20:17:01<12:44:59, 16.21s/it]
 40%|████      | 1892/4722 [20:17:17<12:36:37, 16.04s/it]
 40%|████      | 1893/4722 [20:17:35<13:06:18, 16.68s/it]
 40%|████      | 1894/4722 [20:17:50<12:47:26, 16.28s/it]
 40%|████      | 1895/4722 [20:18:08<13:09:30, 16.76s/it]
 40%|████      | 1896/4722 [20:18:24<13:02:15, 16.61s/it]
 40%|████      | 1897/4722 [20:18:45<14:05:30, 17.96s/it]
 40%|████      | 1898/4722 [20:19:02<13:41:55, 17.46s/it]
 40%|████      | 1899/4722 [20:19:21<14:09:11, 18.05s/it]
 40%|████      | 1900/4722 [20:19:40<14:16:47, 18.22s/it]
                                                         
{'loss': 0.5375, 'learning_rate': 0.0001, 'global_step': 1900, 'epoch': 1.21}
+
 40%|████      | 1900/4722 [20:19:40<14:16:47, 18.22s/it]
 40%|████      | 1901/4722 [20:19:58<14:18:42, 18.26s/it]
 40%|████      | 1902/4722 [20:20:13<13:27:02, 17.17s/it]
 40%|████      | 1903/4722 [20:20:33<14:04:14, 17.97s/it]
 40%|████      | 1904/4722 [20:20:48<13:34:35, 17.34s/it]
 40%|████      | 1905/4722 [20:21:09<14:22:02, 18.36s/it]
 40%|████      | 1906/4722 [20:21:26<13:57:35, 17.85s/it]
 40%|████      | 1907/4722 [20:21:44<14:04:23, 18.00s/it]
 40%|████      | 1908/4722 [20:22:04<14:23:09, 18.40s/it]
 40%|████      | 1909/4722 [20:22:21<14:09:53, 18.13s/it]
 40%|████      | 1910/4722 [20:22:40<14:16:10, 18.27s/it]
 40%|████      | 1911/4722 [20:22:57<14:03:02, 17.99s/it]
 40%|████      | 1912/4722 [20:23:16<14:10:32, 18.16s/it]
 41%|████      | 1913/4722 [20:23:40<15:34:13, 19.96s/it]
 41%|████      | 1914/4722 [20:23:55<14:34:44, 18.69s/it]
 41%|████      | 1915/4722 [20:24:14<14:26:24, 18.52s/it]
 41%|████      | 1916/4722 [20:24:32<14:21:28, 18.42s/it]
 41%|████      | 1917/4722 [20:24:45<13:13:40, 16.98s/it]
 41%|████      | 1918/4722 [20:25:02<13:02:42, 16.75s/it]
 41%|████      | 1919/4722 [20:25:17<12:41:38, 16.30s/it]
 41%|████      | 1920/4722 [20:25:35<13:06:30, 16.84s/it]
                                                         
{'loss': 0.554, 'learning_rate': 0.0001, 'global_step': 1920, 'epoch': 1.22}
+
 41%|████      | 1920/4722 [20:25:35<13:06:30, 16.84s/it]
 41%|████      | 1921/4722 [20:25:54<13:42:58, 17.63s/it]
 41%|████      | 1922/4722 [20:26:10<13:21:38, 17.18s/it]
 41%|████      | 1923/4722 [20:26:29<13:37:28, 17.52s/it]
 41%|████      | 1924/4722 [20:26:48<14:07:23, 18.17s/it]
 41%|████      | 1925/4722 [20:27:20<17:11:30, 22.13s/it]
 41%|████      | 1926/4722 [20:27:53<19:40:44, 25.34s/it]
 41%|████      | 1927/4722 [20:28:20<20:09:05, 25.96s/it]
 41%|████      | 1928/4722 [20:28:55<22:20:20, 28.78s/it]
 41%|████      | 1929/4722 [20:29:24<22:11:16, 28.60s/it]
 41%|████      | 1930/4722 [20:29:52<22:08:21, 28.55s/it]
 41%|████      | 1931/4722 [20:30:30<24:12:57, 31.24s/it]
 41%|████      | 1932/4722 [20:30:58<23:34:59, 30.43s/it]
 41%|████      | 1933/4722 [20:31:29<23:45:09, 30.66s/it]
 41%|████      | 1934/4722 [20:32:07<25:20:09, 32.71s/it]
 41%|████      | 1935/4722 [20:32:36<24:33:10, 31.72s/it]
 41%|████      | 1936/4722 [20:33:04<23:32:43, 30.42s/it]
 41%|████      | 1937/4722 [20:33:35<23:46:32, 30.73s/it]
 41%|████      | 1938/4722 [20:34:09<24:24:16, 31.56s/it]
 41%|████      | 1939/4722 [20:34:35<23:11:26, 30.00s/it]
 41%|████      | 1940/4722 [20:35:01<22:22:10, 28.95s/it]
                                                         
{'loss': 0.5071, 'learning_rate': 0.0001, 'global_step': 1940, 'epoch': 1.23}
+
 41%|████      | 1940/4722 [20:35:01<22:22:10, 28.95s/it]
 41%|████      | 1941/4722 [20:35:31<22:37:09, 29.28s/it]
 41%|████      | 1942/4722 [20:36:01<22:46:50, 29.50s/it]
 41%|████      | 1943/4722 [20:36:25<21:26:15, 27.77s/it]
 41%|████      | 1944/4722 [20:36:55<21:47:11, 28.23s/it]
 41%|████      | 1945/4722 [20:37:22<21:38:56, 28.06s/it]
 41%|████      | 1946/4722 [20:37:46<20:32:35, 26.64s/it]
 41%|████      | 1947/4722 [20:38:13<20:48:43, 27.00s/it]
 41%|████▏     | 1948/4722 [20:38:41<20:59:51, 27.25s/it]
 41%|████▏     | 1949/4722 [20:39:13<22:01:37, 28.60s/it]
 41%|████▏     | 1950/4722 [20:39:44<22:34:05, 29.31s/it]
 41%|████▏     | 1951/4722 [20:40:15<22:58:09, 29.84s/it]
 41%|████▏     | 1952/4722 [20:40:44<22:46:47, 29.61s/it]
 41%|████▏     | 1953/4722 [20:41:14<22:45:57, 29.60s/it]
 41%|████▏     | 1954/4722 [20:41:44<22:50:45, 29.71s/it]
 41%|████▏     | 1955/4722 [20:42:17<23:41:11, 30.82s/it]
 41%|████▏     | 1956/4722 [20:42:45<23:03:40, 30.01s/it]
 41%|████▏     | 1957/4722 [20:43:11<22:00:28, 28.65s/it]
 41%|████▏     | 1958/4722 [20:43:37<21:24:43, 27.89s/it]
 41%|████▏     | 1959/4722 [20:44:06<21:39:01, 28.21s/it]
 42%|████▏     | 1960/4722 [20:44:30<20:46:11, 27.07s/it]
                                                         
{'loss': 0.5844, 'learning_rate': 0.0001, 'global_step': 1960, 'epoch': 1.24}
+
 42%|████▏     | 1960/4722 [20:44:30<20:46:11, 27.07s/it]
 42%|████▏     | 1961/4722 [20:45:08<23:16:30, 30.35s/it]
 42%|████▏     | 1962/4722 [20:45:33<22:00:54, 28.72s/it]
 42%|████▏     | 1963/4722 [20:45:58<21:06:32, 27.54s/it]
 42%|████▏     | 1964/4722 [20:46:27<21:32:46, 28.12s/it]
 42%|████▏     | 1965/4722 [20:46:54<21:15:06, 27.75s/it]
 42%|████▏     | 1966/4722 [20:47:25<21:58:42, 28.71s/it]
 42%|████▏     | 1967/4722 [20:47:52<21:39:36, 28.30s/it]
 42%|████▏     | 1968/4722 [20:48:28<23:15:09, 30.40s/it]
 42%|████▏     | 1969/4722 [20:48:58<23:15:55, 30.42s/it]
 42%|████▏     | 1970/4722 [20:49:35<24:44:56, 32.38s/it]
 42%|████▏     | 1971/4722 [20:50:05<24:13:04, 31.69s/it]
 42%|████▏     | 1972/4722 [20:50:33<23:13:17, 30.40s/it]
 42%|████▏     | 1973/4722 [20:51:07<24:14:21, 31.74s/it]
 42%|████▏     | 1974/4722 [20:51:36<23:28:37, 30.76s/it]
 42%|████▏     | 1975/4722 [20:52:05<23:11:21, 30.39s/it]
 42%|████▏     | 1976/4722 [20:52:38<23:46:37, 31.17s/it]
 42%|████▏     | 1977/4722 [20:53:03<22:11:18, 29.10s/it]
 42%|████▏     | 1978/4722 [20:53:29<21:37:09, 28.36s/it]
 42%|████▏     | 1979/4722 [20:53:56<21:06:50, 27.71s/it]
 42%|████▏     | 1980/4722 [20:54:21<20:30:10, 26.92s/it]
                                                         
{'loss': 0.4972, 'learning_rate': 0.0001, 'global_step': 1980, 'epoch': 1.26}
+
 42%|████▏     | 1980/4722 [20:54:21<20:30:10, 26.92s/it]
 42%|████▏     | 1981/4722 [20:54:49<20:42:57, 27.21s/it]
 42%|████▏     | 1982/4722 [20:55:17<21:06:12, 27.73s/it]
 42%|████▏     | 1983/4722 [20:55:47<21:37:13, 28.42s/it]
 42%|████▏     | 1984/4722 [20:56:19<22:16:12, 29.28s/it]
 42%|████▏     | 1985/4722 [20:56:48<22:18:37, 29.35s/it]
 42%|████▏     | 1986/4722 [20:57:20<22:51:11, 30.07s/it]
 42%|████▏     | 1987/4722 [20:57:57<24:19:31, 32.02s/it]
 42%|████▏     | 1988/4722 [20:58:30<24:34:46, 32.37s/it]
 42%|████▏     | 1989/4722 [20:58:56<23:08:30, 30.48s/it]
 42%|████▏     | 1990/4722 [20:59:22<22:09:42, 29.20s/it]
 42%|████▏     | 1991/4722 [20:59:48<21:20:00, 28.12s/it]
 42%|████▏     | 1992/4722 [21:00:17<21:39:42, 28.56s/it]
 42%|████▏     | 1993/4722 [21:00:45<21:27:25, 28.31s/it]
 42%|████▏     | 1994/4722 [21:01:16<22:03:31, 29.11s/it]
 42%|████▏     | 1995/4722 [21:01:46<22:21:08, 29.51s/it]
 42%|████▏     | 1996/4722 [21:02:17<22:38:17, 29.90s/it]
 42%|████▏     | 1997/4722 [21:02:51<23:34:16, 31.14s/it]
 42%|████▏     | 1998/4722 [21:03:20<22:58:14, 30.36s/it]
 42%|████▏     | 1999/4722 [21:03:57<24:34:41, 32.49s/it]
 42%|████▏     | 2000/4722 [21:04:30<24:34:22, 32.50s/it]
                                                         
{'loss': 0.514, 'learning_rate': 0.0001, 'global_step': 2000, 'epoch': 1.27}
+
 42%|████▏     | 2000/4722 [21:04:30<24:34:22, 32.50s/it]
 42%|████▏     | 2001/4722 [21:04:54<22:36:34, 29.91s/it]
 42%|████▏     | 2002/4722 [21:05:24<22:35:04, 29.89s/it]
 42%|████▏     | 2003/4722 [21:05:54<22:36:16, 29.93s/it]
 42%|████▏     | 2004/4722 [21:06:22<22:13:50, 29.44s/it]
 42%|████▏     | 2005/4722 [21:06:55<22:59:18, 30.46s/it]
 42%|████▏     | 2006/4722 [21:07:22<22:20:11, 29.61s/it]
 43%|████▎     | 2007/4722 [21:07:46<21:03:13, 27.92s/it]
 43%|████▎     | 2008/4722 [21:08:16<21:27:10, 28.46s/it]
 43%|████▎     | 2009/4722 [21:08:46<21:49:30, 28.96s/it]
 43%|████▎     | 2010/4722 [21:09:18<22:34:54, 29.98s/it]
 43%|████▎     | 2011/4722 [21:09:46<22:00:51, 29.23s/it]
 43%|████▎     | 2012/4722 [21:10:20<23:11:38, 30.81s/it]
 43%|████▎     | 2013/4722 [21:10:50<22:55:07, 30.46s/it]
 43%|████▎     | 2014/4722 [21:11:16<21:54:20, 29.12s/it]
 43%|████▎     | 2015/4722 [21:11:49<22:41:15, 30.17s/it]
 43%|████▎     | 2016/4722 [21:12:22<23:24:16, 31.14s/it]
 43%|████▎     | 2017/4722 [21:12:57<24:09:04, 32.14s/it]
 43%|████▎     | 2018/4722 [21:13:27<23:46:54, 31.66s/it]
 43%|████▎     | 2019/4722 [21:13:54<22:44:16, 30.28s/it]
 43%|████▎     | 2020/4722 [21:14:27<23:21:54, 31.13s/it]
                                                         
{'loss': 0.5176, 'learning_rate': 0.0001, 'global_step': 2020, 'epoch': 1.28}
+
 43%|████▎     | 2020/4722 [21:14:27<23:21:54, 31.13s/it]
 43%|████▎     | 2021/4722 [21:14:55<22:39:35, 30.20s/it]
 43%|████▎     | 2022/4722 [21:15:24<22:20:08, 29.78s/it]
 43%|████▎     | 2023/4722 [21:15:53<22:01:09, 29.37s/it]
 43%|████▎     | 2024/4722 [21:16:21<21:54:50, 29.24s/it]
 43%|████▎     | 2025/4722 [21:16:49<21:31:16, 28.73s/it]
 43%|████▎     | 2026/4722 [21:17:17<21:17:29, 28.43s/it]
 43%|████▎     | 2027/4722 [21:17:47<21:43:47, 29.03s/it]
 43%|████▎     | 2028/4722 [21:18:12<20:45:45, 27.75s/it]
 43%|████▎     | 2029/4722 [21:18:41<21:08:43, 28.27s/it]
 43%|████▎     | 2030/4722 [21:19:08<20:39:07, 27.62s/it]
 43%|████▎     | 2031/4722 [21:19:35<20:41:37, 27.68s/it]
 43%|████▎     | 2032/4722 [21:20:04<20:55:53, 28.01s/it]
 43%|████▎     | 2033/4722 [21:20:34<21:25:54, 28.69s/it]
 43%|████▎     | 2034/4722 [21:21:07<22:11:45, 29.73s/it]
 43%|████▎     | 2035/4722 [21:21:40<23:00:16, 30.82s/it]
 43%|████▎     | 2036/4722 [21:22:10<22:55:45, 30.73s/it]
 43%|████▎     | 2037/4722 [21:22:43<23:23:58, 31.37s/it]
 43%|████▎     | 2038/4722 [21:23:13<23:02:47, 30.91s/it]
 43%|████▎     | 2039/4722 [21:23:42<22:30:38, 30.20s/it]
 43%|████▎     | 2040/4722 [21:24:14<22:55:44, 30.78s/it]
                                                         
{'loss': 0.5269, 'learning_rate': 0.0001, 'global_step': 2040, 'epoch': 1.3}
+
 43%|████▎     | 2040/4722 [21:24:14<22:55:44, 30.78s/it]
 43%|████▎     | 2041/4722 [21:24:40<21:56:36, 29.47s/it]
 43%|████▎     | 2042/4722 [21:25:05<20:48:26, 27.95s/it]
 43%|████▎     | 2043/4722 [21:25:35<21:17:22, 28.61s/it]
 43%|████▎     | 2044/4722 [21:26:07<22:00:50, 29.59s/it]
 43%|████▎     | 2045/4722 [21:26:37<22:10:58, 29.83s/it]
 43%|████▎     | 2046/4722 [21:27:05<21:47:07, 29.31s/it]
 43%|████▎     | 2047/4722 [21:27:31<21:00:46, 28.28s/it]
 43%|████▎     | 2048/4722 [21:28:00<21:10:34, 28.51s/it]
 43%|████▎     | 2049/4722 [21:28:31<21:42:27, 29.24s/it]
 43%|████▎     | 2050/4722 [21:29:06<23:05:09, 31.10s/it]
 43%|████▎     | 2051/4722 [21:29:30<21:28:54, 28.95s/it]
 43%|████▎     | 2052/4722 [21:29:59<21:29:06, 28.97s/it]
 43%|████▎     | 2053/4722 [21:30:33<22:29:29, 30.34s/it]
 43%|████▎     | 2054/4722 [21:31:03<22:24:09, 30.23s/it]
 44%|████▎     | 2055/4722 [21:31:29<21:34:52, 29.13s/it]
 44%|████▎     | 2056/4722 [21:32:00<21:52:38, 29.54s/it]
 44%|████▎     | 2057/4722 [21:32:27<21:22:51, 28.88s/it]
 44%|████▎     | 2058/4722 [21:33:01<22:21:18, 30.21s/it]
 44%|████▎     | 2059/4722 [21:33:28<21:46:53, 29.45s/it]
 44%|████▎     | 2060/4722 [21:34:02<22:36:54, 30.58s/it]
                                                         
{'loss': 0.5243, 'learning_rate': 0.0001, 'global_step': 2060, 'epoch': 1.31}
+
 44%|████▎     | 2060/4722 [21:34:02<22:36:54, 30.58s/it]
 44%|████▎     | 2061/4722 [21:34:30<22:10:01, 29.99s/it]
 44%|████▎     | 2062/4722 [21:34:58<21:40:01, 29.32s/it]
 44%|████▎     | 2063/4722 [21:35:27<21:38:16, 29.30s/it]
 44%|████▎     | 2064/4722 [21:35:55<21:24:34, 29.00s/it]
 44%|████▎     | 2065/4722 [21:36:24<21:18:31, 28.87s/it]
 44%|████▍     | 2066/4722 [21:36:52<21:10:09, 28.69s/it]
 44%|████▍     | 2067/4722 [21:37:23<21:35:56, 29.29s/it]
 44%|████▍     | 2068/4722 [21:37:54<21:55:09, 29.73s/it]
 44%|████▍     | 2069/4722 [21:38:19<21:00:59, 28.52s/it]
 44%|████▍     | 2070/4722 [21:38:49<21:10:40, 28.75s/it]
 44%|████▍     | 2071/4722 [21:39:19<21:34:05, 29.29s/it]
 44%|████▍     | 2072/4722 [21:39:46<20:55:10, 28.42s/it]
 44%|████▍     | 2073/4722 [21:40:15<21:11:55, 28.81s/it]
 44%|████▍     | 2074/4722 [21:40:42<20:49:04, 28.30s/it]
 44%|████▍     | 2075/4722 [21:41:16<21:53:05, 29.76s/it]
 44%|████▍     | 2076/4722 [21:41:48<22:26:28, 30.53s/it]
 44%|████▍     | 2077/4722 [21:42:17<22:01:26, 29.98s/it]
 44%|████▍     | 2078/4722 [21:42:49<22:29:17, 30.62s/it]
 44%|████▍     | 2079/4722 [21:43:15<21:36:40, 29.44s/it]
 44%|████▍     | 2080/4722 [21:43:41<20:50:04, 28.39s/it]
                                                         
{'loss': 0.5316, 'learning_rate': 0.0001, 'global_step': 2080, 'epoch': 1.32}
+
 44%|████▍     | 2080/4722 [21:43:41<20:50:04, 28.39s/it]
 44%|████▍     | 2081/4722 [21:44:07<20:12:34, 27.55s/it]
 44%|████▍     | 2082/4722 [21:44:35<20:15:25, 27.62s/it]
 44%|████▍     | 2083/4722 [21:45:02<20:13:46, 27.60s/it]
 44%|████▍     | 2084/4722 [21:45:32<20:38:35, 28.17s/it]
 44%|████▍     | 2085/4722 [21:46:03<21:18:49, 29.10s/it]
 44%|████▍     | 2086/4722 [21:46:32<21:12:26, 28.96s/it]
 44%|████▍     | 2087/4722 [21:46:59<20:46:37, 28.39s/it]
 44%|████▍     | 2088/4722 [21:47:28<21:02:24, 28.76s/it]
 44%|████▍     | 2089/4722 [21:47:59<21:27:07, 29.33s/it]
 44%|████▍     | 2090/4722 [21:48:26<20:57:56, 28.68s/it]
 44%|████▍     | 2091/4722 [21:48:59<21:44:56, 29.76s/it]
 44%|████▍     | 2092/4722 [21:49:33<22:45:49, 31.16s/it]
 44%|████▍     | 2093/4722 [21:50:00<21:56:19, 30.04s/it]
 44%|████▍     | 2094/4722 [21:50:33<22:27:10, 30.76s/it]
 44%|████▍     | 2095/4722 [21:51:03<22:19:27, 30.59s/it]
 44%|████▍     | 2096/4722 [21:51:29<21:19:51, 29.24s/it]
 44%|████▍     | 2097/4722 [21:51:58<21:13:56, 29.12s/it]
 44%|████▍     | 2098/4722 [21:52:23<20:17:20, 27.84s/it]
 44%|████▍     | 2099/4722 [21:52:48<19:45:12, 27.11s/it]
 44%|████▍     | 2100/4722 [21:53:24<21:33:12, 29.59s/it]
                                                         
{'loss': 0.4833, 'learning_rate': 0.0001, 'global_step': 2100, 'epoch': 1.33}
+
 44%|████▍     | 2100/4722 [21:53:24<21:33:12, 29.59s/it]
 44%|████▍     | 2101/4722 [21:53:56<22:10:35, 30.46s/it]
 45%|████▍     | 2102/4722 [21:54:25<21:56:34, 30.15s/it]
 45%|████▍     | 2103/4722 [21:54:55<21:41:53, 29.83s/it]
 45%|████▍     | 2104/4722 [21:55:21<20:55:45, 28.78s/it]
 45%|████▍     | 2105/4722 [21:55:49<20:50:24, 28.67s/it]
 45%|████▍     | 2106/4722 [21:56:17<20:38:00, 28.39s/it]
 45%|████▍     | 2107/4722 [21:56:46<20:47:51, 28.63s/it]
 45%|████▍     | 2108/4722 [21:57:15<20:44:05, 28.56s/it]
 45%|████▍     | 2109/4722 [21:57:46<21:23:37, 29.47s/it]
 45%|████▍     | 2110/4722 [21:58:16<21:27:50, 29.58s/it]
 45%|████▍     | 2111/4722 [21:58:47<21:44:28, 29.98s/it]
 45%|████▍     | 2112/4722 [21:59:17<21:39:40, 29.88s/it]
 45%|████▍     | 2113/4722 [21:59:40<20:17:40, 28.00s/it]
 45%|████▍     | 2114/4722 [22:00:13<21:18:11, 29.41s/it]
 45%|████▍     | 2115/4722 [22:00:40<20:52:50, 28.83s/it]
 45%|████▍     | 2116/4722 [22:01:07<20:28:38, 28.29s/it]
 45%|████▍     | 2117/4722 [22:01:36<20:26:58, 28.26s/it]
 45%|████▍     | 2118/4722 [22:02:04<20:22:13, 28.16s/it]
 45%|████▍     | 2119/4722 [22:02:31<20:08:26, 27.85s/it]
 45%|████▍     | 2120/4722 [22:02:58<20:00:01, 27.67s/it]
                                                         
{'loss': 0.4978, 'learning_rate': 0.0001, 'global_step': 2120, 'epoch': 1.35}
+
 45%|████▍     | 2120/4722 [22:02:58<20:00:01, 27.67s/it]
 45%|████▍     | 2121/4722 [22:03:25<19:51:18, 27.48s/it]
 45%|████▍     | 2122/4722 [22:03:54<20:13:26, 28.00s/it]
 45%|████▍     | 2123/4722 [22:04:20<19:45:58, 27.38s/it]
 45%|████▍     | 2124/4722 [22:04:52<20:41:23, 28.67s/it]
 45%|████▌     | 2125/4722 [22:05:17<20:01:31, 27.76s/it]
 45%|████▌     | 2126/4722 [22:05:46<20:14:33, 28.07s/it]
 45%|████▌     | 2127/4722 [22:06:17<20:44:39, 28.78s/it]
 45%|████▌     | 2128/4722 [22:06:51<21:52:56, 30.37s/it]
 45%|████▌     | 2129/4722 [22:07:19<21:26:44, 29.77s/it]
 45%|████▌     | 2130/4722 [22:07:53<22:15:14, 30.91s/it]
 45%|████▌     | 2131/4722 [22:08:22<21:47:49, 30.29s/it]
 45%|████▌     | 2132/4722 [22:08:53<21:56:31, 30.50s/it]
 45%|████▌     | 2133/4722 [22:09:21<21:29:42, 29.89s/it]
 45%|████▌     | 2134/4722 [22:09:50<21:24:05, 29.77s/it]
 45%|████▌     | 2135/4722 [22:10:21<21:36:33, 30.07s/it]
 45%|████▌     | 2136/4722 [22:10:49<21:05:04, 29.35s/it]
 45%|████▌     | 2137/4722 [22:11:21<21:34:46, 30.05s/it]
 45%|████▌     | 2138/4722 [22:11:50<21:19:31, 29.71s/it]
 45%|████▌     | 2139/4722 [22:12:19<21:20:32, 29.75s/it]
 45%|████▌     | 2140/4722 [22:12:47<20:57:14, 29.22s/it]
                                                         
{'loss': 0.4742, 'learning_rate': 0.0001, 'global_step': 2140, 'epoch': 1.36}
+
 45%|████▌     | 2140/4722 [22:12:47<20:57:14, 29.22s/it]
 45%|████▌     | 2141/4722 [22:13:14<20:27:27, 28.53s/it]
 45%|████▌     | 2142/4722 [22:13:40<19:53:36, 27.76s/it]
 45%|████▌     | 2143/4722 [22:14:12<20:44:42, 28.96s/it]
 45%|████▌     | 2144/4722 [22:14:37<19:47:34, 27.64s/it]
 45%|████▌     | 2145/4722 [22:15:05<20:00:52, 27.96s/it]
 45%|████▌     | 2146/4722 [22:15:35<20:17:54, 28.37s/it]
 45%|████▌     | 2147/4722 [22:16:03<20:22:59, 28.50s/it]
 45%|████▌     | 2148/4722 [22:16:28<19:33:03, 27.34s/it]
 46%|████▌     | 2149/4722 [22:16:56<19:41:35, 27.55s/it]
 46%|████▌     | 2150/4722 [22:17:23<19:38:49, 27.50s/it]
 46%|████▌     | 2151/4722 [22:17:49<19:08:30, 26.80s/it]
 46%|████▌     | 2152/4722 [22:18:19<19:52:31, 27.84s/it]
 46%|████▌     | 2153/4722 [22:18:48<20:07:27, 28.20s/it]
 46%|████▌     | 2154/4722 [22:19:14<19:34:22, 27.44s/it]
 46%|████▌     | 2155/4722 [22:19:42<19:46:52, 27.74s/it]
 46%|████▌     | 2156/4722 [22:20:11<20:01:41, 28.10s/it]
 46%|████▌     | 2157/4722 [22:20:36<19:21:44, 27.18s/it]
 46%|████▌     | 2158/4722 [22:21:10<20:55:06, 29.37s/it]
 46%|████▌     | 2159/4722 [22:21:40<21:01:14, 29.53s/it]
 46%|████▌     | 2160/4722 [22:22:14<21:58:13, 30.87s/it]
                                                         
{'loss': 0.5633, 'learning_rate': 0.0001, 'global_step': 2160, 'epoch': 1.37}
+
 46%|████▌     | 2160/4722 [22:22:14<21:58:13, 30.87s/it]
 46%|████▌     | 2161/4722 [22:22:44<21:40:11, 30.46s/it]
 46%|████▌     | 2162/4722 [22:23:11<20:51:29, 29.33s/it]
 46%|████▌     | 2163/4722 [22:23:43<21:33:41, 30.33s/it]
 46%|████▌     | 2164/4722 [22:24:12<21:17:42, 29.97s/it]
 46%|████▌     | 2165/4722 [22:24:42<21:16:21, 29.95s/it]
 46%|████▌     | 2166/4722 [22:25:12<21:17:50, 30.00s/it]
 46%|████▌     | 2167/4722 [22:25:40<20:42:56, 29.19s/it]
 46%|████▌     | 2168/4722 [22:26:04<19:42:14, 27.77s/it]
 46%|████▌     | 2169/4722 [22:26:33<19:49:25, 27.95s/it]
 46%|████▌     | 2170/4722 [22:26:58<19:18:52, 27.25s/it]
 46%|████▌     | 2171/4722 [22:27:28<19:56:48, 28.15s/it]
 46%|████▌     | 2172/4722 [22:27:57<19:57:55, 28.19s/it]
 46%|████▌     | 2173/4722 [22:28:22<19:21:54, 27.35s/it]
 46%|████▌     | 2174/4722 [22:28:51<19:44:31, 27.89s/it]
 46%|████▌     | 2175/4722 [22:29:17<19:15:48, 27.23s/it]
 46%|████▌     | 2176/4722 [22:29:46<19:33:19, 27.65s/it]
 46%|████▌     | 2177/4722 [22:30:14<19:37:28, 27.76s/it]
 46%|████▌     | 2178/4722 [22:30:43<19:57:38, 28.25s/it]
 46%|████▌     | 2179/4722 [22:31:14<20:35:50, 29.16s/it]
 46%|████▌     | 2180/4722 [22:31:44<20:45:50, 29.41s/it]
                                                         
{'loss': 0.5113, 'learning_rate': 0.0001, 'global_step': 2180, 'epoch': 1.38}
+
 46%|████▌     | 2180/4722 [22:31:44<20:45:50, 29.41s/it]
 46%|████▌     | 2181/4722 [22:32:13<20:32:23, 29.10s/it]
 46%|████▌     | 2182/4722 [22:32:40<20:04:36, 28.46s/it]
 46%|████▌     | 2183/4722 [22:33:07<19:49:17, 28.10s/it]
 46%|████▋     | 2184/4722 [22:33:34<19:32:57, 27.73s/it]
 46%|████▋     | 2185/4722 [22:34:00<19:20:25, 27.44s/it]
 46%|████▋     | 2186/4722 [22:34:30<19:46:20, 28.07s/it]
 46%|████▋     | 2187/4722 [22:34:56<19:17:15, 27.39s/it]
 46%|████▋     | 2188/4722 [22:35:25<19:36:36, 27.86s/it]
 46%|████▋     | 2189/4722 [22:35:52<19:33:34, 27.80s/it]
 46%|████▋     | 2190/4722 [22:36:22<20:00:53, 28.46s/it]
 46%|████▋     | 2191/4722 [22:36:50<19:55:29, 28.34s/it]
 46%|████▋     | 2192/4722 [22:37:22<20:30:34, 29.18s/it]
 46%|████▋     | 2193/4722 [22:37:49<20:12:40, 28.77s/it]
 46%|████▋     | 2194/4722 [22:38:15<19:34:13, 27.87s/it]
 46%|████▋     | 2195/4722 [22:38:48<20:34:21, 29.31s/it]
 47%|████▋     | 2196/4722 [22:39:18<20:46:59, 29.62s/it]
 47%|████▋     | 2197/4722 [22:39:39<18:49:27, 26.84s/it]
 47%|████▋     | 2198/4722 [22:39:55<16:35:15, 23.66s/it]
 47%|████▋     | 2199/4722 [22:40:12<15:15:03, 21.76s/it]
 47%|████▋     | 2200/4722 [22:40:28<13:56:25, 19.90s/it]
                                                         
{'loss': 0.4774, 'learning_rate': 0.0001, 'global_step': 2200, 'epoch': 1.4}
+
 47%|████▋     | 2200/4722 [22:40:28<13:56:25, 19.90s/it]
 47%|████▋     | 2201/4722 [22:40:45<13:28:27, 19.24s/it]
 47%|████▋     | 2202/4722 [22:41:02<12:50:31, 18.35s/it]
 47%|████▋     | 2203/4722 [22:41:18<12:29:06, 17.84s/it]
 47%|████▋     | 2204/4722 [22:41:35<12:13:03, 17.47s/it]
 47%|████▋     | 2205/4722 [22:41:51<11:54:35, 17.03s/it]
 47%|████▋     | 2206/4722 [22:42:06<11:36:01, 16.60s/it]
 47%|████▋     | 2207/4722 [22:42:21<11:14:32, 16.09s/it]
 47%|████▋     | 2208/4722 [22:42:40<11:51:30, 16.98s/it]
 47%|████▋     | 2209/4722 [22:42:58<11:55:48, 17.09s/it]
 47%|████▋     | 2210/4722 [22:43:14<11:46:22, 16.87s/it]
 47%|████▋     | 2211/4722 [22:43:32<11:54:25, 17.07s/it]
 47%|████▋     | 2212/4722 [22:43:49<12:00:08, 17.21s/it]
 47%|████▋     | 2213/4722 [22:44:06<11:55:00, 17.10s/it]
 47%|████▋     | 2214/4722 [22:44:25<12:15:04, 17.59s/it]
 47%|████▋     | 2215/4722 [22:44:41<12:03:29, 17.32s/it]
 47%|████▋     | 2216/4722 [22:45:01<12:29:19, 17.94s/it]
 47%|████▋     | 2217/4722 [22:45:17<12:00:32, 17.26s/it]
 47%|████▋     | 2218/4722 [22:45:34<12:00:04, 17.25s/it]
 47%|████▋     | 2219/4722 [22:45:52<12:06:22, 17.41s/it]
 47%|████▋     | 2220/4722 [22:46:14<13:03:01, 18.78s/it]
                                                         
{'loss': 0.4837, 'learning_rate': 0.0001, 'global_step': 2220, 'epoch': 1.41}
+
 47%|████▋     | 2220/4722 [22:46:14<13:03:01, 18.78s/it]
 47%|████▋     | 2221/4722 [22:46:31<12:40:49, 18.25s/it]
 47%|████▋     | 2222/4722 [22:46:48<12:30:51, 18.02s/it]
 47%|████▋     | 2223/4722 [22:47:05<12:17:52, 17.72s/it]
 47%|████▋     | 2224/4722 [22:47:22<12:03:53, 17.39s/it]
 47%|████▋     | 2225/4722 [22:47:38<11:46:54, 16.99s/it]
 47%|████▋     | 2226/4722 [22:47:54<11:36:52, 16.75s/it]
 47%|████▋     | 2227/4722 [22:48:14<12:21:57, 17.84s/it]
 47%|████▋     | 2228/4722 [22:48:29<11:38:51, 16.81s/it]
 47%|████▋     | 2229/4722 [22:48:43<11:11:06, 16.15s/it]
 47%|████▋     | 2230/4722 [22:49:02<11:46:06, 17.00s/it]
 47%|████▋     | 2231/4722 [22:49:18<11:24:39, 16.49s/it]
 47%|████▋     | 2232/4722 [22:49:36<11:52:20, 17.17s/it]
 47%|████▋     | 2233/4722 [22:49:53<11:45:02, 17.00s/it]
 47%|████▋     | 2234/4722 [22:50:10<11:43:36, 16.97s/it]
 47%|████▋     | 2235/4722 [22:50:27<11:49:59, 17.13s/it]
 47%|████▋     | 2236/4722 [22:50:48<12:31:21, 18.13s/it]
 47%|████▋     | 2237/4722 [22:51:06<12:27:36, 18.05s/it]
 47%|████▋     | 2238/4722 [22:51:25<12:40:49, 18.38s/it]
 47%|████▋     | 2239/4722 [22:51:42<12:28:32, 18.09s/it]
 47%|████▋     | 2240/4722 [22:52:00<12:26:03, 18.04s/it]
                                                         
{'loss': 0.5512, 'learning_rate': 0.0001, 'global_step': 2240, 'epoch': 1.42}
+
 47%|████▋     | 2240/4722 [22:52:00<12:26:03, 18.04s/it]
 47%|████▋     | 2241/4722 [22:52:20<12:50:35, 18.64s/it]
 47%|████▋     | 2242/4722 [22:52:36<12:15:05, 17.78s/it]
 48%|████▊     | 2243/4722 [22:52:54<12:16:48, 17.83s/it]
 48%|████▊     | 2244/4722 [22:53:11<12:11:44, 17.72s/it]
 48%|████▊     | 2245/4722 [22:53:26<11:38:35, 16.92s/it]
 48%|████▊     | 2246/4722 [22:53:46<12:12:30, 17.75s/it]
 48%|████▊     | 2247/4722 [22:54:02<11:42:52, 17.04s/it]
 48%|████▊     | 2248/4722 [22:54:19<11:52:48, 17.29s/it]
 48%|████▊     | 2249/4722 [22:54:37<12:02:32, 17.53s/it]
 48%|████▊     | 2250/4722 [22:54:55<12:06:13, 17.63s/it]
 48%|████▊     | 2251/4722 [22:55:12<11:59:36, 17.47s/it]
 48%|████▊     | 2252/4722 [22:55:29<11:53:22, 17.33s/it]
 48%|████▊     | 2253/4722 [22:55:50<12:28:58, 18.20s/it]
 48%|████▊     | 2254/4722 [22:56:10<12:57:52, 18.91s/it]
 48%|████▊     | 2255/4722 [22:56:28<12:39:56, 18.48s/it]
 48%|████▊     | 2256/4722 [22:56:44<12:10:30, 17.77s/it]
 48%|████▊     | 2257/4722 [22:57:01<12:00:17, 17.53s/it]
 48%|████▊     | 2258/4722 [22:57:17<11:39:22, 17.03s/it]
 48%|████▊     | 2259/4722 [22:57:33<11:25:35, 16.70s/it]
 48%|████▊     | 2260/4722 [22:57:49<11:17:05, 16.50s/it]
                                                         
{'loss': 0.4818, 'learning_rate': 0.0001, 'global_step': 2260, 'epoch': 1.44}
+
 48%|████▊     | 2260/4722 [22:57:49<11:17:05, 16.50s/it]
 48%|████▊     | 2261/4722 [22:58:04<10:57:00, 16.02s/it]
 48%|████▊     | 2262/4722 [22:58:23<11:33:56, 16.93s/it]
 48%|████▊     | 2263/4722 [22:58:40<11:39:25, 17.07s/it]
 48%|████▊     | 2264/4722 [22:58:55<11:16:09, 16.51s/it]
 48%|████▊     | 2265/4722 [22:59:15<12:01:57, 17.63s/it]
 48%|████▊     | 2266/4722 [22:59:35<12:27:03, 18.25s/it]
 48%|████▊     | 2267/4722 [22:59:52<12:11:15, 17.87s/it]
 48%|████▊     | 2268/4722 [23:00:08<11:48:08, 17.31s/it]
 48%|████▊     | 2269/4722 [23:00:43<15:29:02, 22.72s/it]
 48%|████▊     | 2270/4722 [23:01:14<17:07:13, 25.14s/it]
 48%|████▊     | 2271/4722 [23:01:46<18:22:22, 26.99s/it]
 48%|████▊     | 2272/4722 [23:02:21<20:07:19, 29.57s/it]
 48%|████▊     | 2273/4722 [23:02:55<21:00:03, 30.87s/it]
 48%|████▊     | 2274/4722 [23:03:31<22:07:19, 32.53s/it]
 48%|████▊     | 2275/4722 [23:04:04<22:03:15, 32.45s/it]
 48%|████▊     | 2276/4722 [23:04:46<23:59:47, 35.32s/it]
 48%|████▊     | 2277/4722 [23:05:30<25:46:00, 37.94s/it]
 48%|████▊     | 2278/4722 [23:06:01<24:25:30, 35.98s/it]
 48%|████▊     | 2279/4722 [23:06:30<22:59:12, 33.87s/it]
 48%|████▊     | 2280/4722 [23:07:02<22:37:52, 33.36s/it]
                                                         
{'loss': 0.5282, 'learning_rate': 0.0001, 'global_step': 2280, 'epoch': 1.45}
+
 48%|████▊     | 2280/4722 [23:07:02<22:37:52, 33.36s/it]
 48%|████▊     | 2281/4722 [23:07:37<22:48:13, 33.63s/it]
 48%|████▊     | 2282/4722 [23:08:04<21:31:01, 31.75s/it]
 48%|████▊     | 2283/4722 [23:08:47<23:44:42, 35.05s/it]
 48%|████▊     | 2284/4722 [23:09:22<23:46:08, 35.10s/it]
 48%|████▊     | 2285/4722 [23:09:52<22:39:09, 33.46s/it]
 48%|████▊     | 2286/4722 [23:10:24<22:32:02, 33.30s/it]
 48%|████▊     | 2287/4722 [23:10:55<22:00:15, 32.53s/it]
 48%|████▊     | 2288/4722 [23:11:29<22:11:56, 32.83s/it]
 48%|████▊     | 2289/4722 [23:12:01<22:01:18, 32.58s/it]
 48%|████▊     | 2290/4722 [23:12:36<22:37:26, 33.49s/it]
 49%|████▊     | 2291/4722 [23:13:13<23:16:30, 34.47s/it]
 49%|████▊     | 2292/4722 [23:13:43<22:22:48, 33.16s/it]
 49%|████▊     | 2293/4722 [23:14:16<22:15:52, 33.00s/it]
 49%|████▊     | 2294/4722 [23:14:51<22:46:56, 33.78s/it]
 49%|████▊     | 2295/4722 [23:15:27<23:08:28, 34.33s/it]
 49%|████▊     | 2296/4722 [23:16:04<23:44:42, 35.24s/it]
 49%|████▊     | 2297/4722 [23:16:34<22:30:59, 33.43s/it]
 49%|████▊     | 2298/4722 [23:17:03<21:43:04, 32.25s/it]
 49%|████▊     | 2299/4722 [23:17:32<20:59:21, 31.19s/it]
 49%|████▊     | 2300/4722 [23:18:02<20:45:17, 30.85s/it]
                                                         
{'loss': 0.468, 'learning_rate': 0.0001, 'global_step': 2300, 'epoch': 1.46}
+
 49%|████▊     | 2300/4722 [23:18:02<20:45:17, 30.85s/it]
 49%|████▊     | 2301/4722 [23:18:42<22:38:00, 33.66s/it]
 49%|████▉     | 2302/4722 [23:19:20<23:34:22, 35.07s/it]
 49%|████▉     | 2303/4722 [23:19:51<22:41:12, 33.76s/it]
 49%|████▉     | 2304/4722 [23:20:30<23:38:28, 35.20s/it]
 49%|████▉     | 2305/4722 [23:21:01<22:52:26, 34.07s/it]
 49%|████▉     | 2306/4722 [23:21:35<22:55:08, 34.15s/it]
 49%|████▉     | 2307/4722 [23:22:07<22:28:13, 33.50s/it]
 49%|████▉     | 2308/4722 [23:22:40<22:21:18, 33.34s/it]
 49%|████▉     | 2309/4722 [23:23:12<22:00:30, 32.83s/it]
 49%|████▉     | 2310/4722 [23:23:48<22:31:22, 33.62s/it]
 49%|████▉     | 2311/4722 [23:24:24<23:07:01, 34.52s/it]
 49%|████▉     | 2312/4722 [23:24:55<22:17:36, 33.30s/it]
 49%|████▉     | 2313/4722 [23:25:31<22:52:20, 34.18s/it]
 49%|████▉     | 2314/4722 [23:26:07<23:10:29, 34.65s/it]
 49%|████▉     | 2315/4722 [23:26:38<22:37:23, 33.84s/it]
 49%|████▉     | 2316/4722 [23:27:05<21:08:41, 31.64s/it]
 49%|████▉     | 2317/4722 [23:27:39<21:40:25, 32.44s/it]
 49%|████▉     | 2318/4722 [23:28:15<22:19:31, 33.43s/it]
 49%|████▉     | 2319/4722 [23:28:47<22:02:35, 33.02s/it]
 49%|████▉     | 2320/4722 [23:29:21<22:09:42, 33.21s/it]
                                                         
{'loss': 0.5227, 'learning_rate': 0.0001, 'global_step': 2320, 'epoch': 1.47}
+
 49%|████▉     | 2320/4722 [23:29:21<22:09:42, 33.21s/it]
 49%|████▉     | 2321/4722 [23:29:59<23:05:27, 34.62s/it]
 49%|████▉     | 2322/4722 [23:30:28<22:05:54, 33.15s/it]
 49%|████▉     | 2323/4722 [23:30:57<21:15:03, 31.89s/it]
 49%|████▉     | 2324/4722 [23:31:26<20:33:24, 30.86s/it]
 49%|████▉     | 2325/4722 [23:31:59<21:06:00, 31.69s/it]
 49%|████▉     | 2326/4722 [23:32:31<21:00:45, 31.57s/it]
 49%|████▉     | 2327/4722 [23:33:04<21:22:07, 32.12s/it]
 49%|████▉     | 2328/4722 [23:33:36<21:12:38, 31.90s/it]
 49%|████▉     | 2329/4722 [23:34:09<21:28:38, 32.31s/it]
 49%|████▉     | 2330/4722 [23:34:41<21:26:07, 32.26s/it]
 49%|████▉     | 2331/4722 [23:35:12<21:17:01, 32.05s/it]
 49%|████▉     | 2332/4722 [23:35:44<21:11:26, 31.92s/it]
 49%|████▉     | 2333/4722 [23:36:19<21:46:48, 32.82s/it]
 49%|████▉     | 2334/4722 [23:36:58<22:54:29, 34.54s/it]
 49%|████▉     | 2335/4722 [23:37:31<22:39:21, 34.17s/it]
 49%|████▉     | 2336/4722 [23:38:00<21:38:58, 32.67s/it]
 49%|████▉     | 2337/4722 [23:38:36<22:11:49, 33.51s/it]
 50%|████▉     | 2338/4722 [23:39:14<23:13:32, 35.07s/it]
 50%|████▉     | 2339/4722 [23:39:50<23:24:38, 35.37s/it]
 50%|████▉     | 2340/4722 [23:40:19<22:02:27, 33.31s/it]
                                                         
{'loss': 0.4959, 'learning_rate': 0.0001, 'global_step': 2340, 'epoch': 1.49}
+
 50%|████▉     | 2340/4722 [23:40:19<22:02:27, 33.31s/it]
 50%|████▉     | 2341/4722 [23:40:51<21:49:18, 32.99s/it]
 50%|████▉     | 2342/4722 [23:41:25<22:05:35, 33.42s/it]
 50%|████▉     | 2343/4722 [23:42:01<22:24:49, 33.92s/it]
 50%|████▉     | 2344/4722 [23:42:35<22:28:14, 34.02s/it]
 50%|████▉     | 2345/4722 [23:43:06<21:50:07, 33.07s/it]
 50%|████▉     | 2346/4722 [23:43:41<22:21:14, 33.87s/it]
 50%|████▉     | 2347/4722 [23:44:15<22:13:26, 33.69s/it]
 50%|████▉     | 2348/4722 [23:44:49<22:16:00, 33.77s/it]
 50%|████▉     | 2349/4722 [23:45:21<22:04:30, 33.49s/it]
 50%|████▉     | 2350/4722 [23:45:57<22:26:19, 34.06s/it]
 50%|████▉     | 2351/4722 [23:46:31<22:30:42, 34.18s/it]
 50%|████▉     | 2352/4722 [23:47:04<22:13:43, 33.77s/it]
 50%|████▉     | 2353/4722 [23:47:34<21:25:26, 32.56s/it]
 50%|████▉     | 2354/4722 [23:48:13<22:39:25, 34.45s/it]
 50%|████▉     | 2355/4722 [23:48:46<22:29:10, 34.20s/it]
 50%|████▉     | 2356/4722 [23:49:22<22:44:02, 34.59s/it]
 50%|████▉     | 2357/4722 [23:49:50<21:27:59, 32.68s/it]
 50%|████▉     | 2358/4722 [23:50:24<21:38:42, 32.96s/it]
 50%|████▉     | 2359/4722 [23:50:55<21:16:31, 32.41s/it]
 50%|████▉     | 2360/4722 [23:51:27<21:14:29, 32.37s/it]
                                                         
{'loss': 0.5313, 'learning_rate': 0.0001, 'global_step': 2360, 'epoch': 1.5}
+
 50%|████▉     | 2360/4722 [23:51:27<21:14:29, 32.37s/it]
 50%|█████     | 2361/4722 [23:52:03<21:51:49, 33.34s/it]
 50%|█████     | 2362/4722 [23:52:34<21:26:20, 32.70s/it]
 50%|█████     | 2363/4722 [23:53:10<22:06:22, 33.74s/it]
 50%|█████     | 2364/4722 [23:53:39<21:08:31, 32.28s/it]
 50%|██��██     | 2365/4722 [23:54:07<20:14:15, 30.91s/it]
 50%|█████     | 2366/4722 [23:54:35<19:47:33, 30.24s/it]
 50%|█████     | 2367/4722 [23:55:06<19:53:26, 30.41s/it]
 50%|█████     | 2368/4722 [23:55:38<20:15:44, 30.99s/it]
 50%|█████     | 2369/4722 [23:56:11<20:31:37, 31.41s/it]
 50%|█████     | 2370/4722 [23:56:45<21:01:36, 32.18s/it]
 50%|█████     | 2371/4722 [23:57:15<20:34:09, 31.50s/it]
 50%|█████     | 2372/4722 [23:57:49<21:02:51, 32.24s/it]
 50%|█████     | 2373/4722 [23:58:19<20:42:57, 31.75s/it]
 50%|█████     | 2374/4722 [23:59:01<22:42:17, 34.81s/it]
 50%|█████     | 2375/4722 [23:59:28<21:08:03, 32.42s/it]
 50%|█████     | 2376/4722 [24:00:08<22:31:24, 34.56s/it]
 50%|█████     | 2377/4722 [24:00:46<23:10:42, 35.58s/it]
 50%|█████     | 2378/4722 [24:01:12<21:25:10, 32.90s/it]
 50%|█████     | 2379/4722 [24:01:47<21:41:50, 33.34s/it]
 50%|█████     | 2380/4722 [24:02:24<22:25:48, 34.48s/it]
                                                         
{'loss': 0.4999, 'learning_rate': 0.0001, 'global_step': 2380, 'epoch': 1.51}
+
 50%|█████     | 2380/4722 [24:02:24<22:25:48, 34.48s/it]
 50%|█████     | 2381/4722 [24:02:54<21:37:24, 33.25s/it]
 50%|█████     | 2382/4722 [24:03:25<21:11:08, 32.59s/it]
 50%|█████     | 2383/4722 [24:03:59<21:27:26, 33.03s/it]
 50%|█████     | 2384/4722 [24:04:35<22:03:22, 33.96s/it]
 51%|█████     | 2385/4722 [24:05:14<22:55:42, 35.32s/it]
 51%|█████     | 2386/4722 [24:05:50<23:04:09, 35.55s/it]
 51%|█████     | 2387/4722 [24:06:28<23:38:16, 36.44s/it]
 51%|█████     | 2388/4722 [24:07:04<23:28:17, 36.20s/it]
 51%|█████     | 2389/4722 [24:07:36<22:32:11, 34.78s/it]
 51%|█████     | 2390/4722 [24:08:14<23:09:27, 35.75s/it]
 51%|█████     | 2391/4722 [24:08:59<24:58:15, 38.57s/it]
 51%|█████     | 2392/4722 [24:09:33<24:07:00, 37.26s/it]
 51%|█████     | 2393/4722 [24:10:07<23:27:16, 36.25s/it]
 51%|█████     | 2394/4722 [24:10:40<22:45:32, 35.19s/it]
 51%|█████     | 2395/4722 [24:11:17<23:16:32, 36.01s/it]
 51%|█████     | 2396/4722 [24:11:53<23:09:49, 35.85s/it]
 51%|█████     | 2397/4722 [24:12:23<21:58:20, 34.02s/it]
 51%|█████     | 2398/4722 [24:12:58<22:07:26, 34.27s/it]
 51%|█████     | 2399/4722 [24:13:27<21:12:31, 32.87s/it]
 51%|█████     | 2400/4722 [24:14:03<21:42:40, 33.66s/it]
                                                         
{'loss': 0.5336, 'learning_rate': 0.0001, 'global_step': 2400, 'epoch': 1.52}
+
 51%|█████     | 2400/4722 [24:14:03<21:42:40, 33.66s/it]
 51%|█████     | 2401/4722 [24:14:43<22:59:27, 35.66s/it]
 51%|█████     | 2402/4722 [24:15:15<22:13:09, 34.48s/it]
 51%|█████     | 2403/4722 [24:15:50<22:27:27, 34.86s/it]
 51%|█████     | 2404/4722 [24:16:26<22:30:55, 34.97s/it]
 51%|█████     | 2405/4722 [24:17:01<22:34:33, 35.08s/it]
 51%|█████     | 2406/4722 [24:17:32<21:46:30, 33.85s/it]
 51%|█████     | 2407/4722 [24:18:11<22:49:07, 35.48s/it]
 51%|█████     | 2408/4722 [24:18:42<21:57:10, 34.15s/it]
 51%|█████     | 2409/4722 [24:19:17<22:02:29, 34.31s/it]
 51%|█████     | 2410/4722 [24:19:50<21:45:17, 33.87s/it]
 51%|█████     | 2411/4722 [24:20:11<19:13:01, 29.94s/it]
 51%|█████     | 2412/4722 [24:20:24<16:01:13, 24.97s/it]
 51%|█████     | 2413/4722 [24:20:40<14:19:34, 22.34s/it]
 51%|█████     | 2414/4722 [24:20:57<13:09:55, 20.54s/it]
 51%|█████     | 2415/4722 [24:21:17<13:14:24, 20.66s/it]
 51%|█████     | 2416/4722 [24:21:35<12:40:58, 19.80s/it]
 51%|█████     | 2417/4722 [24:21:52<12:10:40, 19.02s/it]
 51%|█████     | 2418/4722 [24:22:10<11:47:38, 18.43s/it]
 51%|█████     | 2419/4722 [24:22:26<11:23:43, 17.81s/it]
 51%|█████     | 2420/4722 [24:22:45<11:34:28, 18.10s/it]
                                                         
{'loss': 0.4878, 'learning_rate': 0.0001, 'global_step': 2420, 'epoch': 1.54}
+
 51%|█████     | 2420/4722 [24:22:45<11:34:28, 18.10s/it]
 51%|█████▏    | 2421/4722 [24:23:03<11:41:37, 18.30s/it]
 51%|█████▏    | 2422/4722 [24:23:20<11:21:27, 17.78s/it]
 51%|█████▏    | 2423/4722 [24:23:38<11:18:05, 17.70s/it]
 51%|█████▏    | 2424/4722 [24:23:55<11:18:10, 17.71s/it]
 51%|█████▏    | 2425/4722 [24:24:11<11:00:07, 17.24s/it]
 51%|█████▏    | 2426/4722 [24:24:30<11:12:57, 17.59s/it]
 51%|█████▏    | 2427/4722 [24:24:47<11:11:42, 17.56s/it]
 51%|█████▏    | 2428/4722 [24:25:04<10:59:43, 17.26s/it]
 51%|█████▏    | 2429/4722 [24:25:23<11:22:33, 17.86s/it]
 51%|█████▏    | 2430/4722 [24:25:40<11:14:57, 17.67s/it]
 51%|��████▏    | 2431/4722 [24:26:01<11:44:27, 18.45s/it]
 52%|█████▏    | 2432/4722 [24:26:17<11:19:18, 17.80s/it]
 52%|█████▏    | 2433/4722 [24:26:33<11:02:51, 17.38s/it]
 52%|█████▏    | 2434/4722 [24:26:50<10:54:52, 17.17s/it]
 52%|█████▏    | 2435/4722 [24:27:10<11:23:06, 17.92s/it]
 52%|█████▏    | 2436/4722 [24:27:26<11:10:35, 17.60s/it]
 52%|█████▏    | 2437/4722 [24:27:45<11:16:56, 17.78s/it]
 52%|█████▏    | 2438/4722 [24:28:03<11:18:02, 17.81s/it]
 52%|█████▏    | 2439/4722 [24:28:20<11:09:52, 17.61s/it]
 52%|█████▏    | 2440/4722 [24:28:34<10:32:35, 16.63s/it]
                                                         
{'loss': 0.4692, 'learning_rate': 0.0001, 'global_step': 2440, 'epoch': 1.55}
+
 52%|█████▏    | 2440/4722 [24:28:34<10:32:35, 16.63s/it]
 52%|█████▏    | 2441/4722 [24:28:51<10:35:32, 16.72s/it]
 52%|█████▏    | 2442/4722 [24:29:10<11:03:54, 17.47s/it]
 52%|█████▏    | 2443/4722 [24:29:29<11:20:36, 17.92s/it]
 52%|█████▏    | 2444/4722 [24:29:46<11:04:36, 17.50s/it]
 52%|█████▏    | 2445/4722 [24:30:05<11:25:52, 18.07s/it]
 52%|█████▏    | 2446/4722 [24:30:20<10:48:29, 17.10s/it]
 52%|█████▏    | 2447/4722 [24:30:40<11:22:33, 18.00s/it]
 52%|█████▏    | 2448/4722 [24:30:56<11:03:57, 17.52s/it]
 52%|█████▏    | 2449/4722 [24:31:15<11:21:06, 17.98s/it]
 52%|█████▏    | 2450/4722 [24:31:32<11:06:22, 17.60s/it]
 52%|█████▏    | 2451/4722 [24:31:50<11:10:12, 17.71s/it]
 52%|█████▏    | 2452/4722 [24:32:06<10:51:44, 17.23s/it]
 52%|█████▏    | 2453/4722 [24:32:26<11:15:05, 17.85s/it]
 52%|█████▏    | 2454/4722 [24:32:41<10:50:47, 17.22s/it]
 52%|█████▏    | 2455/4722 [24:32:59<10:54:05, 17.31s/it]
 52%|█████▏    | 2456/4722 [24:33:18<11:11:20, 17.78s/it]
 52%|█████▏    | 2457/4722 [24:33:36<11:20:25, 18.02s/it]
 52%|█████▏    | 2458/4722 [24:33:54<11:16:17, 17.92s/it]
 52%|█████▏    | 2459/4722 [24:34:09<10:40:31, 16.98s/it]
 52%|█████▏    | 2460/4722 [24:34:25<10:33:29, 16.80s/it]
                                                         
{'loss': 0.4919, 'learning_rate': 0.0001, 'global_step': 2460, 'epoch': 1.56}
+
 52%|█████▏    | 2460/4722 [24:34:25<10:33:29, 16.80s/it]
 52%|█████▏    | 2461/4722 [24:34:42<10:32:11, 16.78s/it]
 52%|█████▏    | 2462/4722 [24:34:59<10:38:43, 16.96s/it]
 52%|█████▏    | 2463/4722 [24:35:16<10:36:50, 16.91s/it]
 52%|█████▏    | 2464/4722 [24:35:31<10:18:39, 16.44s/it]
 52%|█████▏    | 2465/4722 [24:35:49<10:28:37, 16.71s/it]
 52%|█████▏    | 2466/4722 [24:36:06<10:31:38, 16.80s/it]
 52%|█████▏    | 2467/4722 [24:36:25<11:01:40, 17.61s/it]
 52%|█████▏    | 2468/4722 [24:36:46<11:37:06, 18.56s/it]
 52%|█████▏    | 2469/4722 [24:37:05<11:47:07, 18.83s/it]
 52%|█████▏    | 2470/4722 [24:37:25<12:00:17, 19.19s/it]
 52%|█████▏    | 2471/4722 [24:37:41<11:13:00, 17.94s/it]
 52%|█████▏    | 2472/4722 [24:37:57<10:55:34, 17.48s/it]
 52%|█████▏    | 2473/4722 [24:38:13<10:39:32, 17.06s/it]
 52%|█████▏    | 2474/4722 [24:38:30<10:43:01, 17.16s/it]
 52%|█████▏    | 2475/4722 [24:38:50<11:15:20, 18.03s/it]
 52%|█████▏    | 2476/4722 [24:39:08<11:04:22, 17.75s/it]
 52%|█████▏    | 2477/4722 [24:39:24<10:50:52, 17.40s/it]
 52%|█████▏    | 2478/4722 [24:39:41<10:48:04, 17.33s/it]
 52%|█████▏    | 2479/4722 [24:39:59<10:51:32, 17.43s/it]
 53%|█████▎    | 2480/4722 [24:40:17<10:53:14, 17.48s/it]
                                                         
{'loss': 0.5449, 'learning_rate': 0.0001, 'global_step': 2480, 'epoch': 1.58}
+
 53%|█████▎    | 2480/4722 [24:40:17<10:53:14, 17.48s/it]
 53%|█████▎    | 2481/4722 [24:40:42<12:25:45, 19.97s/it]
 53%|█████▎    | 2482/4722 [24:41:01<12:07:07, 19.48s/it]
 53%|█████▎    | 2483/4722 [24:41:17<11:26:25, 18.39s/it]
 53%|█████▎    | 2484/4722 [24:41:35<11:26:35, 18.41s/it]
 53%|█████▎    | 2485/4722 [24:41:50<10:52:58, 17.51s/it]
 53%|█████▎    | 2486/4722 [24:42:08<10:53:17, 17.53s/it]
 53%|█████▎    | 2487/4722 [24:42:25<10:48:15, 17.40s/it]
 53%|█████▎    | 2488/4722 [24:42:44<11:03:07, 17.81s/it]
 53%|█████▎    | 2489/4722 [24:42:59<10:37:16, 17.12s/it]
 53%|█████▎    | 2490/4722 [24:43:17<10:39:37, 17.19s/it]
 53%|█████▎    | 2491/4722 [24:43:36<11:03:53, 17.85s/it]
 53%|█████▎    | 2492/4722 [24:43:54<11:08:57, 18.00s/it]
 53%|█████▎    | 2493/4722 [24:44:09<10:33:47, 17.06s/it]
 53%|█████▎    | 2494/4722 [24:44:25<10:22:45, 16.77s/it]
 53%|█████▎    | 2495/4722 [24:44:42<10:21:29, 16.74s/it]
 53%|█████▎    | 2496/4722 [24:44:56<9:52:12, 15.96s/it] 
 53%|█████▎    | 2497/4722 [24:45:11<9:41:10, 15.67s/it]
 53%|█████▎    | 2498/4722 [24:45:26<9:26:54, 15.29s/it]
 53%|█████▎    | 2499/4722 [24:45:43<9:48:21, 15.88s/it]
 53%|█████▎    | 2500/4722 [24:45:58<9:43:18, 15.75s/it]
                                                        
{'loss': 0.4466, 'learning_rate': 0.0001, 'global_step': 2500, 'epoch': 1.59}
+
 53%|█████▎    | 2500/4722 [24:45:58<9:43:18, 15.75s/it]
 53%|█████▎    | 2501/4722 [24:46:13<9:29:19, 15.38s/it]
 53%|█████▎    | 2502/4722 [24:46:33<10:17:17, 16.68s/it]
 53%|█████▎    | 2503/4722 [24:46:52<10:42:07, 17.36s/it]
 53%|█████▎    | 2504/4722 [24:47:07<10:21:54, 16.82s/it]
 53%|█████▎    | 2505/4722 [24:47:23<10:14:43, 16.64s/it]
 53%|█████▎    | 2506/4722 [24:47:40<10:12:42, 16.59s/it]
 53%|█████▎    | 2507/4722 [24:48:01<11:00:31, 17.89s/it]
 53%|█████▎    | 2508/4722 [24:48:16<10:32:38, 17.14s/it]
 53%|█████▎    | 2509/4722 [24:48:32<10:13:11, 16.63s/it]
 53%|█████▎    | 2510/4722 [24:48:50<10:37:40, 17.30s/it]
 53%|█████▎    | 2511/4722 [24:49:11<11:10:24, 18.19s/it]
 53%|█████▎    | 2512/4722 [24:49:30<11:18:44, 18.43s/it]
 53%|█████▎    | 2513/4722 [24:49:49<11:23:20, 18.56s/it]
 53%|█████▎    | 2514/4722 [24:50:08<11:35:37, 18.90s/it]
 53%|█████▎    | 2515/4722 [24:50:26<11:18:24, 18.44s/it]
 53%|█████▎    | 2516/4722 [24:50:47<11:54:19, 19.43s/it]
 53%|█████▎    | 2517/4722 [24:51:17<13:46:42, 22.50s/it]
 53%|█████▎    | 2518/4722 [24:51:51<15:56:24, 26.04s/it]
 53%|█████▎    | 2519/4722 [24:52:27<17:39:14, 28.85s/it]
 53%|█████▎    | 2520/4722 [24:52:59<18:18:05, 29.92s/it]
                                                         
{'loss': 0.4335, 'learning_rate': 0.0001, 'global_step': 2520, 'epoch': 1.6}
+
 53%|█████▎    | 2520/4722 [24:52:59<18:18:05, 29.92s/it]
 53%|█████▎    | 2521/4722 [24:53:34<19:17:27, 31.55s/it]
 53%|█████▎    | 2522/4722 [24:54:10<19:55:38, 32.61s/it]
 53%|█████▎    | 2523/4722 [24:54:45<20:29:19, 33.54s/it]
 53%|█████▎    | 2524/4722 [24:55:21<20:54:06, 34.23s/it]
 53%|█████▎    | 2525/4722 [24:55:54<20:34:46, 33.72s/it]
 53%|█████▎    | 2526/4722 [24:56:28<20:37:13, 33.80s/it]
 54%|█████▎    | 2527/4722 [24:57:06<21:32:18, 35.33s/it]
 54%|█████▎    | 2528/4722 [24:57:41<21:26:45, 35.19s/it]
 54%|█████▎    | 2529/4722 [24:58:17<21:26:46, 35.21s/it]
 54%|█████▎    | 2530/4722 [24:58:47<20:28:27, 33.63s/it]
 54%|█████▎    | 2531/4722 [24:59:15<19:28:44, 32.01s/it]
 54%|█████▎    | 2532/4722 [24:59:51<20:11:21, 33.19s/it]
 54%|█████▎    | 2533/4722 [25:00:22<19:53:44, 32.72s/it]
 54%|█████▎    | 2534/4722 [25:00:51<19:03:21, 31.35s/it]
 54%|█████▎    | 2535/4722 [25:01:23<19:11:06, 31.58s/it]
 54%|█████▎    | 2536/4722 [25:02:02<20:36:02, 33.93s/it]
 54%|█████▎    | 2537/4722 [25:02:37<20:49:22, 34.31s/it]
 54%|█████▎    | 2538/4722 [25:03:12<20:50:29, 34.35s/it]
 54%|█████▍    | 2539/4722 [25:03:46<20:49:39, 34.35s/it]
 54%|█████▍    | 2540/4722 [25:04:17<20:13:14, 33.36s/it]
                                                         
{'loss': 0.4979, 'learning_rate': 0.0001, 'global_step': 2540, 'epoch': 1.61}
+
 54%|█████▍    | 2540/4722 [25:04:17<20:13:14, 33.36s/it]
 54%|█████▍    | 2541/4722 [25:04:52<20:31:59, 33.89s/it]
 54%|█████▍    | 2542/4722 [25:05:23<19:57:43, 32.97s/it]
 54%|█████▍    | 2543/4722 [25:05:54<19:37:24, 32.42s/it]
 54%|█████▍    | 2544/4722 [25:06:30<20:12:55, 33.41s/it]
 54%|█████▍    | 2545/4722 [25:07:00<19:31:29, 32.29s/it]
 54%|█████▍    | 2546/4722 [25:07:39<20:49:17, 34.45s/it]
 54%|█████▍    | 2547/4722 [25:08:20<21:54:22, 36.26s/it]
 54%|█████▍    | 2548/4722 [25:08:52<21:10:58, 35.08s/it]
 54%|█████▍    | 2549/4722 [25:09:26<20:58:24, 34.75s/it]
 54%|█████▍    | 2550/4722 [25:09:58<20:27:25, 33.91s/it]
 54%|█████▍    | 2551/4722 [25:10:28<19:48:40, 32.85s/it]
 54%|█████▍    | 2552/4722 [25:10:57<19:06:04, 31.69s/it]
 54%|█████▍    | 2553/4722 [25:11:33<19:55:45, 33.08s/it]
 54%|█████▍    | 2554/4722 [25:12:03<19:19:46, 32.10s/it]
 54%|█████▍    | 2555/4722 [25:12:46<21:13:23, 35.26s/it]
 54%|█████▍    | 2556/4722 [25:13:17<20:32:04, 34.13s/it]
 54%|█████▍    | 2557/4722 [25:13:50<20:12:00, 33.59s/it]
 54%|█████▍    | 2558/4722 [25:14:21<19:41:14, 32.75s/it]
 54%|█████▍    | 2559/4722 [25:14:52<19:31:32, 32.50s/it]
 54%|█████▍    | 2560/4722 [25:15:28<20:00:11, 33.31s/it]
                                                         
{'loss': 0.4886, 'learning_rate': 0.0001, 'global_step': 2560, 'epoch': 1.63}
+
 54%|█████▍    | 2560/4722 [25:15:28<20:00:11, 33.31s/it]
 54%|█████▍    | 2561/4722 [25:16:01<20:03:21, 33.41s/it]
 54%|█████▍    | 2562/4722 [25:16:34<19:56:17, 33.23s/it]
 54%|█████▍    | 2563/4722 [25:17:04<19:18:42, 32.20s/it]
 54%|█████▍    | 2564/4722 [25:17:39<19:52:55, 33.17s/it]
 54%|█████▍    | 2565/4722 [25:18:10<19:30:55, 32.57s/it]
 54%|█████▍    | 2566/4722 [25:18:42<19:18:04, 32.23s/it]
 54%|█████▍    | 2567/4722 [25:19:14<19:16:52, 32.21s/it]
 54%|█████▍    | 2568/4722 [25:19:55<20:44:52, 34.68s/it]
 54%|█████▍    | 2569/4722 [25:20:31<21:07:39, 35.33s/it]
 54%|█████▍    | 2570/4722 [25:21:04<20:36:45, 34.48s/it]
 54%|█████▍    | 2571/4722 [25:21:36<20:08:40, 33.72s/it]
 54%|█████▍    | 2572/4722 [25:22:04<19:05:35, 31.97s/it]
 54%|█████▍    | 2573/4722 [25:22:37<19:21:50, 32.44s/it]
 55%|█████▍    | 2574/4722 [25:23:11<19:33:02, 32.77s/it]
 55%|█████▍    | 2575/4722 [25:23:40<18:57:48, 31.80s/it]
 55%|█████▍    | 2576/4722 [25:24:11<18:44:28, 31.44s/it]
 55%|█████▍    | 2577/4722 [25:24:45<19:15:24, 32.32s/it]
 55%|█████▍    | 2578/4722 [25:25:18<19:23:20, 32.56s/it]
 55%|█████▍    | 2579/4722 [25:26:00<20:57:43, 35.21s/it]
 55%|█████▍    | 2580/4722 [25:26:34<20:50:11, 35.02s/it]
                                                         
{'loss': 0.4613, 'learning_rate': 0.0001, 'global_step': 2580, 'epoch': 1.64}
+
 55%|█████▍    | 2580/4722 [25:26:34<20:50:11, 35.02s/it]
 55%|█████▍    | 2581/4722 [25:27:08<20:30:39, 34.49s/it]
 55%|█████▍    | 2582/4722 [25:27:47<21:20:11, 35.89s/it]
 55%|█████▍    | 2583/4722 [25:28:31<22:46:25, 38.33s/it]
 55%|█████▍    | 2584/4722 [25:29:04<21:48:56, 36.73s/it]
 55%|█████▍    | 2585/4722 [25:29:37<21:06:17, 35.55s/it]
 55%|█████▍    | 2586/4722 [25:30:09<20:35:36, 34.71s/it]
 55%|█████▍    | 2587/4722 [25:30:39<19:41:06, 33.19s/it]
 55%|█████▍    | 2588/4722 [25:31:10<19:15:48, 32.50s/it]
 55%|█████▍    | 2589/4722 [25:31:43<19:18:47, 32.60s/it]
 55%|█████▍    | 2590/4722 [25:32:25<20:58:00, 35.40s/it]
 55%|█████▍    | 2591/4722 [25:32:57<20:24:08, 34.47s/it]
 55%|█████▍    | 2592/4722 [25:33:32<20:34:08, 34.76s/it]
 55%|█████▍    | 2593/4722 [25:34:04<20:05:19, 33.97s/it]
 55%|█████▍    | 2594/4722 [25:34:31<18:48:51, 31.83s/it]
 55%|█████▍    | 2595/4722 [25:35:03<18:44:29, 31.72s/it]
 55%|█████▍    | 2596/4722 [25:35:39<19:29:30, 33.01s/it]
 55%|█████▍    | 2597/4722 [25:36:10<19:08:10, 32.42s/it]
 55%|█████▌    | 2598/4722 [25:36:41<18:56:49, 32.11s/it]
 55%|█████▌    | 2599/4722 [25:37:14<19:06:36, 32.41s/it]
 55%|█████▌    | 2600/4722 [25:37:50<19:44:10, 33.48s/it]
                                                         
{'loss': 0.4633, 'learning_rate': 0.0001, 'global_step': 2600, 'epoch': 1.65}
+
 55%|█████▌    | 2600/4722 [25:37:50<19:44:10, 33.48s/it]
 55%|█████▌    | 2601/4722 [25:38:22<19:21:22, 32.85s/it]
 55%|█████▌    | 2602/4722 [25:38:55<19:23:13, 32.92s/it]
 55%|█████▌    | 2603/4722 [25:39:26<19:09:31, 32.55s/it]
 55%|█████▌    | 2604/4722 [25:40:01<19:28:28, 33.10s/it]
 55%|█████▌    | 2605/4722 [25:40:28<18:27:20, 31.38s/it]
 55%|█████▌    | 2606/4722 [25:41:01<18:36:50, 31.67s/it]
 55%|█████▌    | 2607/4722 [25:41:35<19:10:18, 32.63s/it]
 55%|█████▌    | 2608/4722 [25:42:11<19:38:38, 33.45s/it]
 55%|█████▌    | 2609/4722 [25:42:48<20:20:29, 34.66s/it]
 55%|█████▌    | 2610/4722 [25:43:23<20:16:24, 34.56s/it]
 55%|█████▌    | 2611/4722 [25:43:53<19:29:26, 33.24s/it]
 55%|█████▌    | 2612/4722 [25:44:26<19:30:11, 33.28s/it]
 55%|█████▌    | 2613/4722 [25:44:59<19:26:23, 33.18s/it]
 55%|█████▌    | 2614/4722 [25:45:33<19:34:35, 33.43s/it]
 55%|█████▌    | 2615/4722 [25:46:07<19:34:23, 33.44s/it]
 55%|█████▌    | 2616/4722 [25:46:35<18:44:45, 32.04s/it]
 55%|█████▌    | 2617/4722 [25:47:06<18:28:30, 31.60s/it]
 55%|█████▌    | 2618/4722 [25:47:34<17:48:06, 30.46s/it]
 55%|█████▌    | 2619/4722 [25:48:07<18:22:02, 31.44s/it]
 55%|█████▌    | 2620/4722 [25:48:37<17:59:09, 30.80s/it]
                                                         
{'loss': 0.4782, 'learning_rate': 0.0001, 'global_step': 2620, 'epoch': 1.66}
+
 55%|█████▌    | 2620/4722 [25:48:37<17:59:09, 30.80s/it]
 56%|█████▌    | 2621/4722 [25:49:11<18:35:37, 31.86s/it]
 56%|█████▌    | 2622/4722 [25:49:43<18:37:50, 31.94s/it]
 56%|█████▌    | 2623/4722 [25:50:17<18:59:35, 32.58s/it]
 56%|█████▌    | 2624/4722 [25:50:51<19:13:03, 32.98s/it]
 56%|█████▌    | 2625/4722 [25:51:22<18:54:01, 32.45s/it]
 56%|█████▌    | 2626/4722 [25:51:56<19:10:02, 32.92s/it]
 56%|█████▌    | 2627/4722 [25:52:34<20:00:33, 34.38s/it]
 56%|█████▌    | 2628/4722 [25:53:09<20:08:04, 34.62s/it]
 56%|█████▌    | 2629/4722 [25:53:41<19:39:18, 33.81s/it]
 56%|█████▌    | 2630/4722 [25:54:13<19:15:23, 33.14s/it]
 56%|█████▌    | 2631/4722 [25:54:44<18:48:35, 32.38s/it]
 56%|█████▌    | 2632/4722 [25:55:22<19:53:10, 34.25s/it]
 56%|█████▌    | 2633/4722 [25:56:09<22:03:28, 38.01s/it]
 56%|█████▌    | 2634/4722 [25:56:56<23:35:30, 40.68s/it]
 56%|█████▌    | 2635/4722 [25:57:34<23:05:29, 39.83s/it]
 56%|█████▌    | 2636/4722 [25:58:05<21:40:26, 37.40s/it]
 56%|█████▌    | 2637/4722 [25:58:38<20:53:23, 36.07s/it]
 56%|█████▌    | 2638/4722 [25:59:18<21:26:20, 37.03s/it]
 56%|█████▌    | 2639/4722 [25:59:49<20:28:45, 35.39s/it]
 56%|█████▌    | 2640/4722 [26:00:21<19:54:17, 34.42s/it]
                                                         
{'loss': 0.4189, 'learning_rate': 0.0001, 'global_step': 2640, 'epoch': 1.68}
+
 56%|█████▌    | 2640/4722 [26:00:21<19:54:17, 34.42s/it]
 56%|█████▌    | 2641/4722 [26:00:58<20:20:30, 35.19s/it]
 56%|█████▌    | 2642/4722 [26:01:36<20:43:35, 35.87s/it]
 56%|█████▌    | 2643/4722 [26:02:03<19:12:58, 33.27s/it]
 56%|█████▌    | 2644/4722 [26:02:44<20:36:23, 35.70s/it]
 56%|█████▌    | 2645/4722 [26:03:23<21:07:12, 36.61s/it]
 56%|█████▌    | 2646/4722 [26:04:02<21:32:32, 37.36s/it]
 56%|█████▌    | 2647/4722 [26:04:40<21:32:29, 37.37s/it]
 56%|█████▌    | 2648/4722 [26:05:16<21:24:26, 37.16s/it]
 56%|█████▌    | 2649/4722 [26:05:48<20:26:38, 35.50s/it]
 56%|█████▌    | 2650/4722 [26:06:17<19:17:20, 33.51s/it]
 56%|█████▌    | 2651/4722 [26:06:51<19:22:46, 33.69s/it]
 56%|█████▌    | 2652/4722 [26:07:21<18:49:20, 32.73s/it]
 56%|█████▌    | 2653/4722 [26:07:55<18:58:17, 33.01s/it]
 56%|█████▌    | 2654/4722 [26:08:31<19:23:56, 33.77s/it]
 56%|█████▌    | 2655/4722 [26:09:07<19:53:12, 34.64s/it]
 56%|█████▌    | 2656/4722 [26:09:41<19:47:33, 34.49s/it]
 56%|█████▋    | 2657/4722 [26:10:21<20:36:57, 35.94s/it]
 56%|█████▋    | 2658/4722 [26:10:55<20:18:37, 35.43s/it]
 56%|█████▋    | 2659/4722 [26:11:34<20:57:52, 36.58s/it]
 56%|█████▋    | 2660/4722 [26:12:12<21:13:25, 37.05s/it]
                                                         
{'loss': 0.4752, 'learning_rate': 0.0001, 'global_step': 2660, 'epoch': 1.69}
+
 56%|█████▋    | 2660/4722 [26:12:12<21:13:25, 37.05s/it]
 56%|█████▋    | 2661/4722 [26:12:44<20:19:52, 35.51s/it]
 56%|█████▋    | 2662/4722 [26:13:25<21:13:55, 37.10s/it]
 56%|█████▋    | 2663/4722 [26:14:17<23:48:37, 41.63s/it]
 56%|█████▋    | 2664/4722 [26:14:48<21:55:57, 38.37s/it]
 56%|█████▋    | 2665/4722 [26:15:26<21:55:39, 38.38s/it]
 56%|█████▋    | 2666/4722 [26:16:02<21:30:03, 37.65s/it]
 56%|█████▋    | 2667/4722 [26:16:34<20:31:16, 35.95s/it]
 57%|█████▋    | 2668/4722 [26:17:07<19:55:53, 34.93s/it]
 57%|█████▋    | 2669/4722 [26:17:37<19:02:41, 33.40s/it]
 57%|█████▋    | 2670/4722 [26:18:13<19:28:43, 34.17s/it]
 57%|█████▋    | 2671/4722 [26:18:47<19:24:47, 34.08s/it]
 57%|█████▋    | 2672/4722 [26:19:31<21:10:55, 37.20s/it]
 57%|█████▋    | 2673/4722 [26:20:07<20:59:31, 36.88s/it]
 57%|█████▋    | 2674/4722 [26:20:40<20:13:48, 35.56s/it]
 57%|█████▋    | 2675/4722 [26:21:11<19:29:30, 34.28s/it]
 57%|█████▋    | 2676/4722 [26:21:52<20:34:13, 36.19s/it]
 57%|█████▋    | 2677/4722 [26:22:23<19:45:23, 34.78s/it]
 57%|█████▋    | 2678/4722 [26:22:57<19:35:23, 34.50s/it]
 57%|█████▋    | 2679/4722 [26:23:26<18:41:22, 32.93s/it]
 57%|█████▋    | 2680/4722 [26:24:05<19:36:37, 34.57s/it]
                                                         
{'loss': 0.4688, 'learning_rate': 0.0001, 'global_step': 2680, 'epoch': 1.7}
+
 57%|█████▋    | 2680/4722 [26:24:05<19:36:37, 34.57s/it]
 57%|█████▋    | 2681/4722 [26:24:46<20:42:26, 36.52s/it]
 57%|█████▋    | 2682/4722 [26:25:17<19:52:04, 35.06s/it]
 57%|█████▋    | 2683/4722 [26:25:51<19:31:47, 34.48s/it]
 57%|█████▋    | 2684/4722 [26:26:22<19:04:47, 33.70s/it]
 57%|██���██▋    | 2685/4722 [26:26:59<19:32:00, 34.52s/it]
 57%|█████▋    | 2686/4722 [26:27:35<19:51:39, 35.12s/it]
 57%|█████▋    | 2687/4722 [26:28:15<20:34:36, 36.40s/it]
 57%|█████▋    | 2688/4722 [26:28:50<20:19:41, 35.98s/it]
 57%|█████▋    | 2689/4722 [26:29:20<19:18:01, 34.18s/it]
 57%|█████▋    | 2690/4722 [26:29:55<19:32:44, 34.63s/it]
 57%|█████▋    | 2691/4722 [26:30:28<19:10:55, 34.00s/it]
 57%|█████▋    | 2692/4722 [26:31:01<18:59:51, 33.69s/it]
 57%|█████▋    | 2693/4722 [26:31:41<20:06:23, 35.67s/it]
 57%|█████▋    | 2694/4722 [26:32:13<19:29:16, 34.59s/it]
 57%|█████▋    | 2695/4722 [26:32:48<19:27:48, 34.57s/it]
 57%|█████▋    | 2696/4722 [26:33:21<19:15:29, 34.22s/it]
 57%|█████▋    | 2697/4722 [26:33:51<18:26:43, 32.79s/it]
 57%|█████▋    | 2698/4722 [26:34:23<18:23:11, 32.70s/it]
 57%|█████▋    | 2699/4722 [26:34:55<18:12:30, 32.40s/it]
 57%|█████▋    | 2700/4722 [26:35:27<18:04:48, 32.19s/it]
                                                         
{'loss': 0.5238, 'learning_rate': 0.0001, 'global_step': 2700, 'epoch': 1.72}
+
 57%|█████▋    | 2700/4722 [26:35:27<18:04:48, 32.19s/it]
 57%|█████▋    | 2701/4722 [26:36:05<19:07:35, 34.07s/it]
 57%|█████▋    | 2702/4722 [26:36:48<20:39:34, 36.82s/it]
 57%|█████▋    | 2703/4722 [26:37:26<20:49:15, 37.13s/it]
 57%|█████▋    | 2704/4722 [26:37:57<19:43:46, 35.20s/it]
 57%|█████▋    | 2705/4722 [26:38:31<19:35:14, 34.96s/it]
 57%|█████▋    | 2706/4722 [26:39:17<21:27:19, 38.31s/it]
 57%|█████▋    | 2707/4722 [26:40:07<23:25:14, 41.84s/it]
 57%|█████▋    | 2708/4722 [26:40:58<24:50:42, 44.41s/it]
 57%|█████▋    | 2709/4722 [26:41:46<25:25:12, 45.46s/it]
 57%|█████▋    | 2710/4722 [26:42:31<25:23:01, 45.42s/it]
 57%|█████▋    | 2711/4722 [26:43:24<26:41:36, 47.79s/it]
 57%|█████▋    | 2712/4722 [26:44:15<27:08:23, 48.61s/it]
 57%|█████▋    | 2713/4722 [26:45:02<26:49:15, 48.06s/it]
 57%|█████▋    | 2714/4722 [26:45:58<28:08:50, 50.46s/it]
 57%|█████▋    | 2715/4722 [26:46:48<28:07:34, 50.45s/it]
 58%|█████▊    | 2716/4722 [26:47:43<28:53:00, 51.83s/it]
 58%|█████▊    | 2717/4722 [26:48:28<27:44:20, 49.81s/it]
 58%|█████▊    | 2718/4722 [26:49:18<27:41:05, 49.73s/it]
 58%|█████▊    | 2719/4722 [26:50:16<29:08:46, 52.38s/it]
 58%|█████▊    | 2720/4722 [26:51:13<29:45:43, 53.52s/it]
                                                         
{'loss': 0.5027, 'learning_rate': 0.0001, 'global_step': 2720, 'epoch': 1.73}
+
 58%|█████▊    | 2720/4722 [26:51:13<29:45:43, 53.52s/it]
 58%|█████▊    | 2721/4722 [26:52:01<28:57:12, 52.09s/it]
 58%|█████▊    | 2722/4722 [26:52:44<27:17:26, 49.12s/it]
 58%|█████▊    | 2723/4722 [26:53:25<25:58:54, 46.79s/it]
 58%|█████▊    | 2724/4722 [26:54:15<26:31:44, 47.80s/it]
 58%|█████▊    | 2725/4722 [26:55:09<27:28:20, 49.52s/it]
 58%|█████▊    | 2726/4722 [26:55:57<27:18:27, 49.25s/it]
 58%|█████▊    | 2727/4722 [26:56:42<26:32:52, 47.91s/it]
 58%|█████▊    | 2728/4722 [26:57:33<27:02:03, 48.81s/it]
 58%|█████▊    | 2729/4722 [26:58:27<27:52:37, 50.36s/it]
 58%|█████▊    | 2730/4722 [26:59:15<27:28:23, 49.65s/it]
 58%|█████▊    | 2731/4722 [27:00:07<27:48:15, 50.27s/it]
 58%|█████▊    | 2732/4722 [27:00:52<26:57:18, 48.76s/it]
 58%|█████▊    | 2733/4722 [27:01:39<26:43:33, 48.37s/it]
 58%|█████▊    | 2734/4722 [27:02:27<26:34:04, 48.11s/it]
 58%|█████▊    | 2735/4722 [27:03:12<26:09:33, 47.39s/it]
 58%|█████▊    | 2736/4722 [27:03:56<25:30:15, 46.23s/it]
 58%|█████▊    | 2737/4722 [27:04:37<24:37:26, 44.66s/it]
 58%|█████▊    | 2738/4722 [27:05:16<23:38:07, 42.89s/it]
 58%|█████▊    | 2739/4722 [27:05:59<23:39:38, 42.95s/it]
 58%|█████▊    | 2740/4722 [27:06:55<25:52:03, 46.98s/it]
                                                         
{'loss': 0.4318, 'learning_rate': 0.0001, 'global_step': 2740, 'epoch': 1.74}
+
 58%|█████▊    | 2740/4722 [27:06:55<25:52:03, 46.98s/it]
 58%|█████▊    | 2741/4722 [27:07:45<26:21:22, 47.90s/it]
 58%|█████▊    | 2742/4722 [27:08:25<24:59:11, 45.43s/it]
 58%|█████▊    | 2743/4722 [27:09:05<24:02:15, 43.73s/it]
 58%|█████▊    | 2744/4722 [27:09:56<25:15:44, 45.98s/it]
 58%|█████▊    | 2745/4722 [27:10:45<25:41:31, 46.78s/it]
 58%|█████▊    | 2746/4722 [27:11:36<26:29:54, 48.28s/it]
 58%|█████▊    | 2747/4722 [27:12:29<27:08:37, 49.48s/it]
 58%|█████▊    | 2748/4722 [27:13:23<27:53:05, 50.85s/it]
 58%|█████▊    | 2749/4722 [27:14:05<26:28:22, 48.30s/it]
 58%|█████▊    | 2750/4722 [27:14:55<26:44:37, 48.82s/it]
 58%|█████▊    | 2751/4722 [27:15:46<27:02:59, 49.41s/it]
 58%|█████▊    | 2752/4722 [27:16:30<26:06:05, 47.70s/it]
 58%|█████▊    | 2753/4722 [27:17:20<26:35:06, 48.61s/it]
 58%|█████▊    | 2754/4722 [27:18:16<27:48:06, 50.86s/it]
 58%|█████▊    | 2755/4722 [27:19:16<29:10:34, 53.40s/it]
 58%|█████▊    | 2756/4722 [27:20:05<28:29:57, 52.19s/it]
 58%|█████▊    | 2757/4722 [27:21:09<30:27:09, 55.79s/it]
 58%|█████▊    | 2758/4722 [27:21:57<29:08:56, 53.43s/it]
 58%|█████▊    | 2759/4722 [27:22:43<27:53:13, 51.14s/it]
 58%|█████▊    | 2760/4722 [27:23:34<27:55:37, 51.24s/it]
                                                         
{'loss': 0.4194, 'learning_rate': 0.0001, 'global_step': 2760, 'epoch': 1.75}
+
 58%|█████▊    | 2760/4722 [27:23:34<27:55:37, 51.24s/it]
 58%|█████▊    | 2761/4722 [27:24:19<26:53:41, 49.37s/it]
 58%|█████▊    | 2762/4722 [27:24:58<25:09:03, 46.20s/it]
 59%|█████▊    | 2763/4722 [27:25:41<24:35:32, 45.19s/it]
 59%|█████▊    | 2764/4722 [27:26:31<25:23:10, 46.68s/it]
 59%|█████▊    | 2765/4722 [27:27:29<27:10:12, 49.98s/it]
 59%|█████▊    | 2766/4722 [27:28:09<25:29:17, 46.91s/it]
 59%|█████▊    | 2767/4722 [27:28:57<25:38:26, 47.22s/it]
 59%|█████▊    | 2768/4722 [27:29:54<27:17:02, 50.27s/it]
 59%|█████▊    | 2769/4722 [27:30:45<27:22:03, 50.45s/it]
 59%|█████▊    | 2770/4722 [27:31:30<26:31:21, 48.91s/it]
 59%|█████▊    | 2771/4722 [27:32:23<27:04:04, 49.95s/it]
 59%|█████▊    | 2772/4722 [27:33:13<27:07:20, 50.07s/it]
 59%|█████▊    | 2773/4722 [27:34:09<28:07:10, 51.94s/it]
 59%|█████▊    | 2774/4722 [27:35:07<29:02:25, 53.67s/it]
 59%|█████▉    | 2775/4722 [27:35:59<28:48:22, 53.26s/it]
 59%|█████▉    | 2776/4722 [27:36:49<28:16:21, 52.30s/it]
 59%|█████▉    | 2777/4722 [27:37:29<26:12:55, 48.52s/it]
 59%|█████▉    | 2778/4722 [27:38:26<27:37:18, 51.15s/it]
 59%|█████▉    | 2779/4722 [27:39:12<26:44:36, 49.55s/it]
 59%|█████▉    | 2780/4722 [27:40:02<26:49:42, 49.73s/it]
                                                         
{'loss': 0.4614, 'learning_rate': 0.0001, 'global_step': 2780, 'epoch': 1.77}
+
 59%|█████▉    | 2780/4722 [27:40:02<26:49:42, 49.73s/it]
 59%|█████▉    | 2781/4722 [27:40:46<25:54:13, 48.04s/it]
 59%|█████▉    | 2782/4722 [27:41:32<25:30:13, 47.33s/it]
 59%|█████▉    | 2783/4722 [27:42:19<25:29:04, 47.32s/it]
 59%|█████▉    | 2784/4722 [27:43:12<26:16:47, 48.82s/it]
 59%|█████▉    | 2785/4722 [27:43:58<25:47:54, 47.95s/it]
 59%|█████▉    | 2786/4722 [27:44:54<27:11:08, 50.55s/it]
 59%|█████▉    | 2787/4722 [27:45:39<26:15:23, 48.85s/it]
 59%|█████▉    | 2788/4722 [27:46:12<23:41:20, 44.10s/it]
 59%|█████▉    | 2789/4722 [27:46:44<21:39:33, 40.34s/it]
 59%|█████▉    | 2790/4722 [27:47:22<21:17:07, 39.66s/it]
 59%|█████▉    | 2791/4722 [27:47:54<20:07:11, 37.51s/it]
 59%|█████▉    | 2792/4722 [27:48:27<19:20:21, 36.07s/it]
 59%|█████▉    | 2793/4722 [27:49:02<19:09:15, 35.75s/it]
 59%|█████▉    | 2794/4722 [27:49:34<18:34:08, 34.67s/it]
 59%|█████▉    | 2795/4722 [27:50:12<19:06:21, 35.69s/it]
 59%|█████▉    | 2796/4722 [27:50:50<19:30:56, 36.48s/it]
 59%|█████▉    | 2797/4722 [27:51:28<19:38:01, 36.72s/it]
 59%|█████▉    | 2798/4722 [27:52:03<19:27:38, 36.41s/it]
 59%|█████▉    | 2799/4722 [27:52:34<18:31:41, 34.69s/it]
 59%|█████▉    | 2800/4722 [27:53:09<18:36:54, 34.87s/it]
                                                         
{'loss': 0.4679, 'learning_rate': 0.0001, 'global_step': 2800, 'epoch': 1.78}
+
 59%|█████▉    | 2800/4722 [27:53:09<18:36:54, 34.87s/it]
 59%|█████▉    | 2801/4722 [27:53:47<19:06:27, 35.81s/it]
 59%|█████▉    | 2802/4722 [27:54:24<19:15:35, 36.11s/it]
 59%|█████▉    | 2803/4722 [27:54:53<18:08:53, 34.05s/it]
 59%|█████▉    | 2804/4722 [27:55:23<17:25:37, 32.71s/it]
 59%|█████▉    | 2805/4722 [27:56:02<18:21:05, 34.46s/it]
 59%|█████▉    | 2806/4722 [27:56:32<17:44:56, 33.35s/it]
 59%|█████▉    | 2807/4722 [27:57:04<17:24:51, 32.74s/it]
 59%|█████▉    | 2808/4722 [27:57:35<17:11:31, 32.34s/it]
 59%|█████▉    | 2809/4722 [27:58:08<17:12:42, 32.39s/it]
 60%|█████▉    | 2810/4722 [27:58:46<18:10:50, 34.23s/it]
 60%|█████▉    | 2811/4722 [27:59:17<17:34:34, 33.11s/it]
 60%|█████▉    | 2812/4722 [27:59:49<17:27:44, 32.91s/it]
 60%|█████▉    | 2813/4722 [28:00:28<18:24:03, 34.70s/it]
 60%|█████▉    | 2814/4722 [28:01:01<18:07:21, 34.19s/it]
 60%|█████▉    | 2815/4722 [28:01:34<17:58:18, 33.93s/it]
 60%|█████▉    | 2816/4722 [28:02:14<18:50:09, 35.58s/it]
 60%|█████▉    | 2817/4722 [28:02:57<20:00:22, 37.81s/it]
 60%|█████▉    | 2818/4722 [28:03:34<19:56:25, 37.70s/it]
 60%|█████▉    | 2819/4722 [28:04:14<20:19:31, 38.45s/it]
 60%|█████▉    | 2820/4722 [28:04:48<19:28:44, 36.87s/it]
                                                         
{'loss': 0.4475, 'learning_rate': 0.0001, 'global_step': 2820, 'epoch': 1.79}
+
 60%|█████▉    | 2820/4722 [28:04:48<19:28:44, 36.87s/it]
 60%|█████▉    | 2821/4722 [28:05:22<19:03:45, 36.10s/it]
 60%|█████▉    | 2822/4722 [28:06:03<19:55:01, 37.74s/it]
 60%|█████▉    | 2823/4722 [28:06:38<19:24:15, 36.79s/it]
 60%|█████▉    | 2824/4722 [28:07:10<18:43:00, 35.50s/it]
 60%|█████▉    | 2825/4722 [28:07:42<18:07:20, 34.39s/it]
 60%|█████▉    | 2826/4722 [28:08:17<18:09:29, 34.48s/it]
 60%|█████▉    | 2827/4722 [28:08:51<18:08:58, 34.48s/it]
 60%|█████▉    | 2828/4722 [28:09:28<18:27:41, 35.09s/it]
 60%|█████▉    | 2829/4722 [28:10:04<18:39:14, 35.48s/it]
 60%|█████▉    | 2830/4722 [28:10:35<17:54:55, 34.09s/it]
 60%|█████▉    | 2831/4722 [28:11:13<18:30:47, 35.24s/it]
 60%|█████▉    | 2832/4722 [28:11:47<18:13:13, 34.71s/it]
 60%|█████▉    | 2833/4722 [28:12:17<17:29:05, 33.32s/it]
 60%|██████    | 2834/4722 [28:12:50<17:27:42, 33.30s/it]
 60%|██████    | 2835/4722 [28:13:23<17:21:32, 33.12s/it]
 60%|██████    | 2836/4722 [28:13:54<17:04:56, 32.61s/it]
 60%|██████    | 2837/4722 [28:14:29<17:23:28, 33.21s/it]
 60%|██████    | 2838/4722 [28:15:04<17:46:20, 33.96s/it]
 60%|██████    | 2839/4722 [28:15:38<17:44:25, 33.92s/it]
 60%|██████    | 2840/4722 [28:16:13<17:53:55, 34.24s/it]
                                                         
{'loss': 0.4803, 'learning_rate': 0.0001, 'global_step': 2840, 'epoch': 1.8}
+
 60%|██████    | 2840/4722 [28:16:13<17:53:55, 34.24s/it]
 60%|██████    | 2841/4722 [28:16:45<17:29:47, 33.49s/it]
 60%|██████    | 2842/4722 [28:17:19<17:36:57, 33.73s/it]
 60%|██████    | 2843/4722 [28:17:55<17:51:44, 34.22s/it]
 60%|██████    | 2844/4722 [28:18:32<18:18:16, 35.09s/it]
 60%|██████    | 2845/4722 [28:19:04<17:52:50, 34.29s/it]
 60%|██████    | 2846/4722 [28:19:39<17:54:09, 34.35s/it]
 60%|██████    | 2847/4722 [28:20:20<18:56:22, 36.36s/it]
 60%|██████    | 2848/4722 [28:20:48<17:40:12, 33.94s/it]
 60%|██████    | 2849/4722 [28:21:24<17:55:29, 34.45s/it]
 60%|██████    | 2850/4722 [28:21:56<17:38:26, 33.92s/it]
 60%|██████    | 2851/4722 [28:22:35<18:19:45, 35.27s/it]
 60%|██████    | 2852/4722 [28:23:09<18:10:24, 34.99s/it]
 60%|██████    | 2853/4722 [28:23:42<17:47:30, 34.27s/it]
 60%|██████    | 2854/4722 [28:24:12<17:14:41, 33.23s/it]
 60%|██████    | 2855/4722 [28:24:47<17:30:15, 33.75s/it]
 60%|██████    | 2856/4722 [28:25:20<17:23:51, 33.56s/it]
 61%|██████    | 2857/4722 [28:25:57<17:50:39, 34.44s/it]
 61%|██████    | 2858/4722 [28:26:30<17:40:29, 34.14s/it]
 61%|██████    | 2859/4722 [28:27:08<18:12:02, 35.17s/it]
 61%|██████    | 2860/4722 [28:27:49<19:02:26, 36.81s/it]
                                                         
{'loss': 0.4865, 'learning_rate': 0.0001, 'global_step': 2860, 'epoch': 1.82}
+
 61%|██████    | 2860/4722 [28:27:49<19:02:26, 36.81s/it]
 61%|██████    | 2861/4722 [28:28:21<18:22:45, 35.55s/it]
 61%|██████    | 2862/4722 [28:28:55<18:02:25, 34.92s/it]
 61%|██████    | 2863/4722 [28:29:21<16:46:28, 32.48s/it]
 61%|██████    | 2864/4722 [28:29:51<16:13:49, 31.45s/it]
 61%|██████    | 2865/4722 [28:30:25<16:43:49, 32.43s/it]
 61%|██████    | 2866/4722 [28:31:01<17:17:58, 33.56s/it]
 61%|██████    | 2867/4722 [28:31:35<17:18:49, 33.60s/it]
 61%|██████    | 2868/4722 [28:32:07<16:58:49, 32.97s/it]
 61%|██████    | 2869/4722 [28:32:39<16:52:29, 32.78s/it]
 61%|██████    | 2870/4722 [28:33:12<16:58:22, 32.99s/it]
 61%|██████    | 2871/4722 [28:33:42<16:25:44, 31.95s/it]
 61%|██████    | 2872/4722 [28:34:12<16:04:19, 31.28s/it]
 61%|██████    | 2873/4722 [28:34:47<16:39:37, 32.44s/it]
 61%|██████    | 2874/4722 [28:35:20<16:44:21, 32.61s/it]
 61%|██████    | 2875/4722 [28:35:56<17:18:49, 33.75s/it]
 61%|██████    | 2876/4722 [28:36:27<16:55:02, 32.99s/it]
 61%|██████    | 2877/4722 [28:37:01<17:02:17, 33.25s/it]
 61%|██████    | 2878/4722 [28:37:32<16:33:49, 32.34s/it]
 61%|██████    | 2879/4722 [28:38:08<17:08:55, 33.50s/it]
 61%|██████    | 2880/4722 [28:38:41<17:02:29, 33.31s/it]
                                                         
{'loss': 0.4418, 'learning_rate': 0.0001, 'global_step': 2880, 'epoch': 1.83}
+
 61%|██████    | 2880/4722 [28:38:41<17:02:29, 33.31s/it]
 61%|██████    | 2881/4722 [28:39:16<17:21:35, 33.95s/it]
 61%|██████    | 2882/4722 [28:39:54<18:00:04, 35.22s/it]
 61%|██████    | 2883/4722 [28:40:29<17:51:12, 34.95s/it]
 61%|██████    | 2884/4722 [28:41:03<17:48:05, 34.87s/it]
 61%|██████    | 2885/4722 [28:41:44<18:43:58, 36.71s/it]
 61%|██████    | 2886/4722 [28:42:18<18:12:29, 35.70s/it]
 61%|██████    | 2887/4722 [28:42:52<17:59:01, 35.28s/it]
 61%|██████    | 2888/4722 [28:43:20<16:56:01, 33.24s/it]
 61%|██████    | 2889/4722 [28:43:54<17:01:09, 33.43s/it]
 61%|██████    | 2890/4722 [28:44:28<17:06:18, 33.61s/it]
 61%|██████    | 2891/4722 [28:45:06<17:39:33, 34.72s/it]
 61%|██████    | 2892/4722 [28:45:38<17:22:14, 34.17s/it]
 61%|██████▏   | 2893/4722 [28:46:10<16:56:52, 33.36s/it]
 61%|██████▏   | 2894/4722 [28:46:37<15:58:31, 31.46s/it]
 61%|██████▏   | 2895/4722 [28:47:13<16:39:40, 32.83s/it]
 61%|██████▏   | 2896/4722 [28:47:44<16:18:06, 32.14s/it]
 61%|██████▏   | 2897/4722 [28:48:24<17:36:16, 34.73s/it]
 61%|██████▏   | 2898/4722 [28:48:58<17:24:19, 34.35s/it]
 61%|██████▏   | 2899/4722 [28:49:27<16:37:29, 32.83s/it]
 61%|██████▏   | 2900/4722 [28:50:01<16:42:59, 33.03s/it]
                                                         
{'loss': 0.4205, 'learning_rate': 0.0001, 'global_step': 2900, 'epoch': 1.84}
+
 61%|██████▏   | 2900/4722 [28:50:01<16:42:59, 33.03s/it]
 61%|██████▏   | 2901/4722 [28:50:35<16:57:58, 33.54s/it]
 61%|██████▏   | 2902/4722 [28:51:16<18:01:58, 35.67s/it]
 61%|██████▏   | 2903/4722 [28:51:48<17:25:27, 34.48s/it]
 61%|██████▏   | 2904/4722 [28:52:28<18:18:49, 36.27s/it]
 62%|██████▏   | 2905/4722 [28:53:04<18:15:21, 36.17s/it]
 62%|██████▏   | 2906/4722 [28:53:44<18:52:39, 37.42s/it]
 62%|██████▏   | 2907/4722 [28:54:18<18:15:52, 36.23s/it]
 62%|██████▏   | 2908/4722 [28:54:48<17:22:23, 34.48s/it]
 62%|██████▏   | 2909/4722 [28:55:25<17:40:18, 35.09s/it]
 62%|██████▏   | 2910/4722 [28:55:56<17:09:56, 34.10s/it]
 62%|██████▏   | 2911/4722 [28:56:30<16:59:47, 33.79s/it]
 62%|██████▏   | 2912/4722 [28:57:06<17:21:02, 34.51s/it]
 62%|██████▏   | 2913/4722 [28:57:39<17:09:41, 34.15s/it]
 62%|██████▏   | 2914/4722 [28:58:14<17:14:09, 34.32s/it]
 62%|██████▏   | 2915/4722 [28:58:46<16:57:15, 33.78s/it]
 62%|██████▏   | 2916/4722 [28:59:16<16:17:00, 32.46s/it]
 62%|██████▏   | 2917/4722 [28:59:52<16:53:28, 33.69s/it]
 62%|██████▏   | 2918/4722 [29:00:23<16:24:16, 32.74s/it]
 62%|██████▏   | 2919/4722 [29:00:59<16:53:51, 33.74s/it]
 62%|██████▏   | 2920/4722 [29:01:27<16:01:40, 32.02s/it]
                                                         
{'loss': 0.4326, 'learning_rate': 0.0001, 'global_step': 2920, 'epoch': 1.85}
+
 62%|██████▏   | 2920/4722 [29:01:27<16:01:40, 32.02s/it]
 62%|██████▏   | 2921/4722 [29:02:02<16:29:23, 32.96s/it]
 62%|██████▏   | 2922/4722 [29:02:36<16:36:15, 33.21s/it]
 62%|██████▏   | 2923/4722 [29:03:09<16:37:33, 33.27s/it]
 62%|██████▏   | 2924/4722 [29:03:41<16:20:56, 32.73s/it]
 62%|██████▏   | 2925/4722 [29:04:09<15:38:50, 31.35s/it]
 62%|██████▏   | 2926/4722 [29:04:39<15:29:27, 31.05s/it]
 62%|██████▏   | 2927/4722 [29:05:16<16:25:14, 32.93s/it]
 62%|██████▏   | 2928/4722 [29:05:51<16:42:37, 33.53s/it]
 62%|██████▏   | 2929/4722 [29:06:24<16:36:47, 33.36s/it]
 62%|██████▏   | 2930/4722 [29:06:56<16:22:53, 32.91s/it]
 62%|██████▏   | 2931/4722 [29:07:28<16:13:35, 32.62s/it]
 62%|██████▏   | 2932/4722 [29:07:56<15:33:19, 31.28s/it]
 62%|██████▏   | 2933/4722 [29:08:35<16:36:02, 33.41s/it]
 62%|██████▏   | 2934/4722 [29:09:07<16:21:35, 32.94s/it]
 62%|██████▏   | 2935/4722 [29:09:42<16:40:52, 33.61s/it]
 62%|██████▏   | 2936/4722 [29:10:15<16:41:12, 33.64s/it]
 62%|██████▏   | 2937/4722 [29:10:46<16:15:02, 32.77s/it]
 62%|██████▏   | 2938/4722 [29:11:22<16:39:56, 33.63s/it]
 62%|██████▏   | 2939/4722 [29:11:55<16:39:47, 33.64s/it]
 62%|██████▏   | 2940/4722 [29:12:24<15:49:58, 31.99s/it]
                                                         
{'loss': 0.4419, 'learning_rate': 0.0001, 'global_step': 2940, 'epoch': 1.87}
+
 62%|██████▏   | 2940/4722 [29:12:24<15:49:58, 31.99s/it]
 62%|██████▏   | 2941/4722 [29:13:07<17:35:09, 35.55s/it]
 62%|██████▏   | 2942/4722 [29:13:37<16:40:32, 33.73s/it]
 62%|██████▏   | 2943/4722 [29:14:14<17:07:59, 34.67s/it]
 62%|██████▏   | 2944/4722 [29:14:50<17:24:55, 35.26s/it]
 62%|██████▏   | 2945/4722 [29:15:21<16:41:45, 33.82s/it]
 62%|██████▏   | 2946/4722 [29:16:06<18:20:10, 37.17s/it]
 62%|██████▏   | 2947/4722 [29:16:39<17:40:04, 35.83s/it]
 62%|██████▏   | 2948/4722 [29:17:10<16:56:14, 34.37s/it]
 62%|██████▏   | 2949/4722 [29:17:38<16:01:24, 32.53s/it]
 62%|██████▏   | 2950/4722 [29:18:08<15:42:07, 31.90s/it]
 62%|██████▏   | 2951/4722 [29:18:40<15:37:00, 31.75s/it]
 63%|██████▎   | 2952/4722 [29:19:11<15:33:34, 31.65s/it]
 63%|██████▎   | 2953/4722 [29:19:50<16:33:49, 33.71s/it]
 63%|██████▎   | 2954/4722 [29:20:24<16:40:08, 33.94s/it]
 63%|██████▎   | 2955/4722 [29:21:00<16:57:12, 34.54s/it]
 63%|██████▎   | 2956/4722 [29:21:31<16:27:25, 33.55s/it]
 63%|██████▎   | 2957/4722 [29:22:06<16:38:35, 33.95s/it]
 63%|██████▎   | 2958/4722 [29:22:36<15:58:50, 32.61s/it]
 63%|██████▎   | 2959/4722 [29:23:13<16:43:42, 34.16s/it]
 63%|██████▎   | 2960/4722 [29:23:40<15:41:40, 32.07s/it]
                                                         
{'loss': 0.4743, 'learning_rate': 0.0001, 'global_step': 2960, 'epoch': 1.88}
+
 63%|██████▎   | 2960/4722 [29:23:40<15:41:40, 32.07s/it]
 63%|██████▎   | 2961/4722 [29:24:15<16:03:00, 32.81s/it]
 63%|██████▎   | 2962/4722 [29:24:50<16:25:13, 33.59s/it]
 63%|██████▎   | 2963/4722 [29:25:22<16:05:30, 32.93s/it]
 63%|██████▎   | 2964/4722 [29:25:52<15:40:54, 32.11s/it]
 63%|██████▎   | 2965/4722 [29:26:27<16:01:33, 32.84s/it]
 63%|██████▎   | 2966/4722 [29:26:55<15:23:22, 31.55s/it]
 63%|██████▎   | 2967/4722 [29:27:28<15:31:10, 31.83s/it]
 63%|██████▎   | 2968/4722 [29:27:57<15:07:56, 31.06s/it]
 63%|██████▎   | 2969/4722 [29:28:29<15:18:26, 31.44s/it]
 63%|██████▎   | 2970/4722 [29:29:03<15:37:34, 32.11s/it]
 63%|██████▎   | 2971/4722 [29:29:37<15:52:51, 32.65s/it]
 63%|██████▎   | 2972/4722 [29:30:13<16:21:10, 33.64s/it]
 63%|██████▎   | 2973/4722 [29:30:40<15:26:17, 31.78s/it]
 63%|██████▎   | 2974/4722 [29:31:14<15:43:02, 32.37s/it]
 63%|██████▎   | 2975/4722 [29:31:52<16:34:31, 34.16s/it]
 63%|██████▎   | 2976/4722 [29:32:23<16:00:41, 33.01s/it]
 63%|██████▎   | 2977/4722 [29:32:56<16:06:09, 33.22s/it]
 63%|██████▎   | 2978/4722 [29:33:27<15:45:30, 32.53s/it]
 63%|██████▎   | 2979/4722 [29:33:54<14:53:41, 30.76s/it]
 63%|██████▎   | 2980/4722 [29:34:27<15:17:18, 31.60s/it]
                                                         
{'loss': 0.4763, 'learning_rate': 0.0001, 'global_step': 2980, 'epoch': 1.89}
+
 63%|██████▎   | 2980/4722 [29:34:27<15:17:18, 31.60s/it]
 63%|██████▎   | 2981/4722 [29:35:03<15:54:04, 32.88s/it]
 63%|██████▎   | 2982/4722 [29:35:34<15:31:02, 32.10s/it]
 63%|██████▎   | 2983/4722 [29:36:05<15:24:07, 31.88s/it]
 63%|██████▎   | 2984/4722 [29:36:36<15:12:46, 31.51s/it]
 63%|██████▎   | 2985/4722 [29:37:15<16:21:26, 33.90s/it]
 63%|██████▎   | 2986/4722 [29:37:49<16:17:36, 33.79s/it]
 63%|██████▎   | 2987/4722 [29:38:22<16:10:48, 33.57s/it]
 63%|██████▎   | 2988/4722 [29:38:57<16:27:03, 34.15s/it]
 63%|██████▎   | 2989/4722 [29:39:30<16:14:38, 33.74s/it]
 63%|██████▎   | 2990/4722 [29:40:04<16:15:46, 33.80s/it]
 63%|██████▎   | 2991/4722 [29:40:37<16:06:35, 33.50s/it]
 63%|██████▎   | 2992/4722 [29:41:12<16:17:54, 33.92s/it]
 63%|██████▎   | 2993/4722 [29:41:39<15:22:50, 32.02s/it]
 63%|██████▎   | 2994/4722 [29:42:13<15:36:14, 32.51s/it]
 63%|██████▎   | 2995/4722 [29:42:44<15:23:46, 32.09s/it]
 63%|██████▎   | 2996/4722 [29:43:25<16:37:30, 34.68s/it]
 63%|██████▎   | 2997/4722 [29:43:56<16:05:13, 33.57s/it]
 63%|██████▎   | 2998/4722 [29:44:31<16:17:02, 34.00s/it]
 64%|██████▎   | 2999/4722 [29:45:02<15:57:23, 33.34s/it]
 64%|██████▎   | 3000/4722 [29:45:37<16:11:31, 33.85s/it]
                                                         
{'loss': 0.4424, 'learning_rate': 0.0001, 'global_step': 3000, 'epoch': 1.91}
+
 64%|█���████▎   | 3000/4722 [29:45:37<16:11:31, 33.85s/it]
 64%|██████▎   | 3001/4722 [29:46:13<16:29:25, 34.49s/it]
 64%|██████▎   | 3002/4722 [29:46:43<15:47:55, 33.07s/it]
 64%|██████▎   | 3003/4722 [29:47:14<15:26:35, 32.34s/it]
 64%|██████▎   | 3004/4722 [29:47:43<15:02:10, 31.51s/it]
 64%|██████▎   | 3005/4722 [29:48:24<16:20:11, 34.25s/it]
 64%|██████▎   | 3006/4722 [29:48:57<16:05:24, 33.76s/it]
 64%|██████▎   | 3007/4722 [29:49:31<16:08:56, 33.90s/it]
 64%|██████▎   | 3008/4722 [29:50:03<15:50:33, 33.28s/it]
 64%|██████▎   | 3009/4722 [29:50:38<16:05:22, 33.81s/it]
 64%|██████▎   | 3010/4722 [29:51:10<15:51:30, 33.35s/it]
 64%|██████▍   | 3011/4722 [29:51:45<16:03:24, 33.78s/it]
 64%|██████▍   | 3012/4722 [29:52:26<17:01:43, 35.85s/it]
 64%|██████▍   | 3013/4722 [29:52:57<16:19:54, 34.40s/it]
 64%|██████▍   | 3014/4722 [29:53:30<16:15:06, 34.25s/it]
 64%|██████▍   | 3015/4722 [29:54:05<16:15:48, 34.30s/it]
 64%|██████▍   | 3016/4722 [29:54:31<15:05:17, 31.84s/it]
 64%|██████▍   | 3017/4722 [29:55:03<15:04:51, 31.84s/it]
 64%|██████▍   | 3018/4722 [29:55:36<15:13:43, 32.17s/it]
 64%|██████▍   | 3019/4722 [29:56:06<14:53:15, 31.47s/it]
 64%|██████▍   | 3020/4722 [29:56:38<14:56:35, 31.61s/it]
                                                         
{'loss': 0.4982, 'learning_rate': 0.0001, 'global_step': 3020, 'epoch': 1.92}
+
 64%|██████▍   | 3020/4722 [29:56:38<14:56:35, 31.61s/it]
 64%|██████▍   | 3021/4722 [29:57:05<14:24:37, 30.50s/it]
 64%|██████▍   | 3022/4722 [29:57:38<14:41:38, 31.12s/it]
 64%|██████▍   | 3023/4722 [29:58:10<14:50:32, 31.45s/it]
 64%|██████▍   | 3024/4722 [29:58:46<15:25:07, 32.69s/it]
 64%|██████▍   | 3025/4722 [29:59:30<16:59:23, 36.04s/it]
 64%|██████▍   | 3026/4722 [30:00:00<16:11:57, 34.39s/it]
 64%|██████▍   | 3027/4722 [30:00:33<15:55:57, 33.84s/it]
 64%|██████▍   | 3028/4722 [30:01:07<16:01:13, 34.05s/it]
 64%|██████▍   | 3029/4722 [30:01:38<15:34:00, 33.10s/it]
 64%|██████▍   | 3030/4722 [30:02:10<15:21:29, 32.68s/it]
 64%|██████▍   | 3031/4722 [30:02:41<15:07:28, 32.20s/it]
 64%|██████▍   | 3032/4722 [30:03:14<15:17:59, 32.59s/it]
 64%|██████▍   | 3033/4722 [30:03:56<16:31:43, 35.23s/it]
 64%|██████▍   | 3034/4722 [30:04:29<16:10:40, 34.50s/it]
 64%|██████▍   | 3035/4722 [30:05:03<16:05:56, 34.35s/it]
 64%|██████▍   | 3036/4722 [30:05:37<16:02:11, 34.24s/it]
 64%|██████▍   | 3037/4722 [30:06:11<16:04:46, 34.35s/it]
 64%|██████▍   | 3038/4722 [30:06:45<15:56:54, 34.09s/it]
 64%|██████▍   | 3039/4722 [30:07:16<15:34:11, 33.30s/it]
 64%|██████▍   | 3040/4722 [30:07:53<16:01:10, 34.29s/it]
                                                         
{'loss': 0.433, 'learning_rate': 0.0001, 'global_step': 3040, 'epoch': 1.93}
+
 64%|██████▍   | 3040/4722 [30:07:53<16:01:10, 34.29s/it]
 64%|██████▍   | 3041/4722 [30:08:24<15:37:40, 33.47s/it]
 64%|██████▍   | 3042/4722 [30:08:55<15:13:58, 32.64s/it]
 64%|██████▍   | 3043/4722 [30:09:32<15:47:22, 33.86s/it]
 64%|██████▍   | 3044/4722 [30:10:04<15:32:01, 33.33s/it]
 64%|██████▍   | 3045/4722 [30:10:36<15:24:55, 33.09s/it]
 65%|██████▍   | 3046/4722 [30:11:12<15:41:34, 33.71s/it]
 65%|██████▍   | 3047/4722 [30:11:44<15:27:38, 33.23s/it]
 65%|██████▍   | 3048/4722 [30:12:21<16:01:33, 34.46s/it]
 65%|██████▍   | 3049/4722 [30:13:01<16:45:18, 36.05s/it]
 65%|██████▍   | 3050/4722 [30:13:29<15:38:39, 33.68s/it]
 65%|██████▍   | 3051/4722 [30:13:58<14:57:23, 32.22s/it]
 65%|██████▍   | 3052/4722 [30:14:35<15:39:21, 33.75s/it]
 65%|██████▍   | 3053/4722 [30:15:12<16:08:36, 34.82s/it]
 65%|██████▍   | 3054/4722 [30:15:40<15:10:04, 32.74s/it]
 65%|██████▍   | 3055/4722 [30:16:11<14:53:09, 32.15s/it]
 65%|██████▍   | 3056/4722 [30:17:02<17:27:48, 37.74s/it]
 65%|██████▍   | 3057/4722 [30:17:35<16:52:23, 36.48s/it]
 65%|██████▍   | 3058/4722 [30:18:07<16:13:53, 35.12s/it]
 65%|██████▍   | 3059/4722 [30:18:41<16:04:22, 34.79s/it]
 65%|██████▍   | 3060/4722 [30:19:12<15:30:24, 33.59s/it]
                                                         
{'loss': 0.4642, 'learning_rate': 0.0001, 'global_step': 3060, 'epoch': 1.94}
+
 65%|██████▍   | 3060/4722 [30:19:12<15:30:24, 33.59s/it]
 65%|██████▍   | 3061/4722 [30:19:43<15:10:05, 32.87s/it]
 65%|██████▍   | 3062/4722 [30:20:28<16:45:07, 36.33s/it]
 65%|██████▍   | 3063/4722 [30:20:56<15:39:59, 34.00s/it]
 65%|██████▍   | 3064/4722 [30:21:37<16:34:22, 35.98s/it]
 65%|██████▍   | 3065/4722 [30:22:10<16:10:59, 35.16s/it]
 65%|██████▍   | 3066/4722 [30:22:41<15:33:44, 33.83s/it]
 65%|██████▍   | 3067/4722 [30:23:09<14:45:10, 32.09s/it]
 65%|██████▍   | 3068/4722 [30:23:38<14:19:25, 31.18s/it]
 65%|██████▍   | 3069/4722 [30:24:20<15:46:51, 34.37s/it]
 65%|██████▌   | 3070/4722 [30:24:55<15:50:07, 34.51s/it]
 65%|██████▌   | 3071/4722 [30:25:26<15:23:48, 33.57s/it]
 65%|██████▌   | 3072/4722 [30:26:03<15:52:21, 34.63s/it]
 65%|██████▌   | 3073/4722 [30:26:37<15:45:41, 34.41s/it]
 65%|██████▌   | 3074/4722 [30:27:17<16:33:34, 36.17s/it]
 65%|██████▌   | 3075/4722 [30:27:47<15:38:33, 34.19s/it]
 65%|██████▌   | 3076/4722 [30:28:22<15:46:45, 34.51s/it]
 65%|██████▌   | 3077/4722 [30:28:53<15:13:13, 33.31s/it]
 65%|██████▌   | 3078/4722 [30:29:24<14:54:13, 32.64s/it]
 65%|██████▌   | 3079/4722 [30:29:57<15:00:02, 32.87s/it]
 65%|██████▌   | 3080/4722 [30:30:31<15:06:17, 33.12s/it]
                                                         
{'loss': 0.4504, 'learning_rate': 0.0001, 'global_step': 3080, 'epoch': 1.96}
+
 65%|██████▌   | 3080/4722 [30:30:31<15:06:17, 33.12s/it]
 65%|██████▌   | 3081/4722 [30:30:59<14:26:06, 31.67s/it]
 65%|██████▌   | 3082/4722 [30:31:35<14:59:45, 32.92s/it]
 65%|██████▌   | 3083/4722 [30:32:16<16:08:49, 35.47s/it]
 65%|██████▌   | 3084/4722 [30:32:49<15:44:55, 34.61s/it]
 65%|██████▌   | 3085/4722 [30:33:21<15:26:39, 33.96s/it]
 65%|██████▌   | 3086/4722 [30:34:01<16:12:29, 35.67s/it]
 65%|██████▌   | 3087/4722 [30:34:35<15:55:12, 35.05s/it]
 65%|██████▌   | 3088/4722 [30:35:06<15:26:56, 34.04s/it]
 65%|██████▌   | 3089/4722 [30:35:46<16:12:16, 35.72s/it]
 65%|██████▌   | 3090/4722 [30:36:16<15:29:17, 34.17s/it]
 65%|██████▌   | 3091/4722 [30:36:48<15:05:34, 33.31s/it]
 65%|██████▌   | 3092/4722 [30:37:21<15:03:26, 33.26s/it]
 66%|██████▌   | 3093/4722 [30:38:02<16:06:46, 35.61s/it]
 66%|██████▌   | 3094/4722 [30:38:32<15:18:47, 33.86s/it]
 66%|██████▌   | 3095/4722 [30:39:06<15:18:49, 33.88s/it]
 66%|██████▌   | 3096/4722 [30:39:36<14:50:54, 32.87s/it]
 66%|██████▌   | 3097/4722 [30:40:09<14:52:08, 32.94s/it]
 66%|██████▌   | 3098/4722 [30:40:43<15:01:37, 33.31s/it]
 66%|██████▌   | 3099/4722 [30:41:13<14:27:03, 32.05s/it]
 66%|██████▌   | 3100/4722 [30:41:46<14:38:18, 32.49s/it]
                                                         
{'loss': 0.4278, 'learning_rate': 0.0001, 'global_step': 3100, 'epoch': 1.97}
+
 66%|██████▌   | 3100/4722 [30:41:46<14:38:18, 32.49s/it]
 66%|██████▌   | 3101/4722 [30:42:16<14:13:51, 31.60s/it]
 66%|██████▌   | 3102/4722 [30:42:50<14:33:06, 32.34s/it]
 66%|██████▌   | 3103/4722 [30:43:26<15:07:37, 33.64s/it]
 66%|██████▌   | 3104/4722 [30:44:04<15:36:42, 34.74s/it]
 66%|██████▌   | 3105/4722 [30:44:33<14:51:25, 33.08s/it]
 66%|██████▌   | 3106/4722 [30:45:06<14:48:59, 33.01s/it]
 66%|██████▌   | 3107/4722 [30:45:38<14:41:12, 32.74s/it]
 66%|██████▌   | 3108/4722 [30:46:11<14:46:07, 32.94s/it]
 66%|██████▌   | 3109/4722 [30:46:44<14:46:09, 32.96s/it]
 66%|██████▌   | 3110/4722 [30:47:17<14:45:47, 32.97s/it]
 66%|██████▌   | 3111/4722 [30:47:59<15:54:24, 35.55s/it]
 66%|██████▌   | 3112/4722 [30:48:33<15:40:29, 35.05s/it]
 66%|██████▌   | 3113/4722 [30:49:07<15:35:30, 34.89s/it]
 66%|██████▌   | 3114/4722 [30:49:45<15:58:48, 35.78s/it]
 66%|██████▌   | 3115/4722 [30:50:18<15:35:14, 34.92s/it]
 66%|██████▌   | 3116/4722 [30:50:50<15:15:24, 34.20s/it]
 66%|██████▌   | 3117/4722 [30:51:20<14:36:55, 32.78s/it]
 66%|██████▌   | 3118/4722 [30:51:55<14:51:28, 33.35s/it]
 66%|██████▌   | 3119/4722 [30:52:32<15:24:42, 34.61s/it]
 66%|██████▌   | 3120/4722 [30:53:05<15:08:03, 34.01s/it]
                                                         
{'loss': 0.4197, 'learning_rate': 0.0001, 'global_step': 3120, 'epoch': 1.98}
+
 66%|██████▌   | 3120/4722 [30:53:05<15:08:03, 34.01s/it]
 66%|██████▌   | 3121/4722 [30:53:33<14:24:45, 32.41s/it]
 66%|██████▌   | 3122/4722 [30:54:07<14:30:10, 32.63s/it]
 66%|██████▌   | 3123/4722 [30:54:41<14:43:48, 33.16s/it]
 66%|██████▌   | 3124/4722 [30:55:15<14:52:29, 33.51s/it]
 66%|██████▌   | 3125/4722 [30:55:50<15:00:22, 33.83s/it]
 66%|██████▌   | 3126/4722 [30:56:17<14:07:58, 31.88s/it]
 66%|██████▌   | 3127/4722 [30:56:51<14:24:06, 32.51s/it]
 66%|██████▌   | 3128/4722 [30:57:23<14:19:02, 32.34s/it]
 66%|██████▋   | 3129/4722 [30:57:59<14:50:05, 33.52s/it]
 66%|██████▋   | 3130/4722 [30:58:30<14:29:01, 32.75s/it]
 66%|██████▋   | 3131/4722 [30:59:03<14:27:07, 32.70s/it]
 66%|██████▋   | 3132/4722 [30:59:43<15:21:28, 34.77s/it]
 66%|██████▋   | 3133/4722 [31:00:15<14:59:37, 33.97s/it]
 66%|██████▋   | 3134/4722 [31:00:49<15:02:20, 34.09s/it]
 66%|██████▋   | 3135/4722 [31:01:18<14:22:33, 32.61s/it]
 66%|██████▋   | 3136/4722 [31:01:52<14:30:53, 32.95s/it]
 66%|██████▋   | 3137/4722 [31:02:26<14:36:31, 33.18s/it]
 66%|██████▋   | 3138/4722 [31:03:00<14:47:00, 33.60s/it]
 66%|██████▋   | 3139/4722 [31:03:31<14:25:19, 32.80s/it]
 66%|██████▋   | 3140/4722 [31:04:03<14:20:50, 32.65s/it]
                                                         
{'loss': 0.4454, 'learning_rate': 0.0001, 'global_step': 3140, 'epoch': 1.99}
+
 66%|██████▋   | 3140/4722 [31:04:03<14:20:50, 32.65s/it]
 67%|██████▋   | 3141/4722 [31:04:39<14:39:34, 33.38s/it]
 67%|██████▋   | 3142/4722 [31:05:13<14:46:26, 33.66s/it]
 67%|██████▋   | 3143/4722 [31:05:43<14:15:48, 32.52s/it]
 67%|██████▋   | 3144/4722 [31:06:14<14:05:34, 32.15s/it]
 67%|██████▋   | 3145/4722 [31:06:45<13:57:19, 31.86s/it]
 67%|██████▋   | 3146/4722 [31:07:17<13:57:25, 31.88s/it]
 67%|██████▋   | 3147/4722 [31:07:50<14:08:09, 32.31s/it]
 67%|██████▋   | 3148/4722 [31:08:23<14:10:54, 32.44s/it]
 67%|██████▋   | 3149/4722 [31:09:03<15:06:53, 34.59s/it]
 67%|██████▋   | 3150/4722 [31:09:32<14:21:02, 32.86s/it]
 67%|██████▋   | 3151/4722 [31:10:10<15:02:21, 34.46s/it]
 67%|██████▋   | 3152/4722 [31:10:46<15:12:40, 34.88s/it]
 67%|██████▋   | 3153/4722 [31:11:21<15:16:47, 35.06s/it]
 67%|██████▋   | 3154/4722 [31:11:56<15:13:12, 34.94s/it]
 67%|██████▋   | 3155/4722 [31:12:31<15:16:33, 35.09s/it]
 67%|██████▋   | 3156/4722 [31:13:01<14:35:44, 33.55s/it]
 67%|██████▋   | 3157/4722 [31:13:34<14:31:37, 33.42s/it]
 67%|██████▋   | 3158/4722 [31:14:14<15:20:48, 35.33s/it]
 67%|██████▋   | 3159/4722 [31:14:42<14:20:09, 33.02s/it]
 67%|██████▋   | 3160/4722 [31:15:18<14:48:02, 34.11s/it]
                                                         
{'loss': 0.3011, 'learning_rate': 0.0001, 'global_step': 3160, 'epoch': 2.01}
+
 67%|██████▋   | 3160/4722 [31:15:18<14:48:02, 34.11s/it]
 67%|██████▋   | 3161/4722 [31:15:51<14:33:44, 33.58s/it]
 67%|██████▋   | 3162/4722 [31:16:26<14:47:40, 34.14s/it]
 67%|██████▋   | 3163/4722 [31:16:58<14:28:48, 33.44s/it]
 67%|██████▋   | 3164/4722 [31:17:32<14:34:41, 33.69s/it]
 67%|██████▋   | 3165/4722 [31:18:06<14:36:44, 33.79s/it]
 67%|██████▋   | 3166/4722 [31:18:38<14:16:29, 33.03s/it]
 67%|██████▋   | 3167/4722 [31:19:21<15:37:25, 36.17s/it]
 67%|██████▋   | 3168/4722 [31:19:58<15:43:49, 36.44s/it]
 67%|██████▋   | 3169/4722 [31:20:37<16:00:00, 37.09s/it]
 67%|██████▋   | 3170/4722 [31:21:02<14:29:26, 33.61s/it]
 67%|██████▋   | 3171/4722 [31:21:35<14:20:46, 33.30s/it]
 67%|██████▋   | 3172/4722 [31:22:10<14:34:57, 33.87s/it]
 67%|██████▋   | 3173/4722 [31:22:43<14:25:20, 33.52s/it]
 67%|██████▋   | 3174/4722 [31:23:16<14:23:53, 33.48s/it]
 67%|██████▋   | 3175/4722 [31:23:47<14:04:14, 32.74s/it]
 67%|██████▋   | 3176/4722 [31:24:22<14:18:10, 33.31s/it]
 67%|██████▋   | 3177/4722 [31:24:56<14:23:34, 33.54s/it]
 67%|██████▋   | 3178/4722 [31:25:30<14:29:10, 33.78s/it]
 67%|██████▋   | 3179/4722 [31:26:00<13:58:37, 32.61s/it]
 67%|██████▋   | 3180/4722 [31:26:31<13:46:41, 32.17s/it]
                                                         
{'loss': 0.2277, 'learning_rate': 0.0001, 'global_step': 3180, 'epoch': 2.02}
+
 67%|██████▋   | 3180/4722 [31:26:31<13:46:41, 32.17s/it]
 67%|██████▋   | 3181/4722 [31:27:11<14:45:34, 34.48s/it]
 67%|██████▋   | 3182/4722 [31:27:41<14:12:23, 33.21s/it]
 67%|██████▋   | 3183/4722 [31:28:17<14:31:01, 33.96s/it]
 67%|██████▋   | 3184/4722 [31:28:50<14:20:31, 33.57s/it]
 67%|██████▋   | 3185/4722 [31:29:19<13:46:54, 32.28s/it]
 67%|██████▋   | 3186/4722 [31:29:50<13:40:58, 32.07s/it]
 67%|███��██▋   | 3187/4722 [31:30:24<13:51:42, 32.51s/it]
 68%|██████▊   | 3188/4722 [31:30:57<13:54:03, 32.62s/it]
 68%|██████▊   | 3189/4722 [31:31:33<14:16:40, 33.53s/it]
 68%|██████▊   | 3190/4722 [31:32:03<13:56:08, 32.75s/it]
 68%|██████▊   | 3191/4722 [31:32:37<13:58:51, 32.88s/it]
 68%|██████▊   | 3192/4722 [31:33:15<14:39:47, 34.50s/it]
 68%|██████▊   | 3193/4722 [31:33:50<14:42:35, 34.63s/it]
 68%|██████▊   | 3194/4722 [31:34:26<14:52:32, 35.05s/it]
 68%|██████▊   | 3195/4722 [31:35:00<14:47:33, 34.87s/it]
 68%|██████▊   | 3196/4722 [31:35:34<14:34:03, 34.37s/it]
 68%|██████▊   | 3197/4722 [31:36:05<14:10:52, 33.48s/it]
 68%|██████▊   | 3198/4722 [31:36:40<14:25:38, 34.08s/it]
 68%|██████▊   | 3199/4722 [31:37:19<14:56:26, 35.32s/it]
 68%|██████▊   | 3200/4722 [31:37:51<14:34:13, 34.46s/it]
                                                         
{'loss': 0.2198, 'learning_rate': 0.0001, 'global_step': 3200, 'epoch': 2.03}
+
 68%|██████▊   | 3200/4722 [31:37:51<14:34:13, 34.46s/it]
 68%|██████▊   | 3201/4722 [31:38:26<14:38:40, 34.66s/it]
 68%|██████▊   | 3202/4722 [31:39:00<14:31:50, 34.41s/it]
 68%|██████▊   | 3203/4722 [31:39:39<15:02:22, 35.64s/it]
 68%|██████▊   | 3204/4722 [31:40:09<14:24:13, 34.16s/it]
 68%|██████▊   | 3205/4722 [31:40:46<14:46:24, 35.06s/it]
 68%|██████▊   | 3206/4722 [31:41:18<14:21:47, 34.11s/it]
 68%|██████▊   | 3207/4722 [31:41:48<13:50:09, 32.88s/it]
 68%|██████▊   | 3208/4722 [31:42:21<13:46:45, 32.76s/it]
 68%|██████▊   | 3209/4722 [31:42:54<13:50:22, 32.93s/it]
 68%|██████▊   | 3210/4722 [31:43:34<14:39:10, 34.89s/it]
 68%|██████▊   | 3211/4722 [31:44:07<14:26:39, 34.41s/it]
 68%|██████▊   | 3212/4722 [31:44:41<14:20:06, 34.18s/it]
 68%|██████▊   | 3213/4722 [31:45:15<14:23:09, 34.32s/it]
 68%|██████▊   | 3214/4722 [31:45:47<14:02:29, 33.52s/it]
 68%|██████▊   | 3215/4722 [31:46:20<13:58:14, 33.37s/it]
 68%|██████▊   | 3216/4722 [31:46:58<14:36:01, 34.90s/it]
 68%|██████▊   | 3217/4722 [31:47:31<14:15:24, 34.10s/it]
 68%|██████▊   | 3218/4722 [31:48:00<13:41:58, 32.79s/it]
 68%|██████▊   | 3219/4722 [31:48:39<14:23:26, 34.47s/it]
 68%|██████▊   | 3220/4722 [31:49:13<14:20:27, 34.37s/it]
                                                         
{'loss': 0.2302, 'learning_rate': 0.0001, 'global_step': 3220, 'epoch': 2.05}
+
 68%|██████▊   | 3220/4722 [31:49:13<14:20:27, 34.37s/it]
 68%|██████▊   | 3221/4722 [31:49:45<14:04:56, 33.78s/it]
 68%|██████▊   | 3222/4722 [31:50:17<13:52:24, 33.30s/it]
 68%|██████▊   | 3223/4722 [31:50:50<13:49:57, 33.22s/it]
 68%|██████▊   | 3224/4722 [31:51:23<13:47:52, 33.16s/it]
 68%|██████▊   | 3225/4722 [31:51:54<13:24:12, 32.23s/it]
 68%|██████▊   | 3226/4722 [31:52:28<13:38:54, 32.84s/it]
 68%|██████▊   | 3227/4722 [31:53:01<13:38:06, 32.83s/it]
 68%|██████▊   | 3228/4722 [31:53:35<13:47:38, 33.24s/it]
 68%|██████▊   | 3229/4722 [31:54:04<13:15:13, 31.96s/it]
 68%|██████▊   | 3230/4722 [31:54:40<13:49:31, 33.36s/it]
 68%|██████▊   | 3231/4722 [31:55:13<13:41:52, 33.07s/it]
 68%|██████▊   | 3232/4722 [31:55:42<13:12:00, 31.89s/it]
 68%|██████▊   | 3233/4722 [31:56:15<13:17:10, 32.12s/it]
 68%|██████▊   | 3234/4722 [31:56:52<13:58:53, 33.83s/it]
 69%|██████▊   | 3235/4722 [31:57:25<13:46:09, 33.34s/it]
 69%|██████▊   | 3236/4722 [31:58:03<14:21:23, 34.78s/it]
 69%|██████▊   | 3237/4722 [31:58:43<14:57:37, 36.27s/it]
 69%|██████▊   | 3238/4722 [31:59:15<14:28:36, 35.12s/it]
 69%|██████▊   | 3239/4722 [31:59:57<15:18:27, 37.16s/it]
 69%|██████▊   | 3240/4722 [32:00:30<14:45:28, 35.85s/it]
                                                         
{'loss': 0.3219, 'learning_rate': 0.0001, 'global_step': 3240, 'epoch': 2.06}
+
 69%|██████▊   | 3240/4722 [32:00:30<14:45:28, 35.85s/it]
 69%|██████▊   | 3241/4722 [32:00:58<13:50:39, 33.65s/it]
 69%|██████▊   | 3242/4722 [32:01:30<13:39:11, 33.21s/it]
 69%|██████▊   | 3243/4722 [32:02:03<13:37:02, 33.15s/it]
 69%|██████▊   | 3244/4722 [32:02:35<13:28:15, 32.81s/it]
 69%|██████▊   | 3245/4722 [32:03:03<12:48:03, 31.20s/it]
 69%|██████▊   | 3246/4722 [32:03:41<13:36:27, 33.19s/it]
 69%|██████▉   | 3247/4722 [32:04:12<13:22:40, 32.65s/it]
 69%|██████▉   | 3248/4722 [32:04:48<13:47:46, 33.69s/it]
 69%|██████▉   | 3249/4722 [32:05:19<13:23:33, 32.73s/it]
 69%|██████▉   | 3250/4722 [32:05:57<14:01:35, 34.30s/it]
 69%|██████▉   | 3251/4722 [32:06:28<13:40:40, 33.47s/it]
 69%|██████▉   | 3252/4722 [32:06:56<12:56:57, 31.71s/it]
 69%|██████▉   | 3253/4722 [32:07:27<12:49:40, 31.44s/it]
 69%|██████▉   | 3254/4722 [32:08:03<13:22:06, 32.78s/it]
 69%|██████▉   | 3255/4722 [32:08:36<13:30:10, 33.14s/it]
 69%|██████▉   | 3256/4722 [32:09:10<13:32:59, 33.27s/it]
 69%|██████▉   | 3257/4722 [32:09:42<13:22:44, 32.88s/it]
 69%|██████▉   | 3258/4722 [32:10:12<12:59:39, 31.95s/it]
 69%|██████▉   | 3259/4722 [32:10:47<13:20:54, 32.85s/it]
 69%|██████▉   | 3260/4722 [32:11:18<13:08:00, 32.34s/it]
                                                         
{'loss': 0.2044, 'learning_rate': 0.0001, 'global_step': 3260, 'epoch': 2.07}
+
 69%|██████▉   | 3260/4722 [32:11:18<13:08:00, 32.34s/it]
 69%|██████▉   | 3261/4722 [32:11:51<13:10:47, 32.48s/it]
 69%|██████▉   | 3262/4722 [32:12:26<13:31:07, 33.33s/it]
 69%|██████▉   | 3263/4722 [32:12:59<13:30:35, 33.33s/it]
 69%|██████▉   | 3264/4722 [32:13:26<12:41:37, 31.34s/it]
 69%|██████▉   | 3265/4722 [32:13:54<12:14:02, 30.23s/it]
 69%|██████▉   | 3266/4722 [32:14:29<12:47:14, 31.62s/it]
 69%|██████▉   | 3267/4722 [32:15:07<13:36:15, 33.66s/it]
 69%|██████▉   | 3268/4722 [32:15:42<13:46:16, 34.10s/it]
 69%|██████▉   | 3269/4722 [32:16:16<13:45:20, 34.08s/it]
 69%|██████▉   | 3270/4722 [32:16:54<14:15:25, 35.35s/it]
 69%|██████▉   | 3271/4722 [32:17:23<13:29:16, 33.46s/it]
 69%|██████▉   | 3272/4722 [32:17:56<13:18:59, 33.06s/it]
 69%|██████▉   | 3273/4722 [32:18:30<13:31:27, 33.60s/it]
 69%|██████▉   | 3274/4722 [32:19:04<13:32:39, 33.67s/it]
 69%|██████▉   | 3275/4722 [32:19:36<13:15:23, 32.98s/it]
 69%|██████▉   | 3276/4722 [32:20:08<13:12:11, 32.87s/it]
 69%|██████▉   | 3277/4722 [32:20:39<12:55:54, 32.22s/it]
 69%|██████▉   | 3278/4722 [32:21:10<12:49:16, 31.96s/it]
 69%|██████▉   | 3279/4722 [32:21:44<13:00:14, 32.44s/it]
 69%|██████▉   | 3280/4722 [32:22:24<13:56:50, 34.82s/it]
                                                         
{'loss': 0.265, 'learning_rate': 0.0001, 'global_step': 3280, 'epoch': 2.08}
+
 69%|██████▉   | 3280/4722 [32:22:24<13:56:50, 34.82s/it]
 69%|██████▉   | 3281/4722 [32:22:56<13:34:35, 33.92s/it]
 70%|██████▉   | 3282/4722 [32:23:28<13:17:05, 33.21s/it]
 70%|██████▉   | 3283/4722 [32:24:01<13:13:52, 33.10s/it]
 70%|██████▉   | 3284/4722 [32:24:32<12:58:33, 32.49s/it]
 70%|██████▉   | 3285/4722 [32:25:05<13:02:49, 32.69s/it]
 70%|██████▉   | 3286/4722 [32:25:38<13:05:36, 32.82s/it]
 70%|██████▉   | 3287/4722 [32:26:23<14:35:16, 36.60s/it]
 70%|██████▉   | 3288/4722 [32:26:56<14:06:23, 35.41s/it]
 70%|██████▉   | 3289/4722 [32:27:28<13:45:21, 34.56s/it]
 70%|██████▉   | 3290/4722 [32:28:03<13:47:40, 34.68s/it]
 70%|██████▉   | 3291/4722 [32:28:29<12:44:35, 32.06s/it]
 70%|██████▉   | 3292/4722 [32:29:01<12:39:38, 31.87s/it]
 70%|██████▉   | 3293/4722 [32:29:36<12:59:30, 32.73s/it]
 70%|██████▉   | 3294/4722 [32:30:05<12:32:22, 31.61s/it]
 70%|██████▉   | 3295/4722 [32:30:38<12:43:14, 32.09s/it]
 70%|██████▉   | 3296/4722 [32:31:12<13:00:44, 32.85s/it]
 70%|██████▉   | 3297/4722 [32:31:44<12:51:39, 32.49s/it]
 70%|██████▉   | 3298/4722 [32:32:19<13:09:08, 33.25s/it]
 70%|██████▉   | 3299/4722 [32:32:56<13:36:37, 34.43s/it]
 70%|██████▉   | 3300/4722 [32:33:27<13:08:41, 33.28s/it]
                                                         
{'loss': 0.2372, 'learning_rate': 0.0001, 'global_step': 3300, 'epoch': 2.1}
+
 70%|██████▉   | 3300/4722 [32:33:27<13:08:41, 33.28s/it]
 70%|██████▉   | 3301/4722 [32:34:07<13:58:38, 35.41s/it]
 70%|██████▉   | 3302/4722 [32:34:38<13:24:03, 33.97s/it]
 70%|██████▉   | 3303/4722 [32:35:15<13:42:31, 34.78s/it]
 70%|██████▉   | 3304/4722 [32:35:45<13:10:41, 33.46s/it]
 70%|██████▉   | 3305/4722 [32:36:21<13:29:30, 34.28s/it]
 70%|███████   | 3306/4722 [32:36:55<13:23:17, 34.04s/it]
 70%|███████   | 3307/4722 [32:37:37<14:22:19, 36.56s/it]
 70%|███████   | 3308/4722 [32:38:10<13:59:06, 35.61s/it]
 70%|███████   | 3309/4722 [32:38:53<14:49:29, 37.77s/it]
 70%|███████   | 3310/4722 [32:39:25<14:03:27, 35.84s/it]
 70%|███████   | 3311/4722 [32:39:57<13:42:16, 34.97s/it]
 70%|███████   | 3312/4722 [32:40:32<13:40:42, 34.92s/it]
 70%|███████   | 3313/4722 [32:41:05<13:24:59, 34.28s/it]
 70%|███████   | 3314/4722 [32:41:36<13:02:45, 33.36s/it]
 70%|███████   | 3315/4722 [32:42:10<13:05:39, 33.50s/it]
 70%|███████   | 3316/4722 [32:42:44<13:10:19, 33.73s/it]
 70%|███████   | 3317/4722 [32:43:18<13:10:58, 33.78s/it]
 70%|███████   | 3318/4722 [32:43:48<12:41:05, 32.53s/it]
 70%|███████   | 3319/4722 [32:44:19<12:33:43, 32.23s/it]
 70%|███████   | 3320/4722 [32:44:52<12:37:57, 32.44s/it]
                                                         
{'loss': 0.2402, 'learning_rate': 0.0001, 'global_step': 3320, 'epoch': 2.11}
+
 70%|███████   | 3320/4722 [32:44:52<12:37:57, 32.44s/it]
 70%|███████   | 3321/4722 [32:45:31<13:20:11, 34.27s/it]
 70%|███████   | 3322/4722 [32:46:05<13:15:25, 34.09s/it]
 70%|███████   | 3323/4722 [32:46:36<12:58:53, 33.40s/it]
 70%|███████   | 3324/4722 [32:47:09<12:55:48, 33.30s/it]
 70%|███████   | 3325/4722 [32:47:48<13:31:21, 34.85s/it]
 70%|███████   | 3326/4722 [32:48:24<13:37:06, 35.12s/it]
 70%|███████   | 3327/4722 [32:49:05<14:21:33, 37.06s/it]
 70%|███████   | 3328/4722 [32:49:36<13:34:29, 35.06s/it]
 70%|███████   | 3329/4722 [32:50:07<13:05:18, 33.83s/it]
 71%|███████   | 3330/4722 [32:50:44<13:27:34, 34.81s/it]
 71%|███████   | 3331/4722 [32:51:17<13:16:05, 34.34s/it]
 71%|███████   | 3332/4722 [32:51:48<12:52:24, 33.34s/it]
 71%|███████   | 3333/4722 [32:52:17<12:20:02, 31.97s/it]
 71%|███████   | 3334/4722 [32:52:50<12:30:47, 32.45s/it]
 71%|███████   | 3335/4722 [32:53:22<12:26:56, 32.31s/it]
 71%|███████   | 3336/4722 [32:53:54<12:22:23, 32.14s/it]
 71%|███████   | 3337/4722 [32:54:28<12:35:50, 32.74s/it]
 71%|███████   | 3338/4722 [32:54:56<12:04:06, 31.39s/it]
 71%|███████   | 3339/4722 [32:55:27<12:01:44, 31.31s/it]
 71%|███████   | 3340/4722 [32:55:57<11:46:05, 30.66s/it]
                                                         
{'loss': 0.2581, 'learning_rate': 0.0001, 'global_step': 3340, 'epoch': 2.12}
+
 71%|███████   | 3340/4722 [32:55:57<11:46:05, 30.66s/it]
 71%|███████   | 3341/4722 [32:56:31<12:09:54, 31.71s/it]
 71%|███████   | 3342/4722 [32:57:02<12:05:27, 31.54s/it]
 71%|███████   | 3343/4722 [32:57:34<12:11:59, 31.85s/it]
 71%|███████   | 3344/4722 [32:58:10<12:36:49, 32.95s/it]
 71%|███████   | 3345/4722 [32:58:42<12:30:18, 32.69s/it]
 71%|███████   | 3346/4722 [32:59:14<12:26:02, 32.53s/it]
 71%|███████   | 3347/4722 [32:59:50<12:47:57, 33.51s/it]
 71%|███████   | 3348/4722 [33:00:24<12:52:33, 33.74s/it]
 71%|███████   | 3349/4722 [33:00:58<12:53:19, 33.79s/it]
 71%|███████   | 3350/4722 [33:01:26<12:08:03, 31.84s/it]
 71%|███████   | 3351/4722 [33:01:55<11:51:03, 31.12s/it]
 71%|███████   | 3352/4722 [33:02:25<11:41:42, 30.73s/it]
 71%|███████   | 3353/4722 [33:02:54<11:33:56, 30.41s/it]
 71%|███████   | 3354/4722 [33:03:32<12:25:06, 32.68s/it]
 71%|███████   | 3355/4722 [33:04:07<12:38:12, 33.28s/it]
 71%|███████   | 3356/4722 [33:04:38<12:23:47, 32.67s/it]
 71%|███████   | 3357/4722 [33:05:18<13:09:26, 34.70s/it]
 71%|███████   | 3358/4722 [33:05:53<13:09:02, 34.71s/it]
 71%|███████   | 3359/4722 [33:06:27<13:06:32, 34.62s/it]
 71%|███████   | 3360/4722 [33:07:01<13:01:34, 34.43s/it]
                                                         
{'loss': 0.2237, 'learning_rate': 0.0001, 'global_step': 3360, 'epoch': 2.13}
+
 71%|███████   | 3360/4722 [33:07:01<13:01:34, 34.43s/it]
 71%|███████   | 3361/4722 [33:07:32<12:37:09, 33.38s/it]
 71%|███████   | 3362/4722 [33:08:02<12:12:42, 32.33s/it]
 71%|███████   | 3363/4722 [33:08:33<12:05:44, 32.04s/it]
 71%|███████   | 3364/4722 [33:09:06<12:14:00, 32.43s/it]
 71%|███████▏  | 3365/4722 [33:09:39<12:13:39, 32.44s/it]
 71%|███████▏  | 3366/4722 [33:10:09<11:55:38, 31.67s/it]
 71%|███████▏  | 3367/4722 [33:10:51<13:08:18, 34.91s/it]
 71%|███████▏  | 3368/4722 [33:11:28<13:18:28, 35.38s/it]
 71%|███████▏  | 3369/4722 [33:12:04<13:23:32, 35.63s/it]
 71%|███████▏  | 3370/4722 [33:12:45<14:00:59, 37.32s/it]
 71%|███████▏  | 3371/4722 [33:13:18<13:28:48, 35.92s/it]
 71%|███████▏  | 3372/4722 [33:13:53<13:21:19, 35.61s/it]
 71%|███████▏  | 3373/4722 [33:14:32<13:42:17, 36.57s/it]
 71%|���██████▏  | 3374/4722 [33:15:02<12:57:33, 34.61s/it]
 71%|███████▏  | 3375/4722 [33:15:38<13:09:54, 35.19s/it]
 71%|███████▏  | 3376/4722 [33:16:08<12:33:39, 33.60s/it]
 72%|███████▏  | 3377/4722 [33:16:43<12:41:28, 33.97s/it]
 72%|███████▏  | 3378/4722 [33:17:12<12:08:34, 32.53s/it]
 72%|███████▏  | 3379/4722 [33:17:44<12:07:31, 32.50s/it]
 72%|███████▏  | 3380/4722 [33:18:15<11:56:36, 32.04s/it]
                                                         
{'loss': 0.2123, 'learning_rate': 0.0001, 'global_step': 3380, 'epoch': 2.15}
+
 72%|███████▏  | 3380/4722 [33:18:15<11:56:36, 32.04s/it]
 72%|███████▏  | 3381/4722 [33:18:52<12:28:10, 33.48s/it]
 72%|███████▏  | 3382/4722 [33:19:22<12:02:57, 32.37s/it]
 72%|███████▏  | 3383/4722 [33:19:52<11:48:11, 31.73s/it]
 72%|███████▏  | 3384/4722 [33:20:29<12:23:58, 33.36s/it]
 72%|███████▏  | 3385/4722 [33:21:01<12:13:43, 32.93s/it]
 72%|███████▏  | 3386/4722 [33:21:35<12:21:07, 33.28s/it]
 72%|███████▏  | 3387/4722 [33:22:10<12:28:25, 33.64s/it]
 72%|███████▏  | 3388/4722 [33:22:51<13:19:44, 35.97s/it]
 72%|███████▏  | 3389/4722 [33:23:23<12:50:07, 34.66s/it]
 72%|███████▏  | 3390/4722 [33:23:54<12:23:48, 33.50s/it]
 72%|███████▏  | 3391/4722 [33:24:26<12:14:31, 33.11s/it]
 72%|███████▏  | 3392/4722 [33:24:59<12:11:15, 32.99s/it]
 72%|███████▏  | 3393/4722 [33:25:28<11:49:26, 32.03s/it]
 72%|███████▏  | 3394/4722 [33:26:03<12:03:56, 32.71s/it]
 72%|███████▏  | 3395/4722 [33:26:32<11:39:42, 31.64s/it]
 72%|███████▏  | 3396/4722 [33:27:03<11:33:32, 31.38s/it]
 72%|███████▏  | 3397/4722 [33:27:41<12:18:14, 33.43s/it]
 72%|███████▏  | 3398/4722 [33:28:20<12:58:04, 35.26s/it]
 72%|███████▏  | 3399/4722 [33:28:49<12:15:03, 33.34s/it]
 72%|███████▏  | 3400/4722 [33:29:20<11:58:48, 32.62s/it]
                                                         
{'loss': 0.2064, 'learning_rate': 0.0001, 'global_step': 3400, 'epoch': 2.16}
+
 72%|███████▏  | 3400/4722 [33:29:20<11:58:48, 32.62s/it]
 72%|███████▏  | 3401/4722 [33:29:49<11:34:08, 31.53s/it]
 72%|███████▏  | 3402/4722 [33:30:17<11:09:27, 30.43s/it]
 72%|███████▏  | 3403/4722 [33:30:47<11:06:44, 30.33s/it]
 72%|███████▏  | 3404/4722 [33:31:22<11:36:45, 31.72s/it]
 72%|███████▏  | 3405/4722 [33:31:59<12:13:01, 33.40s/it]
 72%|███████▏  | 3406/4722 [33:32:35<12:25:36, 33.99s/it]
 72%|███████▏  | 3407/4722 [33:33:12<12:45:48, 34.94s/it]
 72%|███████▏  | 3408/4722 [33:33:44<12:26:38, 34.09s/it]
 72%|███████▏  | 3409/4722 [33:34:19<12:33:56, 34.45s/it]
 72%|███████▏  | 3410/4722 [33:34:57<12:55:41, 35.47s/it]
 72%|███████▏  | 3411/4722 [33:35:33<12:55:07, 35.47s/it]
 72%|███████▏  | 3412/4722 [33:36:05<12:31:22, 34.41s/it]
 72%|███████▏  | 3413/4722 [33:36:35<12:06:38, 33.31s/it]
 72%|███████▏  | 3414/4722 [33:37:12<12:28:07, 34.32s/it]
 72%|███████▏  | 3415/4722 [33:37:47<12:32:26, 34.54s/it]
 72%|███████▏  | 3416/4722 [33:38:22<12:35:31, 34.71s/it]
 72%|███████▏  | 3417/4722 [33:39:04<13:18:53, 36.73s/it]
 72%|███████▏  | 3418/4722 [33:39:35<12:43:13, 35.12s/it]
 72%|███████▏  | 3419/4722 [33:40:04<12:02:23, 33.26s/it]
 72%|███████▏  | 3420/4722 [33:40:35<11:49:44, 32.71s/it]
                                                         
{'loss': 0.234, 'learning_rate': 0.0001, 'global_step': 3420, 'epoch': 2.17}
+
 72%|███████▏  | 3420/4722 [33:40:35<11:49:44, 32.71s/it]
 72%|███████▏  | 3421/4722 [33:41:06<11:39:03, 32.24s/it]
 72%|███████▏  | 3422/4722 [33:41:42<11:56:40, 33.08s/it]
 72%|███████▏  | 3423/4722 [33:42:11<11:33:58, 32.05s/it]
 73%|███████▎  | 3424/4722 [33:42:45<11:43:12, 32.51s/it]
 73%|███████▎  | 3425/4722 [33:43:15<11:29:25, 31.89s/it]
 73%|███████▎  | 3426/4722 [33:43:49<11:41:56, 32.50s/it]
 73%|███████▎  | 3427/4722 [33:44:25<12:02:01, 33.45s/it]
 73%|███████▎  | 3428/4722 [33:44:57<11:50:26, 32.94s/it]
 73%|███████▎  | 3429/4722 [33:45:31<11:58:40, 33.35s/it]
 73%|███████▎  | 3430/4722 [33:46:05<12:03:15, 33.59s/it]
 73%|███████▎  | 3431/4722 [33:46:39<12:03:22, 33.62s/it]
 73%|███████▎  | 3432/4722 [33:47:17<12:32:49, 35.02s/it]
 73%|███████▎  | 3433/4722 [33:47:50<12:21:17, 34.51s/it]
 73%|███████▎  | 3434/4722 [33:48:21<11:58:12, 33.46s/it]
 73%|███████▎  | 3435/4722 [33:48:58<12:16:18, 34.33s/it]
 73%|███████▎  | 3436/4722 [33:49:30<12:02:21, 33.70s/it]
 73%|███████▎  | 3437/4722 [33:50:06<12:18:52, 34.50s/it]
 73%|███████▎  | 3438/4722 [33:50:37<11:51:29, 33.25s/it]
 73%|███████▎  | 3439/4722 [33:51:15<12:26:37, 34.92s/it]
 73%|███████▎  | 3440/4722 [33:51:45<11:50:45, 33.26s/it]
                                                         
{'loss': 0.2014, 'learning_rate': 0.0001, 'global_step': 3440, 'epoch': 2.19}
+
 73%|███████▎  | 3440/4722 [33:51:45<11:50:45, 33.26s/it]
 73%|███████▎  | 3441/4722 [33:52:22<12:12:31, 34.31s/it]
 73%|███████▎  | 3442/4722 [33:52:55<12:06:14, 34.04s/it]
 73%|███████▎  | 3443/4722 [33:53:29<12:04:26, 33.98s/it]
 73%|███████▎  | 3444/4722 [33:53:59<11:42:06, 32.96s/it]
 73%|███████▎  | 3445/4722 [33:54:30<11:25:34, 32.21s/it]
 73%|███████▎  | 3446/4722 [33:55:07<11:58:06, 33.77s/it]
 73%|███████▎  | 3447/4722 [33:55:36<11:26:47, 32.32s/it]
 73%|███████▎  | 3448/4722 [33:56:10<11:34:31, 32.71s/it]
 73%|███████▎  | 3449/4722 [33:56:43<11:37:15, 32.86s/it]
 73%|███████▎  | 3450/4722 [33:57:17<11:43:59, 33.21s/it]
 73%|███████▎  | 3451/4722 [33:57:54<12:08:06, 34.37s/it]
 73%|███████▎  | 3452/4722 [33:58:29<12:13:42, 34.66s/it]
 73%|███████▎  | 3453/4722 [33:58:58<11:33:30, 32.79s/it]
 73%|███████▎  | 3454/4722 [33:59:34<11:54:22, 33.80s/it]
 73%|███████▎  | 3455/4722 [34:00:06<11:43:53, 33.33s/it]
 73%|███████▎  | 3456/4722 [34:00:36<11:20:34, 32.25s/it]
 73%|███████▎  | 3457/4722 [34:01:10<11:29:46, 32.72s/it]
 73%|███████▎  | 3458/4722 [34:01:45<11:46:35, 33.54s/it]
 73%|███████▎  | 3459/4722 [34:02:18<11:40:15, 33.27s/it]
 73%|███████▎  | 3460/4722 [34:02:47<11:14:44, 32.08s/it]
                                                         
{'loss': 0.2686, 'learning_rate': 0.0001, 'global_step': 3460, 'epoch': 2.2}
+
 73%|███████▎  | 3460/4722 [34:02:47<11:14:44, 32.08s/it]
 73%|███████▎  | 3461/4722 [34:03:19<11:11:11, 31.94s/it]
 73%|███████▎  | 3462/4722 [34:03:49<11:00:38, 31.46s/it]
 73%|███████▎  | 3463/4722 [34:04:20<10:55:40, 31.25s/it]
 73%|███████▎  | 3464/4722 [34:04:50<10:47:59, 30.91s/it]
 73%|███████▎  | 3465/4722 [34:05:23<11:02:01, 31.60s/it]
 73%|███████▎  | 3466/4722 [34:06:03<11:51:06, 33.97s/it]
 73%|███████▎  | 3467/4722 [34:06:32<11:23:17, 32.67s/it]
 73%|███████▎  | 3468/4722 [34:07:11<12:00:48, 34.49s/it]
 73%|███████▎  | 3469/4722 [34:07:43<11:41:39, 33.60s/it]
 73%|███████▎  | 3470/4722 [34:08:19<11:58:38, 34.44s/it]
 74%|███████▎  | 3471/4722 [34:08:50<11:34:23, 33.30s/it]
 74%|███████▎  | 3472/4722 [34:09:23<11:32:34, 33.24s/it]
 74%|███████▎  | 3473/4722 [34:09:54<11:20:24, 32.69s/it]
 74%|███████▎  | 3474/4722 [34:10:24<11:00:21, 31.75s/it]
 74%|███████▎  | 3475/4722 [34:11:00<11:29:21, 33.17s/it]
 74%|███████▎  | 3476/4722 [34:11:40<12:11:39, 35.23s/it]
 74%|███████▎  | 3477/4722 [34:12:15<12:10:14, 35.19s/it]
 74%|███████▎  | 3478/4722 [34:12:49<11:59:08, 34.69s/it]
 74%|███████▎  | 3479/4722 [34:13:18<11:24:35, 33.05s/it]
 74%|███████▎  | 3480/4722 [34:13:49<11:11:28, 32.44s/it]
                                                         
{'loss': 0.1797, 'learning_rate': 0.0001, 'global_step': 3480, 'epoch': 2.21}
+
 74%|███████▎  | 3480/4722 [34:13:49<11:11:28, 32.44s/it]
 74%|███████▎  | 3481/4722 [34:14:20<10:58:11, 31.82s/it]
 74%|███████▎  | 3482/4722 [34:14:49<10:41:17, 31.03s/it]
 74%|███████▍  | 3483/4722 [34:15:22<10:52:52, 31.62s/it]
 74%|███████▍  | 3484/4722 [34:15:52<10:43:34, 31.19s/it]
 74%|███████▍  | 3485/4722 [34:16:23<10:43:00, 31.19s/it]
 74%|███████▍  | 3486/4722 [34:17:02<11:30:41, 33.53s/it]
 74%|███████▍  | 3487/4722 [34:17:36<11:30:09, 33.53s/it]
 74%|███████▍  | 3488/4722 [34:18:05<11:02:18, 32.20s/it]
 74%|███████▍  | 3489/4722 [34:18:40<11:22:11, 33.20s/it]
 74%|███████▍  | 3490/4722 [34:19:10<11:02:19, 32.26s/it]
 74%|███████▍  | 3491/4722 [34:19:49<11:41:40, 34.20s/it]
 74%|███████▍  | 3492/4722 [34:20:22<11:36:34, 33.98s/it]
 74%|███████▍  | 3493/4722 [34:20:51<11:01:51, 32.31s/it]
 74%|███████▍  | 3494/4722 [34:21:22<10:50:52, 31.80s/it]
 74%|█████��█▍  | 3495/4722 [34:21:53<10:48:20, 31.70s/it]
 74%|███████▍  | 3496/4722 [34:22:31<11:28:48, 33.71s/it]
 74%|███████▍  | 3497/4722 [34:23:04<11:21:46, 33.39s/it]
 74%|███████▍  | 3498/4722 [34:23:39<11:28:40, 33.76s/it]
 74%|███████▍  | 3499/4722 [34:24:21<12:20:48, 36.34s/it]
 74%|███████▍  | 3500/4722 [34:24:56<12:13:48, 36.03s/it]
                                                         
{'loss': 0.221, 'learning_rate': 0.0001, 'global_step': 3500, 'epoch': 2.22}
+
 74%|███████▍  | 3500/4722 [34:24:56<12:13:48, 36.03s/it]
 74%|███████▍  | 3501/4722 [34:25:40<12:58:23, 38.25s/it]
 74%|███████▍  | 3502/4722 [34:26:16<12:45:04, 37.63s/it]
 74%|███████▍  | 3503/4722 [34:26:55<12:54:51, 38.14s/it]
 74%|███████▍  | 3504/4722 [34:27:28<12:22:51, 36.59s/it]
 74%|███████▍  | 3505/4722 [34:28:00<11:52:47, 35.14s/it]
 74%|███████▍  | 3506/4722 [34:28:38<12:11:57, 36.12s/it]
 74%|███████▍  | 3507/4722 [34:29:12<11:56:58, 35.41s/it]
 74%|███████▍  | 3508/4722 [34:29:49<12:04:44, 35.82s/it]
 74%|███████▍  | 3509/4722 [34:30:18<11:22:30, 33.76s/it]
 74%|███████▍  | 3510/4722 [34:30:48<10:59:09, 32.63s/it]
 74%|███████▍  | 3511/4722 [34:31:25<11:26:01, 33.99s/it]
 74%|███████▍  | 3512/4722 [34:31:58<11:16:44, 33.56s/it]
 74%|███████▍  | 3513/4722 [34:32:27<10:53:58, 32.45s/it]
 74%|███████▍  | 3514/4722 [34:33:06<11:28:02, 34.17s/it]
 74%|███████▍  | 3515/4722 [34:33:47<12:12:44, 36.42s/it]
 74%|███████▍  | 3516/4722 [34:34:20<11:49:02, 35.28s/it]
 74%|███████▍  | 3517/4722 [34:34:51<11:26:11, 34.17s/it]
 75%|███████▍  | 3518/4722 [34:35:28<11:37:41, 34.77s/it]
 75%|███████▍  | 3519/4722 [34:35:59<11:18:07, 33.82s/it]
 75%|███████▍  | 3520/4722 [34:36:32<11:09:36, 33.42s/it]
                                                         
{'loss': 0.2378, 'learning_rate': 0.0001, 'global_step': 3520, 'epoch': 2.24}
+
 75%|███████▍  | 3520/4722 [34:36:32<11:09:36, 33.42s/it]
 75%|███████▍  | 3521/4722 [34:37:06<11:15:34, 33.75s/it]
 75%|███████▍  | 3522/4722 [34:37:38<11:00:50, 33.04s/it]
 75%|███████▍  | 3523/4722 [34:38:07<10:39:22, 32.00s/it]
 75%|███████▍  | 3524/4722 [34:38:37<10:26:18, 31.37s/it]
 75%|███████▍  | 3525/4722 [34:39:19<11:29:17, 34.55s/it]
 75%|███████▍  | 3526/4722 [34:39:50<11:09:15, 33.57s/it]
 75%|███████▍  | 3527/4722 [34:40:20<10:46:39, 32.47s/it]
 75%|███████▍  | 3528/4722 [34:40:52<10:41:53, 32.26s/it]
 75%|███████▍  | 3529/4722 [34:41:24<10:41:20, 32.26s/it]
 75%|███████▍  | 3530/4722 [34:41:53<10:19:20, 31.18s/it]
 75%|███████▍  | 3531/4722 [34:42:25<10:23:59, 31.44s/it]
 75%|███████▍  | 3532/4722 [34:42:53<10:05:47, 30.54s/it]
 75%|███████▍  | 3533/4722 [34:43:28<10:26:33, 31.62s/it]
 75%|███████▍  | 3534/4722 [34:43:59<10:26:46, 31.66s/it]
 75%|███████▍  | 3535/4722 [34:44:32<10:29:26, 31.82s/it]
 75%|███████▍  | 3536/4722 [34:45:00<10:10:51, 30.90s/it]
 75%|███████▍  | 3537/4722 [34:45:37<10:44:37, 32.64s/it]
 75%|███████▍  | 3538/4722 [34:46:13<11:03:42, 33.63s/it]
 75%|███████▍  | 3539/4722 [34:46:40<10:22:41, 31.58s/it]
 75%|███████▍  | 3540/4722 [34:47:17<10:53:00, 33.15s/it]
                                                         
{'loss': 0.307, 'learning_rate': 0.0001, 'global_step': 3540, 'epoch': 2.25}
+
 75%|███████▍  | 3540/4722 [34:47:17<10:53:00, 33.15s/it]
 75%|███████▍  | 3541/4722 [34:47:48<10:45:02, 32.77s/it]
 75%|███████▌  | 3542/4722 [34:48:25<11:08:27, 33.99s/it]
 75%|███████▌  | 3543/4722 [34:48:55<10:43:41, 32.76s/it]
 75%|███████▌  | 3544/4722 [34:49:26<10:33:33, 32.27s/it]
 75%|███████▌  | 3545/4722 [34:50:03<10:59:03, 33.60s/it]
 75%|███████▌  | 3546/4722 [34:50:32<10:29:18, 32.11s/it]
 75%|███████▌  | 3547/4722 [34:51:08<10:52:14, 33.31s/it]
 75%|███████▌  | 3548/4722 [34:51:37<10:26:24, 32.01s/it]
 75%|███████▌  | 3549/4722 [34:52:07<10:14:43, 31.44s/it]
 75%|███████▌  | 3550/4722 [34:52:38<10:12:58, 31.38s/it]
 75%|███████▌  | 3551/4722 [34:53:10<10:18:05, 31.67s/it]
 75%|███████▌  | 3552/4722 [34:53:43<10:21:40, 31.88s/it]
 75%|███████▌  | 3553/4722 [34:54:12<10:06:08, 31.11s/it]
 75%|███████▌  | 3554/4722 [34:54:45<10:18:54, 31.79s/it]
 75%|███████▌  | 3555/4722 [34:55:18<10:23:29, 32.06s/it]
 75%|███████▌  | 3556/4722 [34:55:52<10:36:19, 32.74s/it]
 75%|███████▌  | 3557/4722 [34:56:23<10:21:34, 32.01s/it]
 75%|███████▌  | 3558/4722 [34:56:58<10:40:17, 33.01s/it]
 75%|███████▌  | 3559/4722 [34:57:27<10:15:54, 31.77s/it]
 75%|███████▌  | 3560/4722 [34:57:59<10:16:52, 31.85s/it]
                                                         
{'loss': 0.1915, 'learning_rate': 0.0001, 'global_step': 3560, 'epoch': 2.26}
+
 75%|███████▌  | 3560/4722 [34:57:59<10:16:52, 31.85s/it]
 75%|███████▌  | 3561/4722 [34:58:31<10:16:45, 31.87s/it]
 75%|███████▌  | 3562/4722 [34:59:12<11:10:26, 34.68s/it]
 75%|███████▌  | 3563/4722 [34:59:50<11:25:22, 35.48s/it]
 75%|███████▌  | 3564/4722 [35:00:25<11:25:50, 35.54s/it]
 75%|███████▌  | 3565/4722 [35:00:59<11:16:35, 35.09s/it]
 76%|███████▌  | 3566/4722 [35:01:30<10:49:38, 33.72s/it]
 76%|███████▌  | 3567/4722 [35:01:59<10:21:24, 32.28s/it]
 76%|███████▌  | 3568/4722 [35:02:28<10:01:30, 31.27s/it]
 76%|███████▌  | 3569/4722 [35:03:09<10:57:27, 34.21s/it]
 76%|███████▌  | 3570/4722 [35:03:47<11:17:34, 35.29s/it]
 76%|███████▌  | 3571/4722 [35:04:22<11:16:41, 35.28s/it]
 76%|███████▌  | 3572/4722 [35:04:53<10:55:34, 34.20s/it]
 76%|███████▌  | 3573/4722 [35:05:25<10:37:53, 33.31s/it]
 76%|███████▌  | 3574/4722 [35:06:05<11:16:59, 35.38s/it]
 76%|███████▌  | 3575/4722 [35:06:37<10:55:22, 34.28s/it]
 76%|███████▌  | 3576/4722 [35:07:05<10:22:58, 32.62s/it]
 76%|███████▌  | 3577/4722 [35:07:37<10:18:36, 32.42s/it]
 76%|███████▌  | 3578/4722 [35:08:17<10:58:40, 34.55s/it]
 76%|███████▌  | 3579/4722 [35:08:47<10:34:33, 33.31s/it]
 76%|███████▌  | 3580/4722 [35:09:23<10:48:25, 34.07s/it]
                                                         
{'loss': 0.2686, 'learning_rate': 0.0001, 'global_step': 3580, 'epoch': 2.27}
+
 76%|███████▌  | 3580/4722 [35:09:23<10:48:25, 34.07s/it]
 76%|███████▌  | 3581/4722 [35:10:00<11:03:36, 34.90s/it]
 76%|███████▌  | 3582/4722 [35:10:33<10:54:53, 34.47s/it]
 76%|███████▌  | 3583/4722 [35:11:10<11:06:45, 35.12s/it]
 76%|███████▌  | 3584/4722 [35:11:45<11:05:43, 35.10s/it]
 76%|███████▌  | 3585/4722 [35:12:16<10:40:42, 33.81s/it]
 76%|███████▌  | 3586/4722 [35:12:45<10:13:26, 32.40s/it]
 76%|███████▌  | 3587/4722 [35:13:21<10:31:48, 33.40s/it]
 76%|███████▌  | 3588/4722 [35:14:01<11:09:46, 35.44s/it]
 76%|███████▌  | 3589/4722 [35:14:37<11:13:13, 35.65s/it]
 76%|███████▌  | 3590/4722 [35:15:08<10:44:02, 34.14s/it]
 76%|███████▌  | 3591/4722 [35:15:39<10:28:14, 33.33s/it]
 76%|███████▌  | 3592/4722 [35:16:12<10:25:21, 33.20s/it]
 76%|███████▌  | 3593/4722 [35:16:39<9:51:14, 31.42s/it] 
 76%|███████▌  | 3594/4722 [35:17:14<10:08:25, 32.36s/it]
 76%|███████▌  | 3595/4722 [35:17:44<9:58:04, 31.84s/it] 
 76%|███████▌  | 3596/4722 [35:18:21<10:22:06, 33.15s/it]
 76%|███████▌  | 3597/4722 [35:18:57<10:37:25, 34.00s/it]
 76%|███████▌  | 3598/4722 [35:19:35<10:59:16, 35.19s/it]
 76%|███████▌  | 3599/4722 [35:20:14<11:21:48, 36.43s/it]
 76%|███████▌  | 3600/4722 [35:20:54<11:39:23, 37.40s/it]
                                                         
{'loss': 0.2719, 'learning_rate': 0.0001, 'global_step': 3600, 'epoch': 2.29}
+
 76%|███████▌  | 3600/4722 [35:20:54<11:39:23, 37.40s/it]
 76%|███████▋  | 3601/4722 [35:21:33<11:48:08, 37.90s/it]
 76%|███████▋  | 3602/4722 [35:22:05<11:15:12, 36.17s/it]
 76%|███████▋  | 3603/4722 [35:22:38<10:55:51, 35.17s/it]
 76%|███████▋  | 3604/4722 [35:23:11<10:43:49, 34.55s/it]
 76%|███████▋  | 3605/4722 [35:23:46<10:49:37, 34.89s/it]
 76%|███████▋  | 3606/4722 [35:24:20<10:42:53, 34.56s/it]
 76%|███████▋  | 3607/4722 [35:24:56<10:48:32, 34.90s/it]
 76%|███████▋  | 3608/4722 [35:25:28<10:29:57, 33.93s/it]
 76%|███████▋  | 3609/4722 [35:26:07<10:57:24, 35.44s/it]
 76%|███████▋  | 3610/4722 [35:26:36<10:24:31, 33.70s/it]
 76%|███████▋  | 3611/4722 [35:27:05<9:54:34, 32.11s/it] 
 76%|███████▋  | 3612/4722 [35:27:37<9:53:39, 32.09s/it]
 77%|███████▋  | 3613/4722 [35:28:09<9:56:49, 32.29s/it]
 77%|███████▋  | 3614/4722 [35:28:43<10:01:49, 32.59s/it]
 77%|███████▋  | 3615/4722 [35:29:22<10:39:33, 34.66s/it]
 77%|███████▋  | 3616/4722 [35:29:56<10:33:47, 34.38s/it]
 77%|███████▋  | 3617/4722 [35:30:30<10:29:32, 34.18s/it]
 77%|███████▋  | 3618/4722 [35:31:02<10:18:07, 33.59s/it]
 77%|███████▋  | 3619/4722 [35:31:36<10:19:57, 33.72s/it]
 77%|███████▋  | 3620/4722 [35:32:12<10:30:54, 34.35s/it]
                                                         
{'loss': 0.2416, 'learning_rate': 0.0001, 'global_step': 3620, 'epoch': 2.3}
+
 77%|███████▋  | 3620/4722 [35:32:12<10:30:54, 34.35s/it]
 77%|███████▋  | 3621/4722 [35:32:51<10:55:12, 35.71s/it]
 77%|███████▋  | 3622/4722 [35:33:25<10:48:19, 35.36s/it]
 77%|███████▋  | 3623/4722 [35:33:58<10:32:39, 34.54s/it]
 77%|███████▋  | 3624/4722 [35:34:33<10:34:14, 34.66s/it]
 77%|███████▋  | 3625/4722 [35:35:06<10:29:06, 34.41s/it]
 77%|███████▋  | 3626/4722 [35:35:30<9:31:21, 31.28s/it] 
 77%|███████▋  | 3627/4722 [35:36:11<10:22:20, 34.10s/it]
 77%|███████▋  | 3628/4722 [35:36:41<9:58:15, 32.81s/it] 
 77%|███████▋  | 3629/4722 [35:37:14<10:00:52, 32.98s/it]
 77%|███████▋  | 3630/4722 [35:37:50<10:15:12, 33.80s/it]
 77%|███████▋  | 3631/4722 [35:38:21<10:00:42, 33.04s/it]
 77%|███████▋  | 3632/4722 [35:38:53<9:51:36, 32.57s/it] 
 77%|███████▋  | 3633/4722 [35:39:27<9:59:34, 33.03s/it]
 77%|███████▋  | 3634/4722 [35:40:08<10:42:20, 35.42s/it]
 77%|███████▋  | 3635/4722 [35:40:43<10:42:46, 35.48s/it]
 77%|███████▋  | 3636/4722 [35:41:14<10:13:20, 33.89s/it]
 77%|███████▋  | 3637/4722 [35:41:46<10:04:42, 33.44s/it]
 77%|███████▋  | 3638/4722 [35:42:21<10:12:21, 33.89s/it]
 77%|███████▋  | 3639/4722 [35:42:56<10:15:24, 34.09s/it]
 77%|███████▋  | 3640/4722 [35:43:26<9:54:15, 32.95s/it] 
                                                        
{'loss': 0.2409, 'learning_rate': 0.0001, 'global_step': 3640, 'epoch': 2.31}
+
 77%|███████▋  | 3640/4722 [35:43:26<9:54:15, 32.95s/it]
 77%|███████▋  | 3641/4722 [35:44:00<9:58:19, 33.21s/it]
 77%|███████▋  | 3642/4722 [35:44:34<10:04:26, 33.58s/it]
 77%|███████▋  | 3643/4722 [35:45:08<10:03:47, 33.58s/it]
 77%|███████▋  | 3644/4722 [35:45:44<10:20:16, 34.52s/it]
 77%|███████▋  | 3645/4722 [35:46:16<10:02:01, 33.54s/it]
 77%|███████▋  | 3646/4722 [35:46:51<10:12:21, 34.15s/it]
 77%|███████▋  | 3647/4722 [35:47:22<9:56:05, 33.27s/it] 
 77%|███████▋  | 3648/4722 [35:48:04<10:40:11, 35.76s/it]
 77%|███████▋  | 3649/4722 [35:48:38<10:30:28, 35.25s/it]
 77%|███████▋  | 3650/4722 [35:49:10<10:13:08, 34.32s/it]
 77%|███████▋  | 3651/4722 [35:49:50<10:43:17, 36.04s/it]
 77%|███████▋  | 3652/4722 [35:50:22<10:21:50, 34.87s/it]
 77%|███████▋  | 3653/4722 [35:50:55<10:06:47, 34.06s/it]
 77%|███████▋  | 3654/4722 [35:51:27<9:58:05, 33.60s/it] 
 77%|███████▋  | 3655/4722 [35:51:59<9:46:48, 33.00s/it]
 77%|███████▋  | 3656/4722 [35:52:33<9:52:35, 33.35s/it]
 77%|███████▋  | 3657/4722 [35:53:10<10:09:55, 34.36s/it]
 77%|███████▋  | 3658/4722 [35:53:46<10:21:47, 35.06s/it]
 77%|███████▋  | 3659/4722 [35:54:24<10:36:11, 35.91s/it]
 78%|███████▊  | 3660/4722 [35:54:58<10:23:04, 35.20s/it]
                                                         
{'loss': 0.2486, 'learning_rate': 0.0001, 'global_step': 3660, 'epoch': 2.32}
+
 78%|███████▊  | 3660/4722 [35:54:58<10:23:04, 35.20s/it]
 78%|███████▊  | 3661/4722 [35:55:37<10:41:47, 36.29s/it]
 78%|███████▊  | 3662/4722 [35:56:10<10:25:14, 35.39s/it]
 78%|███████▊  | 3663/4722 [35:56:45<10:22:41, 35.28s/it]
 78%|███████▊  | 3664/4722 [35:57:18<10:10:54, 34.64s/it]
 78%|███████▊  | 3665/4722 [35:57:50<9:57:30, 33.92s/it] 
 78%|███████▊  | 3666/4722 [35:58:20<9:34:17, 32.63s/it]
 78%|███████▊  | 3667/4722 [35:58:52<9:30:47, 32.46s/it]
 78%|███████▊  | 3668/4722 [35:59:23<9:20:31, 31.91s/it]
 78%|███████▊  | 3669/4722 [35:59:56<9:27:09, 32.32s/it]
 78%|███████▊  | 3670/4722 [36:00:31<9:43:28, 33.28s/it]
 78%|███████▊  | 3671/4722 [36:01:06<9:48:52, 33.62s/it]
 78%|███████▊  | 3672/4722 [36:01:41<9:58:48, 34.22s/it]
 78%|███████▊  | 3673/4722 [36:02:19<10:13:58, 35.12s/it]
 78%|███████▊  | 3674/4722 [36:02:55<10:21:27, 35.58s/it]
 78%|███████▊  | 3675/4722 [36:03:25<9:48:22, 33.72s/it] 
 78%|███████▊  | 3676/4722 [36:03:55<9:32:24, 32.83s/it]
 78%|██████��▊  | 3677/4722 [36:04:28<9:30:33, 32.76s/it]
 78%|███████▊  | 3678/4722 [36:05:01<9:30:57, 32.81s/it]
 78%|███████▊  | 3679/4722 [36:05:29<9:06:29, 31.44s/it]
 78%|███████▊  | 3680/4722 [36:06:03<9:16:42, 32.06s/it]
                                                        
{'loss': 0.2249, 'learning_rate': 0.0001, 'global_step': 3680, 'epoch': 2.34}
+
 78%|███████▊  | 3680/4722 [36:06:03<9:16:42, 32.06s/it]
 78%|███████▊  | 3681/4722 [36:06:35<9:18:35, 32.20s/it]
 78%|███████▊  | 3682/4722 [36:07:06<9:08:36, 31.65s/it]
 78%|███████▊  | 3683/4722 [36:07:38<9:13:56, 31.99s/it]
 78%|███████▊  | 3684/4722 [36:08:06<8:48:41, 30.56s/it]
 78%|███████▊  | 3685/4722 [36:08:39<9:03:23, 31.44s/it]
 78%|███████▊  | 3686/4722 [36:09:15<9:23:37, 32.64s/it]
 78%|███████▊  | 3687/4722 [36:09:59<10:22:43, 36.10s/it]
 78%|███████▊  | 3688/4722 [36:10:27<9:39:51, 33.65s/it] 
 78%|███████▊  | 3689/4722 [36:10:59<9:32:47, 33.27s/it]
 78%|███████▊  | 3690/4722 [36:11:31<9:25:14, 32.86s/it]
 78%|███████▊  | 3691/4722 [36:12:13<10:10:37, 35.54s/it]
 78%|███████▊  | 3692/4722 [36:12:46<9:56:58, 34.78s/it] 
 78%|███████▊  | 3693/4722 [36:13:16<9:31:51, 33.34s/it]
 78%|███████▊  | 3694/4722 [36:13:44<9:03:40, 31.73s/it]
 78%|███████▊  | 3695/4722 [36:14:24<9:45:28, 34.20s/it]
 78%|███████▊  | 3696/4722 [36:15:00<9:54:15, 34.75s/it]
 78%|███████▊  | 3697/4722 [36:15:37<10:07:29, 35.56s/it]
 78%|███████▊  | 3698/4722 [36:16:10<9:52:30, 34.72s/it] 
 78%|███████▊  | 3699/4722 [36:16:40<9:29:14, 33.39s/it]
 78%|███████▊  | 3700/4722 [36:17:22<10:11:00, 35.87s/it]
                                                         
{'loss': 0.2383, 'learning_rate': 0.0001, 'global_step': 3700, 'epoch': 2.35}
+
 78%|███████▊  | 3700/4722 [36:17:22<10:11:00, 35.87s/it]
 78%|███████▊  | 3701/4722 [36:18:00<10:22:34, 36.59s/it]
 78%|███████▊  | 3702/4722 [36:18:38<10:28:35, 36.98s/it]
 78%|███████▊  | 3703/4722 [36:19:04<9:30:40, 33.60s/it] 
 78%|███████▊  | 3704/4722 [36:19:43<9:57:04, 35.19s/it]
 78%|███████▊  | 3705/4722 [36:20:22<10:16:37, 36.38s/it]
 78%|███████▊  | 3706/4722 [36:20:58<10:13:07, 36.21s/it]
 79%|███████▊  | 3707/4722 [36:21:38<10:33:04, 37.42s/it]
 79%|███████▊  | 3708/4722 [36:22:09<10:01:10, 35.57s/it]
 79%|███████▊  | 3709/4722 [36:22:41<9:39:56, 34.35s/it] 
 79%|███████▊  | 3710/4722 [36:23:12<9:22:44, 33.36s/it]
 79%|███████▊  | 3711/4722 [36:23:47<9:31:57, 33.94s/it]
 79%|███████▊  | 3712/4722 [36:24:28<10:08:27, 36.15s/it]
 79%|███████▊  | 3713/4722 [36:24:57<9:30:54, 33.95s/it] 
 79%|███████▊  | 3714/4722 [36:25:29<9:18:34, 33.25s/it]
 79%|███████▊  | 3715/4722 [36:26:06<9:40:31, 34.59s/it]
 79%|███████▊  | 3716/4722 [36:26:37<9:18:41, 33.32s/it]
 79%|███████▊  | 3717/4722 [36:27:17<9:51:21, 35.31s/it]
 79%|███████▊  | 3718/4722 [36:27:48<9:30:24, 34.09s/it]
 79%|███████▉  | 3719/4722 [36:28:23<9:36:53, 34.51s/it]
 79%|███████▉  | 3720/4722 [36:28:52<9:05:51, 32.69s/it]
                                                        
{'loss': 0.2195, 'learning_rate': 0.0001, 'global_step': 3720, 'epoch': 2.36}
+
 79%|███████▉  | 3720/4722 [36:28:52<9:05:51, 32.69s/it]
 79%|███████▉  | 3721/4722 [36:29:24<9:02:30, 32.52s/it]
 79%|███████▉  | 3722/4722 [36:29:55<8:55:31, 32.13s/it]
 79%|███████▉  | 3723/4722 [36:30:27<8:53:43, 32.06s/it]
 79%|███████▉  | 3724/4722 [36:31:01<9:03:49, 32.69s/it]
 79%|███████▉  | 3725/4722 [36:31:31<8:50:48, 31.94s/it]
 79%|███████▉  | 3726/4722 [36:32:05<8:59:04, 32.47s/it]
 79%|███████▉  | 3727/4722 [36:32:35<8:43:26, 31.56s/it]
 79%|███████▉  | 3728/4722 [36:33:10<9:01:17, 32.67s/it]
 79%|███████▉  | 3729/4722 [36:33:40<8:47:57, 31.90s/it]
 79%|███████▉  | 3730/4722 [36:34:11<8:44:09, 31.70s/it]
 79%|███████▉  | 3731/4722 [36:34:47<9:03:09, 32.89s/it]
 79%|███████▉  | 3732/4722 [36:35:26<9:33:53, 34.78s/it]
 79%|███████▉  | 3733/4722 [36:35:59<9:22:03, 34.10s/it]
 79%|███████▉  | 3734/4722 [36:36:30<9:10:18, 33.42s/it]
 79%|███████▉  | 3735/4722 [36:37:02<9:02:03, 32.95s/it]
 79%|███████▉  | 3736/4722 [36:37:33<8:52:03, 32.38s/it]
 79%|███████▉  | 3737/4722 [36:38:08<9:01:10, 32.97s/it]
 79%|███████▉  | 3738/4722 [36:38:46<9:26:39, 34.55s/it]
 79%|███████▉  | 3739/4722 [36:39:24<9:44:15, 35.66s/it]
 79%|███████▉  | 3740/4722 [36:39:55<9:18:32, 34.13s/it]
                                                        
{'loss': 0.2611, 'learning_rate': 0.0001, 'global_step': 3740, 'epoch': 2.38}
+
 79%|███████▉  | 3740/4722 [36:39:55<9:18:32, 34.13s/it]
 79%|███████▉  | 3741/4722 [36:40:30<9:21:45, 34.36s/it]
 79%|███████▉  | 3742/4722 [36:41:02<9:12:08, 33.80s/it]
 79%|███████▉  | 3743/4722 [36:41:33<8:55:57, 32.85s/it]
 79%|███████▉  | 3744/4722 [36:42:08<9:06:18, 33.52s/it]
 79%|███████▉  | 3745/4722 [36:42:39<8:54:13, 32.81s/it]
 79%|███████▉  | 3746/4722 [36:43:11<8:48:38, 32.50s/it]
 79%|███████▉  | 3747/4722 [36:43:41<8:36:00, 31.75s/it]
 79%|███████▉  | 3748/4722 [36:44:18<9:01:25, 33.35s/it]
 79%|███████▉  | 3749/4722 [36:44:47<8:42:03, 32.19s/it]
 79%|███████▉  | 3750/4722 [36:45:21<8:50:25, 32.74s/it]
 79%|███████▉  | 3751/4722 [36:45:56<8:59:46, 33.35s/it]
 79%|███████▉  | 3752/4722 [36:46:27<8:47:47, 32.65s/it]
 79%|███████▉  | 3753/4722 [36:46:59<8:44:12, 32.46s/it]
 80%|███████▉  | 3754/4722 [36:47:39<9:19:13, 34.66s/it]
 80%|███████▉  | 3755/4722 [36:48:09<8:54:35, 33.17s/it]
 80%|███████▉  | 3756/4722 [36:48:46<9:15:15, 34.49s/it]
 80%|███████▉  | 3757/4722 [36:49:19<9:06:59, 34.01s/it]
 80%|███████▉  | 3758/4722 [36:49:53<9:05:13, 33.93s/it]
 80%|███████▉  | 3759/4722 [36:50:21<8:37:09, 32.22s/it]
 80%|███████▉  | 3760/4722 [36:50:54<8:40:45, 32.48s/it]
                                                        
{'loss': 0.2851, 'learning_rate': 0.0001, 'global_step': 3760, 'epoch': 2.39}
+
 80%|███████▉  | 3760/4722 [36:50:54<8:40:45, 32.48s/it]
 80%|███████▉  | 3761/4722 [36:51:28<8:46:00, 32.84s/it]
 80%|███████▉  | 3762/4722 [36:52:02<8:50:34, 33.16s/it]
 80%|███████▉  | 3763/4722 [36:52:30<8:26:50, 31.71s/it]
 80%|███████▉  | 3764/4722 [36:53:07<8:50:53, 33.25s/it]
 80%|███████▉  | 3765/4722 [36:53:37<8:37:29, 32.44s/it]
 80%|███████▉  | 3766/4722 [36:54:14<8:55:39, 33.62s/it]
 80%|███████▉  | 3767/4722 [36:54:45<8:41:24, 32.76s/it]
 80%|███████▉  | 3768/4722 [36:55:13<8:18:03, 31.32s/it]
 80%|███████▉  | 3769/4722 [36:55:46<8:29:32, 32.08s/it]
 80%|███████▉  | 3770/4722 [36:56:20<8:35:35, 32.50s/it]
 80%|███████▉  | 3771/4722 [36:56:53<8:36:23, 32.58s/it]
 80%|███████▉  | 3772/4722 [36:57:24<8:32:13, 32.35s/it]
 80%|███████▉  | 3773/4722 [36:57:57<8:32:29, 32.40s/it]
 80%|███████▉  | 3774/4722 [36:58:30<8:33:53, 32.53s/it]
 80%|███████▉  | 3775/4722 [36:59:06<8:51:17, 33.66s/it]
 80%|███████▉  | 3776/4722 [36:59:39<8:49:20, 33.57s/it]
 80%|███████▉  | 3777/4722 [37:00:07<8:20:22, 31.77s/it]
 80%|████████  | 3778/4722 [37:00:36<8:04:19, 30.78s/it]
 80%|████████  | 3779/4722 [37:01:15<8:43:00, 33.28s/it]
 80%|████████  | 3780/4722 [37:01:47<8:39:48, 33.11s/it]
                                                        
{'loss': 0.1962, 'learning_rate': 0.0001, 'global_step': 3780, 'epoch': 2.4}
+
 80%|████████  | 3780/4722 [37:01:47<8:39:48, 33.11s/it]
 80%|████████  | 3781/4722 [37:02:20<8:37:02, 32.97s/it]
 80%|████████  | 3782/4722 [37:02:58<8:58:56, 34.40s/it]
 80%|████████  | 3783/4722 [37:03:30<8:50:41, 33.91s/it]
 80%|████████  | 3784/4722 [37:04:12<9:27:31, 36.30s/it]
 80%|████████  | 3785/4722 [37:04:45<9:10:51, 35.27s/it]
 80%|████████  | 3786/4722 [37:05:19<9:04:37, 34.91s/it]
 80%|████████  | 3787/4722 [37:05:54<9:01:44, 34.76s/it]
 80%|████████  | 3788/4722 [37:06:32<9:17:19, 35.80s/it]
 80%|████████  | 3789/4722 [37:07:03<8:54:34, 34.38s/it]
 80%|████████  | 3790/4722 [37:07:36<8:47:40, 33.97s/it]
 80%|████████  | 3791/4722 [37:08:14<9:07:59, 35.32s/it]
 80%|████████  | 3792/4722 [37:08:44<8:40:38, 33.59s/it]
 80%|████████  | 3793/4722 [37:09:15<8:27:55, 32.81s/it]
 80%|████████  | 3794/4722 [37:09:49<8:34:41, 33.28s/it]
 80%|████████  | 3795/4722 [37:10:24<8:40:00, 33.66s/it]
 80%|████████  | 3796/4722 [37:11:01<8:57:33, 34.83s/it]
 80%|████████  | 3797/4722 [37:11:30<8:25:24, 32.78s/it]
 80%|████████  | 3798/4722 [37:12:00<8:16:00, 32.21s/it]
 80%|████████  | 3799/4722 [37:12:38<8:40:34, 33.84s/it]
 80%|████████  | 3800/4722 [37:13:12<8:40:29, 33.87s/it]
                                                        
{'loss': 0.2459, 'learning_rate': 0.0001, 'global_step': 3800, 'epoch': 2.41}
+
 80%|████████  | 3800/4722 [37:13:12<8:40:29, 33.87s/it]
 80%|████████  | 3801/4722 [37:13:49<8:52:33, 34.69s/it]
 81%|████████  | 3802/4722 [37:14:23<8:50:02, 34.57s/it]
 81%|████████  | 3803/4722 [37:14:57<8:45:25, 34.30s/it]
 81%|████████  | 3804/4722 [37:15:26<8:22:03, 32.81s/it]
 81%|████████  | 3805/4722 [37:16:04<8:48:00, 34.55s/it]
 81%|████████  | 3806/4722 [37:16:44<9:10:20, 36.05s/it]
 81%|████████  | 3807/4722 [37:17:16<8:53:04, 34.96s/it]
 81%|████████  | 3808/4722 [37:17:52<8:53:33, 35.03s/it]
 81%|████████  | 3809/4722 [37:18:24<8:39:17, 34.13s/it]
 81%|████████  | 3810/4722 [37:18:59<8:42:50, 34.40s/it]
 81%|████████  | 3811/4722 [37:19:34<8:46:37, 34.68s/it]
 81%|████████  | 3812/4722 [37:20:09<8:47:59, 34.81s/it]
 81%|████████  | 3813/4722 [37:20:39<8:24:29, 33.30s/it]
 81%|████████  | 3814/4722 [37:21:15<8:35:26, 34.06s/it]
 81%|████████  | 3815/4722 [37:21:49<8:37:18, 34.22s/it]
 81%|████████  | 3816/4722 [37:22:27<8:52:07, 35.24s/it]
 81%|████████  | 3817/4722 [37:22:59<8:38:11, 34.36s/it]
 81%|████████  | 3818/4722 [37:23:24<7:53:41, 31.44s/it]
 81%|████████  | 3819/4722 [37:23:58<8:07:15, 32.38s/it]
 81%|████████  | 3820/4722 [37:24:36<8:29:52, 33.92s/it]
                                                        
{'loss': 0.3119, 'learning_rate': 0.0001, 'global_step': 3820, 'epoch': 2.43}
+
 81%|████████  | 3820/4722 [37:24:36<8:29:52, 33.92s/it]
 81%|████████  | 3821/4722 [37:25:07<8:15:13, 32.98s/it]
 81%|████████  | 3822/4722 [37:25:44<8:34:20, 34.29s/it]
 81%|████████  | 3823/4722 [37:26:21<8:44:08, 34.98s/it]
 81%|████████  | 3824/4722 [37:26:53<8:30:16, 34.09s/it]
 81%|████████  | 3825/4722 [37:27:23<8:14:27, 33.07s/it]
 81%|████████  | 3826/4722 [37:28:02<8:40:11, 34.83s/it]
 81%|████████  | 3827/4722 [37:28:34<8:24:56, 33.85s/it]
 81%|████████  | 3828/4722 [37:29:04<8:08:40, 32.80s/it]
 81%|████████  | 3829/4722 [37:29:40<8:20:22, 33.62s/it]
 81%|████████  | 3830/4722 [37:30:12<8:14:18, 33.25s/it]
 81%|████████  | 3831/4722 [37:30:41<7:54:55, 31.98s/it]
 81%|████████  | 3832/4722 [37:31:15<8:00:20, 32.38s/it]
 81%|████████  | 3833/4722 [37:31:38<7:19:17, 29.65s/it]
 81%|████████  | 3834/4722 [37:32:00<6:45:37, 27.41s/it]
 81%|████████  | 3835/4722 [37:32:23<6:24:38, 26.02s/it]
 81%|████████  | 3836/4722 [37:32:49<6:26:08, 26.15s/it]
 81%|████████▏ | 3837/4722 [37:33:13<6:13:33, 25.33s/it]
 81%|████████▏ | 3838/4722 [37:33:37<6:07:07, 24.92s/it]
 81%|████████▏ | 3839/4722 [37:34:03<6:14:49, 25.47s/it]
 81%|████████▏ | 3840/4722 [37:34:31<6:24:09, 26.13s/it]
                                                        
{'loss': 0.2237, 'learning_rate': 0.0001, 'global_step': 3840, 'epoch': 2.44}
+
 81%|████████▏ | 3840/4722 [37:34:31<6:24:09, 26.13s/it]
 81%|████████▏ | 3841/4722 [37:34:56<6:20:26, 25.91s/it]
 81%|████████▏ | 3842/4722 [37:35:22<6:17:56, 25.77s/it]
 81%|████████▏ | 3843/4722 [37:35:51<6:32:55, 26.82s/it]
 81%|████████▏ | 3844/4722 [37:36:17<6:26:38, 26.42s/it]
 81%|████████▏ | 3845/4722 [37:36:40<6:11:31, 25.42s/it]
 81%|████████▏ | 3846/4722 [37:37:03<6:01:59, 24.79s/it]
 81%|████████▏ | 3847/4722 [37:37:28<6:01:52, 24.81s/it]
 81%|████████▏ | 3848/4722 [37:37:51<5:55:30, 24.41s/it]
 82%|████████▏ | 3849/4722 [37:38:15<5:53:29, 24.29s/it]
 82%|████████▏ | 3850/4722 [37:38:44<6:13:20, 25.69s/it]
 82%|████████▏ | 3851/4722 [37:39:07<6:00:53, 24.86s/it]
 82%|████████▏ | 3852/4722 [37:39:29<5:46:43, 23.91s/it]
 82%|████████▏ | 3853/4722 [37:39:58<6:10:38, 25.59s/it]
 82%|████████▏ | 3854/4722 [37:40:20<5:52:54, 24.39s/it]
 82%|████████▏ | 3855/4722 [37:40:50<6:14:43, 25.93s/it]
 82%|████████▏ | 3856/4722 [37:41:15<6:13:23, 25.87s/it]
 82%|████████▏ | 3857/4722 [37:41:49<6:46:17, 28.18s/it]
 82%|████████▏ | 3858/4722 [37:42:24<7:15:46, 30.26s/it]
 82%|████████▏ | 3859/4722 [37:43:00<7:39:14, 31.93s/it]
 82%|████████▏ | 3860/4722 [37:43:29<7:26:08, 31.05s/it]
                                                        
{'loss': 0.2777, 'learning_rate': 0.0001, 'global_step': 3860, 'epoch': 2.45}
+
 82%|████████▏ | 3860/4722 [37:43:29<7:26:08, 31.05s/it]
 82%|████████▏ | 3861/4722 [37:44:10<8:07:31, 33.97s/it]
 82%|████████▏ | 3862/4722 [37:44:44<8:08:14, 34.06s/it]
 82%|████████▏ | 3863/4722 [37:45:17<8:01:58, 33.67s/it]
 82%|████████▏ | 3864/4722 [37:45:47<7:45:40, 32.56s/it]
 82%|████████▏ | 3865/4722 [37:46:22<7:58:45, 33.52s/it]
 82%|████████▏ | 3866/4722 [37:46:52<7:42:04, 32.39s/it]
 82%|████████▏ | 3867/4722 [37:47:25<7:44:22, 32.59s/it]
 82%|████████▏ | 3868/4722 [37:47:56<7:34:57, 31.96s/it]
 82%|████████▏ | 3869/4722 [37:48:30<7:46:32, 32.82s/it]
 82%|████████▏ | 3870/4722 [37:49:09<8:08:23, 34.39s/it]
 82%|████████▏ | 3871/4722 [37:49:41<7:58:32, 33.74s/it]
 82%|████████▏ | 3872/4722 [37:50:15<8:01:15, 33.97s/it]
 82%|████████▏ | 3873/4722 [37:50:54<8:19:16, 35.29s/it]
 82%|████████▏ | 3874/4722 [37:51:23<7:52:40, 33.44s/it]
 82%|████████▏ | 3875/4722 [37:51:55<7:47:57, 33.15s/it]
 82%|████████▏ | 3876/4722 [37:52:27<7:41:36, 32.74s/it]
 82%|████████▏ | 3877/4722 [37:53:00<7:43:31, 32.91s/it]
 82%|████████▏ | 3878/4722 [37:53:31<7:33:43, 32.25s/it]
 82%|████████▏ | 3879/4722 [37:54:04<7:36:17, 32.48s/it]
 82%|████████▏ | 3880/4722 [37:54:37<7:36:01, 32.50s/it]
                                                        
{'loss': 0.2371, 'learning_rate': 0.0001, 'global_step': 3880, 'epoch': 2.46}
+
 82%|████████▏ | 3880/4722 [37:54:37<7:36:01, 32.50s/it]
 82%|████████▏ | 3881/4722 [37:55:11<7:45:18, 33.20s/it]
 82%|████████▏ | 3882/4722 [37:55:42<7:35:39, 32.55s/it]
 82%|████████▏ | 3883/4722 [37:56:17<7:43:54, 33.18s/it]
 82%|████████▏ | 3884/4722 [37:56:48<7:34:32, 32.55s/it]
 82%|████████▏ | 3885/4722 [37:57:19<7:28:17, 32.14s/it]
 82%|████████▏ | 3886/4722 [37:57:58<7:57:00, 34.24s/it]
 82%|████████▏ | 3887/4722 [37:58:34<8:00:57, 34.56s/it]
 82%|████████▏ | 3888/4722 [37:59:06<7:51:03, 33.89s/it]
 82%|████████▏ | 3889/4722 [37:59:43<8:03:18, 34.81s/it]
 82%|████████▏ | 3890/4722 [38:00:17<7:57:50, 34.46s/it]
 82%|████████▏ | 3891/4722 [38:00:55<8:12:14, 35.54s/it]
 82%|████████▏ | 3892/4722 [38:01:26<7:53:58, 34.26s/it]
 82%|████████▏ | 3893/4722 [38:02:04<8:07:50, 35.31s/it]
 82%|████████▏ | 3894/4722 [38:02:35<7:48:37, 33.96s/it]
 82%|████████▏ | 3895/4722 [38:03:04<7:28:39, 32.55s/it]
 83%|████████▎ | 3896/4722 [38:03:33<7:14:47, 31.58s/it]
 83%|████████▎ | 3897/4722 [38:03:59<6:49:13, 29.76s/it]
 83%|████████▎ | 3898/4722 [38:04:36<7:21:36, 32.16s/it]
 83%|████████▎ | 3899/4722 [38:05:08<7:20:21, 32.10s/it]
 83%|████████▎ | 3900/4722 [38:05:43<7:30:40, 32.90s/it]
                                                        
{'loss': 0.2181, 'learning_rate': 0.0001, 'global_step': 3900, 'epoch': 2.48}
+
 83%|████████▎ | 3900/4722 [38:05:43<7:30:40, 32.90s/it]
 83%|████████▎ | 3901/4722 [38:06:27<8:16:13, 36.27s/it]
 83%|████████▎ | 3902/4722 [38:07:02<8:08:11, 35.72s/it]
 83%|████████▎ | 3903/4722 [38:07:33<7:47:27, 34.25s/it]
 83%|████████▎ | 3904/4722 [38:08:07<7:47:32, 34.29s/it]
 83%|████████▎ | 3905/4722 [38:08:42<7:50:38, 34.56s/it]
 83%|████████▎ | 3906/4722 [38:09:23<8:14:50, 36.39s/it]
 83%|████████▎ | 3907/4722 [38:10:03<8:30:01, 37.55s/it]
 83%|████████▎ | 3908/4722 [38:10:40<8:25:03, 37.23s/it]
 83%|████████▎ | 3909/4722 [38:11:21<8:40:02, 38.38s/it]
 83%|████████▎ | 3910/4722 [38:11:55<8:21:46, 37.08s/it]
 83%|████████▎ | 3911/4722 [38:12:28<8:07:59, 36.10s/it]
 83%|████████▎ | 3912/4722 [38:13:02<7:56:40, 35.31s/it]
 83%|████████▎ | 3913/4722 [38:13:34<7:44:26, 34.45s/it]
 83%|████████▎ | 3914/4722 [38:14:03<7:21:00, 32.75s/it]
 83%|████████▎ | 3915/4722 [38:14:37<7:23:45, 32.99s/it]
 83%|████████▎ | 3916/4722 [38:15:16<7:49:40, 34.96s/it]
 83%|████████▎ | 3917/4722 [38:15:47<7:31:34, 33.66s/it]
 83%|████████▎ | 3918/4722 [38:16:14<7:02:47, 31.55s/it]
 83%|████████▎ | 3919/4722 [38:16:50<7:22:44, 33.08s/it]
 83%|████████▎ | 3920/4722 [38:17:23<7:19:25, 32.87s/it]
                                                        
{'loss': 0.2603, 'learning_rate': 0.0001, 'global_step': 3920, 'epoch': 2.49}
+
 83%|████████▎ | 3920/4722 [38:17:23<7:19:25, 32.87s/it]
 83%|████████▎ | 3921/4722 [38:17:53<7:07:26, 32.02s/it]
 83%|████████▎ | 3922/4722 [38:18:25<7:10:09, 32.26s/it]
 83%|████████▎ | 3923/4722 [38:18:59<7:16:06, 32.75s/it]
 83%|████████▎ | 3924/4722 [38:19:28<7:00:37, 31.63s/it]
 83%|████████▎ | 3925/4722 [38:20:04<7:15:31, 32.79s/it]
 83%|████████▎ | 3926/4722 [38:20:36<7:11:04, 32.49s/it]
 83%|████████▎ | 3927/4722 [38:21:14<7:34:28, 34.30s/it]
 83%|████████▎ | 3928/4722 [38:21:49<7:34:31, 34.35s/it]
 83%|████████▎ | 3929/4722 [38:22:18<7:13:53, 32.83s/it]
 83%|████████▎ | 3930/4722 [38:22:50<7:09:38, 32.55s/it]
 83%|████████▎ | 3931/4722 [38:23:22<7:08:32, 32.51s/it]
 83%|████████▎ | 3932/4722 [38:23:58<7:21:29, 33.53s/it]
 83%|████████▎ | 3933/4722 [38:24:26<6:59:46, 31.92s/it]
 83%|████████▎ | 3934/4722 [38:24:59<7:02:48, 32.19s/it]
 83%|████████▎ | 3935/4722 [38:25:34<7:12:48, 33.00s/it]
 83%|████████▎ | 3936/4722 [38:26:07<7:13:42, 33.11s/it]
 83%|████████▎ | 3937/4722 [38:26:44<7:26:20, 34.12s/it]
 83%|████████▎ | 3938/4722 [38:27:12<7:02:58, 32.37s/it]
 83%|████████▎ | 3939/4722 [38:27:47<7:13:42, 33.23s/it]
 83%|████████▎ | 3940/4722 [38:28:22<7:19:23, 33.71s/it]
                                                        
{'loss': 0.2764, 'learning_rate': 0.0001, 'global_step': 3940, 'epoch': 2.5}
+
 83%|████████▎ | 3940/4722 [38:28:22<7:19:23, 33.71s/it]
 83%|████████▎ | 3941/4722 [38:28:54<7:12:00, 33.19s/it]
 83%|████████▎ | 3942/4722 [38:29:29<7:18:25, 33.73s/it]
 84%|████████▎ | 3943/4722 [38:29:59<7:03:30, 32.62s/it]
 84%|████████▎ | 3944/4722 [38:30:31<6:58:13, 32.25s/it]
 84%|████████▎ | 3945/4722 [38:31:03<6:57:48, 32.26s/it]
 84%|████████▎ | 3946/4722 [38:31:38<7:08:05, 33.10s/it]
 84%|████████▎ | 3947/4722 [38:32:08<6:55:08, 32.14s/it]
 84%|████████▎ | 3948/4722 [38:32:38<6:45:17, 31.42s/it]
 84%|████████▎ | 3949/4722 [38:33:10<6:49:00, 31.75s/it]
 84%|████████▎ | 3950/4722 [38:33:39<6:38:07, 30.94s/it]
 84%|████████▎ | 3951/4722 [38:34:19<7:10:45, 33.52s/it]
 84%|████████▎ | 3952/4722 [38:34:49<6:56:41, 32.47s/it]
 84%|████████▎ | 3953/4722 [38:35:19<6:46:34, 31.72s/it]
 84%|████████▎ | 3954/4722 [38:35:54<7:01:05, 32.90s/it]
 84%|████████▍ | 3955/4722 [38:36:26<6:54:07, 32.40s/it]
 84%|████████▍ | 3956/4722 [38:36:56<6:47:41, 31.93s/it]
 84%|████████▍ | 3957/4722 [38:37:39<7:27:51, 35.13s/it]
 84%|████████▍ | 3958/4722 [38:38:16<7:36:17, 35.83s/it]
 84%|████████▍ | 3959/4722 [38:38:49<7:23:03, 34.84s/it]
 84%|████████▍ | 3960/4722 [38:39:23<7:18:41, 34.54s/it]
                                                        
{'loss': 0.2124, 'learning_rate': 0.0001, 'global_step': 3960, 'epoch': 2.52}
+
 84%|████████▍ | 3960/4722 [38:39:23<7:18:41, 34.54s/it]
 84%|████████▍ | 3961/4722 [38:39:56<7:13:11, 34.15s/it]
 84%|████████▍ | 3962/4722 [38:40:27<7:00:15, 33.18s/it]
 84%|████████▍ | 3963/4722 [38:40:56<6:43:47, 31.92s/it]
 84%|████████▍ | 3964/4722 [38:41:29<6:48:36, 32.34s/it]
 84%|████████▍ | 3965/4722 [38:42:00<6:41:45, 31.84s/it]
 84%|████████▍ | 3966/4722 [38:42:32<6:43:42, 32.04s/it]
 84%|████████▍ | 3967/4722 [38:43:01<6:31:50, 31.14s/it]
 84%|████████▍ | 3968/4722 [38:43:39<6:53:57, 32.94s/it]
 84%|████████▍ | 3969/4722 [38:44:11<6:49:27, 32.63s/it]
 84%|████████▍ | 3970/4722 [38:44:39<6:34:35, 31.48s/it]
 84%|████████▍ | 3971/4722 [38:45:17<6:57:46, 33.38s/it]
 84%|████████▍ | 3972/4722 [38:45:55<7:15:28, 34.84s/it]
 84%|████████▍ | 3973/4722 [38:46:27<7:01:30, 33.77s/it]
 84%|████████▍ | 3974/4722 [38:46:58<6:51:00, 32.97s/it]
 84%|████████▍ | 3975/4722 [38:47:34<7:04:21, 34.09s/it]
 84%|████████▍ | 3976/4722 [38:48:10<7:08:03, 34.43s/it]
 84%|████████▍ | 3977/4722 [38:48:42<7:00:40, 33.88s/it]
 84%|████████▍ | 3978/4722 [38:49:19<7:12:12, 34.86s/it]
 84%|████████▍ | 3979/4722 [38:49:49<6:52:39, 33.32s/it]
 84%|████████▍ | 3980/4722 [38:50:20<6:43:40, 32.64s/it]
                                                        
{'loss': 0.2239, 'learning_rate': 0.0001, 'global_step': 3980, 'epoch': 2.53}
+
 84%|████████▍ | 3980/4722 [38:50:20<6:43:40, 32.64s/it]
 84%|████████▍ | 3981/4722 [38:50:58<7:00:30, 34.05s/it]
 84%|████████▍ | 3982/4722 [38:51:28<6:47:40, 33.06s/it]
 84%|████████▍ | 3983/4722 [38:52:03<6:54:04, 33.62s/it]
 84%|████████▍ | 3984/4722 [38:52:40<7:03:51, 34.46s/it]
 84%|████████▍ | 3985/4722 [38:53:19<7:20:47, 35.88s/it]
 84%|████████▍ | 3986/4722 [38:53:52<7:09:58, 35.05s/it]
 84%|████████▍ | 3987/4722 [38:54:28<7:11:37, 35.24s/it]
 84%|████████▍ | 3988/4722 [38:54:56<6:47:21, 33.30s/it]
 84%|████████▍ | 3989/4722 [38:55:28<6:39:34, 32.71s/it]
 84%|████████▍ | 3990/4722 [38:56:03<6:47:42, 33.42s/it]
 85%|████████▍ | 3991/4722 [38:56:40<7:00:12, 34.49s/it]
 85%|████████▍ | 3992/4722 [38:57:13<6:53:43, 34.00s/it]
 85%|████████▍ | 3993/4722 [38:57:54<7:18:09, 36.06s/it]
 85%|████████▍ | 3994/4722 [38:58:26<7:03:40, 34.92s/it]
 85%|████████▍ | 3995/4722 [38:59:01<7:04:47, 35.06s/it]
 85%|████████▍ | 3996/4722 [38:59:31<6:43:35, 33.35s/it]
 85%|████████▍ | 3997/4722 [38:59:59<6:25:04, 31.87s/it]
 85%|████████▍ | 3998/4722 [39:00:32<6:29:55, 32.31s/it]
 85%|████████▍ | 3999/4722 [39:01:05<6:32:11, 32.55s/it]
 85%|████████▍ | 4000/4722 [39:01:38<6:32:04, 32.58s/it]
                                                        
{'loss': 0.2812, 'learning_rate': 0.0001, 'global_step': 4000, 'epoch': 2.54}
+
 85%|████████▍ | 4000/4722 [39:01:38<6:32:04, 32.58s/it]
 85%|████████▍ | 4001/4722 [39:02:12<6:34:52, 32.86s/it]
 85%|████████▍ | 4002/4722 [39:02:54<7:09:10, 35.76s/it]
 85%|████████▍ | 4003/4722 [39:03:24<6:47:24, 34.00s/it]
 85%|████████▍ | 4004/4722 [39:04:04<7:07:22, 35.71s/it]
 85%|████████▍ | 4005/4722 [39:04:37<6:57:49, 34.96s/it]
 85%|████████▍ | 4006/4722 [39:05:10<6:49:47, 34.34s/it]
 85%|████████▍ | 4007/4722 [39:05:43<6:44:50, 33.97s/it]
 85%|████████▍ | 4008/4722 [39:06:20<6:54:44, 34.85s/it]
 85%|████████▍ | 4009/4722 [39:06:55<6:57:00, 35.09s/it]
 85%|████████▍ | 4010/4722 [39:07:30<6:55:16, 35.00s/it]
 85%|████████▍ | 4011/4722 [39:08:04<6:51:26, 34.72s/it]
 85%|████████▍ | 4012/4722 [39:08:38<6:48:12, 34.50s/it]
 85%|████████▍ | 4013/4722 [39:09:17<7:03:51, 35.87s/it]
 85%|████████▌ | 4014/4722 [39:09:57<7:17:04, 37.04s/it]
 85%|████████▌ | 4015/4722 [39:10:30<7:00:01, 35.65s/it]
 85%|████████▌ | 4016/4722 [39:10:58<6:32:51, 33.39s/it]
 85%|████████▌ | 4017/4722 [39:11:29<6:23:37, 32.65s/it]
 85%|████████▌ | 4018/4722 [39:12:19<7:25:40, 37.98s/it]
 85%|████████▌ | 4019/4722 [39:12:49<6:58:26, 35.71s/it]
 85%|████████▌ | 4020/4722 [39:13:22<6:46:35, 34.75s/it]
                                                        
{'loss': 0.1968, 'learning_rate': 0.0001, 'global_step': 4020, 'epoch': 2.55}
+
 85%|████████▌ | 4020/4722 [39:13:22<6:46:35, 34.75s/it]
 85%|████████▌ | 4021/4722 [39:13:52<6:31:02, 33.47s/it]
 85%|████████▌ | 4022/4722 [39:14:28<6:38:21, 34.14s/it]
 85%|████████▌ | 4023/4722 [39:15:03<6:39:14, 34.27s/it]
 85%|████████▌ | 4024/4722 [39:15:41<6:54:13, 35.61s/it]
 85%|████████▌ | 4025/4722 [39:16:13<6:41:15, 34.54s/it]
 85%|████████▌ | 4026/4722 [39:16:45<6:30:27, 33.66s/it]
 85%|████████▌ | 4027/4722 [39:17:19<6:31:16, 33.78s/it]
 85%|████████▌ | 4028/4722 [39:18:01<6:59:43, 36.29s/it]
 85%|████████▌ | 4029/4722 [39:18:33<6:44:58, 35.06s/it]
 85%|████████▌ | 4030/4722 [39:19:05<6:33:11, 34.09s/it]
 85%|████████▌ | 4031/4722 [39:19:37<6:26:04, 33.52s/it]
 85%|████████▌ | 4032/4722 [39:20:06<6:09:10, 32.10s/it]
 85%|████████▌ | 4033/4722 [39:20:44<6:27:13, 33.72s/it]
 85%|████████▌ | 4034/4722 [39:21:19<6:31:27, 34.14s/it]
 85%|████████▌ | 4035/4722 [39:21:51<6:23:44, 33.51s/it]
 85%|████████▌ | 4036/4722 [39:22:26<6:27:34, 33.90s/it]
 85%|████████▌ | 4037/4722 [39:22:54<6:08:42, 32.30s/it]
 86%|████████▌ | 4038/4722 [39:23:28<6:11:57, 32.63s/it]
 86%|████████▌ | 4039/4722 [39:23:59<6:05:38, 32.12s/it]
 86%|████████▌ | 4040/4722 [39:24:30<6:02:40, 31.91s/it]
                                                        
{'loss': 0.2648, 'learning_rate': 0.0001, 'global_step': 4040, 'epoch': 2.57}
+
 86%|████████▌ | 4040/4722 [39:24:30<6:02:40, 31.91s/it]
 86%|████████▌ | 4041/4722 [39:25:00<5:57:09, 31.47s/it]
 86%|████████▌ | 4042/4722 [39:25:30<5:51:00, 30.97s/it]
 86%|████████▌ | 4043/4722 [39:26:03<5:55:37, 31.43s/it]
 86%|████████▌ | 4044/4722 [39:26:44<6:28:11, 34.35s/it]
 86%|████████▌ | 4045/4722 [39:27:14<6:12:03, 32.97s/it]
 86%|████████▌ | 4046/4722 [39:27:46<6:08:50, 32.74s/it]
 86%|████████▌ | 4047/4722 [39:28:16<5:58:21, 31.85s/it]
 86%|████████▌ | 4048/4722 [39:28:53<6:15:19, 33.41s/it]
 86%|████████▌ | 4049/4722 [39:29:21<5:58:27, 31.96s/it]
 86%|████████▌ | 4050/4722 [39:29:59<6:17:26, 33.70s/it]
 86%|████████▌ | 4051/4722 [39:30:39<6:38:35, 35.64s/it]
 86%|████████▌ | 4052/4722 [39:31:10<6:20:15, 34.05s/it]
 86%|████████▌ | 4053/4722 [39:31:42<6:12:50, 33.44s/it]
 86%|████████▌ | 4054/4722 [39:32:19<6:25:34, 34.63s/it]
 86%|████████▌ | 4055/4722 [39:32:49<6:08:25, 33.14s/it]
 86%|████████▌ | 4056/4722 [39:33:23<6:11:58, 33.51s/it]
 86%|████████▌ | 4057/4722 [39:34:06<6:44:23, 36.49s/it]
 86%|████████▌ | 4058/4722 [39:34:39<6:30:40, 35.30s/it]
 86%|████████▌ | 4059/4722 [39:35:07<6:05:40, 33.09s/it]
 86%|████████▌ | 4060/4722 [39:35:41<6:08:22, 33.39s/it]
                                                        
{'loss': 0.2645, 'learning_rate': 0.0001, 'global_step': 4060, 'epoch': 2.58}
+
 86%|████████▌ | 4060/4722 [39:35:41<6:08:22, 33.39s/it]
 86%|████████▌ | 4061/4722 [39:36:11<5:57:15, 32.43s/it]
 86%|████████▌ | 4062/4722 [39:36:50<6:16:57, 34.27s/it]
 86%|████████▌ | 4063/4722 [39:37:25<6:20:59, 34.69s/it]
 86%|████████▌ | 4064/4722 [39:37:59<6:16:03, 34.29s/it]
 86%|████████▌ | 4065/4722 [39:38:31<6:07:50, 33.59s/it]
 86%|████████▌ | 4066/4722 [39:39:04<6:04:44, 33.36s/it]
 86%|████████▌ | 4067/4722 [39:39:42<6:22:16, 35.02s/it]
 86%|████████▌ | 4068/4722 [39:40:10<5:55:47, 32.64s/it]
 86%|████████▌ | 4069/4722 [39:40:39<5:43:53, 31.60s/it]
 86%|████████▌ | 4070/4722 [39:41:10<5:42:47, 31.54s/it]
 86%|████████▌ | 4071/4722 [39:41:45<5:54:31, 32.68s/it]
 86%|████████▌ | 4072/4722 [39:42:20<5:59:23, 33.17s/it]
 86%|████████▋ | 4073/4722 [39:42:54<6:03:02, 33.56s/it]
 86%|████████▋ | 4074/4722 [39:43:31<6:12:58, 34.53s/it]
 86%|████████▋ | 4075/4722 [39:44:08<6:19:18, 35.18s/it]
 86%|████████▋ | 4076/4722 [39:44:42<6:15:03, 34.83s/it]
 86%|████████▋ | 4077/4722 [39:45:10<5:54:17, 32.96s/it]
 86%|████████▋ | 4078/4722 [39:45:45<5:59:36, 33.50s/it]
 86%|████████▋ | 4079/4722 [39:46:19<6:01:28, 33.73s/it]
 86%|████████▋ | 4080/4722 [39:46:49<5:47:01, 32.43s/it]
                                                        
{'loss': 0.1914, 'learning_rate': 0.0001, 'global_step': 4080, 'epoch': 2.59}
+
 86%|████████▋ | 4080/4722 [39:46:49<5:47:01, 32.43s/it]
 86%|████████▋ | 4081/4722 [39:47:17<5:34:01, 31.27s/it]
 86%|████████▋ | 4082/4722 [39:47:54<5:50:21, 32.85s/it]
 86%|████████▋ | 4083/4722 [39:48:23<5:38:55, 31.82s/it]
 86%|████████▋ | 4084/4722 [39:48:56<5:40:34, 32.03s/it]
 87%|████████▋ | 4085/4722 [39:49:31<5:49:46, 32.95s/it]
 87%|████████▋ | 4086/4722 [39:50:06<5:57:00, 33.68s/it]
 87%|████████▋ | 4087/4722 [39:50:36<5:42:51, 32.40s/it]
 87%|████████▋ | 4088/4722 [39:51:07<5:38:54, 32.07s/it]
 87%|████████▋ | 4089/4722 [39:51:44<5:54:29, 33.60s/it]
 87%|████████▋ | 4090/4722 [39:52:16<5:47:50, 33.02s/it]
 87%|████████▋ | 4091/4722 [39:52:52<5:58:18, 34.07s/it]
 87%|████████▋ | 4092/4722 [39:53:24<5:50:13, 33.35s/it]
 87%|████████▋ | 4093/4722 [39:53:55<5:40:55, 32.52s/it]
 87%|████████▋ | 4094/4722 [39:54:28<5:42:59, 32.77s/it]
 87%|████████▋ | 4095/4722 [39:55:02<5:46:34, 33.16s/it]
 87%|████████▋ | 4096/4722 [39:55:33<5:39:48, 32.57s/it]
 87%|████████▋ | 4097/4722 [39:56:05<5:35:28, 32.21s/it]
 87%|████████▋ | 4098/4722 [39:56:38<5:40:03, 32.70s/it]
 87%|████████▋ | 4099/4722 [39:57:14<5:47:37, 33.48s/it]
 87%|��███████▋ | 4100/4722 [39:57:46<5:42:38, 33.05s/it]
                                                        
{'loss': 0.2733, 'learning_rate': 0.0001, 'global_step': 4100, 'epoch': 2.6}
+
 87%|████████▋ | 4100/4722 [39:57:46<5:42:38, 33.05s/it]
 87%|████████▋ | 4101/4722 [39:58:32<6:21:30, 36.86s/it]
 87%|████████▋ | 4102/4722 [39:59:04<6:06:11, 35.44s/it]
 87%|████████▋ | 4103/4722 [39:59:38<6:03:27, 35.23s/it]
 87%|████████▋ | 4104/4722 [40:00:14<6:03:45, 35.32s/it]
 87%|████████▋ | 4105/4722 [40:00:43<5:44:35, 33.51s/it]
 87%|████████▋ | 4106/4722 [40:01:10<5:22:39, 31.43s/it]
 87%|████████▋ | 4107/4722 [40:01:39<5:14:41, 30.70s/it]
 87%|████████▋ | 4108/4722 [40:02:10<5:15:16, 30.81s/it]
 87%|████████▋ | 4109/4722 [40:02:43<5:23:15, 31.64s/it]
 87%|████████▋ | 4110/4722 [40:03:16<5:25:50, 31.95s/it]
 87%|████████▋ | 4111/4722 [40:03:47<5:22:07, 31.63s/it]
 87%|████████▋ | 4112/4722 [40:04:20<5:26:49, 32.15s/it]
 87%|████████▋ | 4113/4722 [40:04:55<5:34:24, 32.95s/it]
 87%|████████▋ | 4114/4722 [40:05:25<5:23:12, 31.90s/it]
 87%|████████▋ | 4115/4722 [40:05:57<5:24:56, 32.12s/it]
 87%|████████▋ | 4116/4722 [40:06:27<5:17:13, 31.41s/it]
 87%|████████▋ | 4117/4722 [40:07:00<5:22:54, 32.02s/it]
 87%|████████▋ | 4118/4722 [40:07:45<6:00:48, 35.84s/it]
 87%|████████▋ | 4119/4722 [40:08:13<5:35:56, 33.43s/it]
 87%|████████▋ | 4120/4722 [40:08:50<5:46:10, 34.50s/it]
                                                        
{'loss': 0.2374, 'learning_rate': 0.0001, 'global_step': 4120, 'epoch': 2.62}
+
 87%|████████▋ | 4120/4722 [40:08:50<5:46:10, 34.50s/it]
 87%|████████▋ | 4121/4722 [40:09:20<5:33:11, 33.26s/it]
 87%|████████▋ | 4122/4722 [40:09:55<5:37:09, 33.72s/it]
 87%|████████▋ | 4123/4722 [40:10:24<5:22:55, 32.35s/it]
 87%|████████▋ | 4124/4722 [40:10:53<5:10:37, 31.17s/it]
 87%|████████▋ | 4125/4722 [40:11:23<5:07:04, 30.86s/it]
 87%|████████▋ | 4126/4722 [40:11:55<5:10:28, 31.26s/it]
 87%|████████▋ | 4127/4722 [40:12:24<5:02:54, 30.55s/it]
 87%|████████▋ | 4128/4722 [40:13:00<5:19:32, 32.28s/it]
 87%|████████▋ | 4129/4722 [40:13:33<5:19:59, 32.38s/it]
 87%|████████▋ | 4130/4722 [40:14:04<5:15:31, 31.98s/it]
 87%|████████▋ | 4131/4722 [40:14:39<5:24:24, 32.94s/it]
 88%|████████▊ | 4132/4722 [40:15:14<5:30:28, 33.61s/it]
 88%|████████▊ | 4133/4722 [40:15:55<5:51:10, 35.77s/it]
 88%|████████▊ | 4134/4722 [40:16:33<5:55:56, 36.32s/it]
 88%|████████▊ | 4135/4722 [40:17:04<5:39:27, 34.70s/it]
 88%|████████▊ | 4136/4722 [40:17:39<5:40:55, 34.91s/it]
 88%|████████▊ | 4137/4722 [40:18:11<5:32:16, 34.08s/it]
 88%|████████▊ | 4138/4722 [40:18:43<5:25:27, 33.44s/it]
 88%|████████▊ | 4139/4722 [40:19:24<5:46:19, 35.64s/it]
 88%|████████▊ | 4140/4722 [40:19:55<5:33:37, 34.39s/it]
                                                        
{'loss': 0.208, 'learning_rate': 0.0001, 'global_step': 4140, 'epoch': 2.63}
+
 88%|████████▊ | 4140/4722 [40:19:55<5:33:37, 34.39s/it]
 88%|████████▊ | 4141/4722 [40:20:27<5:24:18, 33.49s/it]
 88%|████████▊ | 4142/4722 [40:21:00<5:22:26, 33.36s/it]
 88%|████████▊ | 4143/4722 [40:21:38<5:35:18, 34.75s/it]
 88%|████████▊ | 4144/4722 [40:22:19<5:52:25, 36.58s/it]
 88%|████████▊ | 4145/4722 [40:22:48<5:31:33, 34.48s/it]
 88%|████████▊ | 4146/4722 [40:23:19<5:19:24, 33.27s/it]
 88%|████████▊ | 4147/4722 [40:23:50<5:12:00, 32.56s/it]
 88%|████████▊ | 4148/4722 [40:24:32<5:38:37, 35.40s/it]
 88%|████████▊ | 4149/4722 [40:25:02<5:22:48, 33.80s/it]
 88%|████████▊ | 4150/4722 [40:25:35<5:20:43, 33.64s/it]
 88%|████████▊ | 4151/4722 [40:26:16<5:42:36, 36.00s/it]
 88%|████████▊ | 4152/4722 [40:26:52<5:39:55, 35.78s/it]
 88%|████████▊ | 4153/4722 [40:27:25<5:31:52, 35.00s/it]
 88%|████████▊ | 4154/4722 [40:27:54<5:14:21, 33.21s/it]
 88%|████████▊ | 4155/4722 [40:28:26<5:10:54, 32.90s/it]
 88%|████████▊ | 4156/4722 [40:28:59<5:09:08, 32.77s/it]
 88%|████████▊ | 4157/4722 [40:29:31<5:07:55, 32.70s/it]
 88%|████████▊ | 4158/4722 [40:30:01<4:59:39, 31.88s/it]
 88%|████████▊ | 4159/4722 [40:30:35<5:03:33, 32.35s/it]
 88%|████████▊ | 4160/4722 [40:31:05<4:58:28, 31.87s/it]
                                                        
{'loss': 0.2691, 'learning_rate': 0.0001, 'global_step': 4160, 'epoch': 2.64}
+
 88%|████████▊ | 4160/4722 [40:31:05<4:58:28, 31.87s/it]
 88%|████████▊ | 4161/4722 [40:31:31<4:40:42, 30.02s/it]
 88%|████████▊ | 4162/4722 [40:32:05<4:52:05, 31.30s/it]
 88%|████████▊ | 4163/4722 [40:32:36<4:48:46, 30.99s/it]
 88%|████████▊ | 4164/4722 [40:33:08<4:51:59, 31.40s/it]
 88%|████████▊ | 4165/4722 [40:33:45<5:07:24, 33.11s/it]
 88%|████████▊ | 4166/4722 [40:34:20<5:11:06, 33.57s/it]
 88%|████████▊ | 4167/4722 [40:34:50<5:00:53, 32.53s/it]
 88%|████████▊ | 4168/4722 [40:35:22<4:59:33, 32.44s/it]
 88%|████████▊ | 4169/4722 [40:35:58<5:07:49, 33.40s/it]
 88%|████████▊ | 4170/4722 [40:36:32<5:10:34, 33.76s/it]
 88%|████████▊ | 4171/4722 [40:37:13<5:30:24, 35.98s/it]
 88%|████████▊ | 4172/4722 [40:37:44<5:15:40, 34.44s/it]
 88%|████████▊ | 4173/4722 [40:38:13<4:58:55, 32.67s/it]
 88%|████████▊ | 4174/4722 [40:38:42<4:49:23, 31.69s/it]
 88%|████████▊ | 4175/4722 [40:39:12<4:45:05, 31.27s/it]
 88%|████████▊ | 4176/4722 [40:39:46<4:50:09, 31.89s/it]
 88%|████████▊ | 4177/4722 [40:40:22<5:02:49, 33.34s/it]
 88%|████████▊ | 4178/4722 [40:40:56<5:03:40, 33.49s/it]
 89%|████████▊ | 4179/4722 [40:41:37<5:22:13, 35.61s/it]
 89%|████████▊ | 4180/4722 [40:42:10<5:14:53, 34.86s/it]
                                                        
{'loss': 0.2268, 'learning_rate': 0.0001, 'global_step': 4180, 'epoch': 2.66}
+
 89%|████████▊ | 4180/4722 [40:42:10<5:14:53, 34.86s/it]
 89%|████████▊ | 4181/4722 [40:42:43<5:08:34, 34.22s/it]
 89%|████████▊ | 4182/4722 [40:43:25<5:29:04, 36.56s/it]
 89%|████████▊ | 4183/4722 [40:43:58<5:18:40, 35.47s/it]
 89%|████████▊ | 4184/4722 [40:44:36<5:26:28, 36.41s/it]
 89%|████████▊ | 4185/4722 [40:45:07<5:09:42, 34.60s/it]
 89%|████████▊ | 4186/4722 [40:45:41<5:08:25, 34.53s/it]
 89%|████████▊ | 4187/4722 [40:46:19<5:17:12, 35.57s/it]
 89%|████████▊ | 4188/4722 [40:46:52<5:10:12, 34.85s/it]
 89%|████████▊ | 4189/4722 [40:47:24<5:01:41, 33.96s/it]
 89%|████████▊ | 4190/4722 [40:47:55<4:52:08, 32.95s/it]
 89%|████████▉ | 4191/4722 [40:48:27<4:50:11, 32.79s/it]
 89%|████████▉ | 4192/4722 [40:49:04<5:01:15, 34.10s/it]
 89%|████████▉ | 4193/4722 [40:49:37<4:56:20, 33.61s/it]
 89%|████████▉ | 4194/4722 [40:50:05<4:40:40, 31.90s/it]
 89%|████████▉ | 4195/4722 [40:50:34<4:32:36, 31.04s/it]
 89%|████████▉ | 4196/4722 [40:51:06<4:35:01, 31.37s/it]
 89%|████████▉ | 4197/4722 [40:51:41<4:44:35, 32.52s/it]
 89%|████████▉ | 4198/4722 [40:52:12<4:39:52, 32.05s/it]
 89%|████████▉ | 4199/4722 [40:52:49<4:53:20, 33.65s/it]
 89%|████████▉ | 4200/4722 [40:53:22<4:50:16, 33.37s/it]
                                                        
{'loss': 0.2565, 'learning_rate': 0.0001, 'global_step': 4200, 'epoch': 2.67}
+
 89%|████████▉ | 4200/4722 [40:53:22<4:50:16, 33.37s/it]
 89%|████████▉ | 4201/4722 [40:54:03<5:08:28, 35.52s/it]
 89%|████████▉ | 4202/4722 [40:54:37<5:05:16, 35.22s/it]
 89%|████████▉ | 4203/4722 [40:55:16<5:14:54, 36.41s/it]
 89%|████████▉ | 4204/4722 [40:55:47<5:00:32, 34.81s/it]
 89%|████████▉ | 4205/4722 [40:56:19<4:52:19, 33.93s/it]
 89%|████████▉ | 4206/4722 [40:56:56<4:59:01, 34.77s/it]
 89%|████████▉ | 4207/4722 [40:57:30<4:56:23, 34.53s/it]
 89%|████████▉ | 4208/4722 [40:58:06<5:00:44, 35.11s/it]
 89%|████████▉ | 4209/4722 [40:58:40<4:56:23, 34.67s/it]
 89%|████████▉ | 4210/4722 [40:59:12<4:48:54, 33.86s/it]
 89%|████████▉ | 4211/4722 [40:59:48<4:54:53, 34.62s/it]
 89%|████████▉ | 4212/4722 [41:00:22<4:52:15, 34.38s/it]
 89%|████████▉ | 4213/4722 [41:00:51<4:38:20, 32.81s/it]
 89%|████████▉ | 4214/4722 [41:01:25<4:40:20, 33.11s/it]
 89%|████████▉ | 4215/4722 [41:01:57<4:37:41, 32.86s/it]
 89%|████████▉ | 4216/4722 [41:02:36<4:51:24, 34.55s/it]
 89%|████████▉ | 4217/4722 [41:03:12<4:54:06, 34.94s/it]
 89%|████████▉ | 4218/4722 [41:03:43<4:45:15, 33.96s/it]
 89%|████████▉ | 4219/4722 [41:04:22<4:56:31, 35.37s/it]
 89%|████████▉ | 4220/4722 [41:05:00<5:00:58, 35.97s/it]
                                                        
{'loss': 0.2758, 'learning_rate': 0.0001, 'global_step': 4220, 'epoch': 2.68}
+
 89%|████████▉ | 4220/4722 [41:05:00<5:00:58, 35.97s/it]
 89%|████████▉ | 4221/4722 [41:05:30<4:46:05, 34.26s/it]
 89%|████████▉ | 4222/4722 [41:06:04<4:44:38, 34.16s/it]
 89%|████████▉ | 4223/4722 [41:06:44<4:58:05, 35.84s/it]
 89%|████████▉ | 4224/4722 [41:07:17<4:51:19, 35.10s/it]
 89%|████████▉ | 4225/4722 [41:07:48<4:41:52, 34.03s/it]
 89%|████████▉ | 4226/4722 [41:08:26<4:51:11, 35.22s/it]
 90%|████████▉ | 4227/4722 [41:09:04<4:56:42, 35.96s/it]
 90%|████████▉ | 4228/4722 [41:09:48<5:15:22, 38.30s/it]
 90%|████████▉ | 4229/4722 [41:10:22<5:03:45, 36.97s/it]
 90%|████████▉ | 4230/4722 [41:10:57<5:00:03, 36.59s/it]
 90%|████████▉ | 4231/4722 [41:11:34<4:59:19, 36.58s/it]
 90%|████████▉ | 4232/4722 [41:12:07<4:49:06, 35.40s/it]
 90%|████████▉ | 4233/4722 [41:12:45<4:55:02, 36.20s/it]
 90%|████████▉ | 4234/4722 [41:13:14<4:38:01, 34.18s/it]
 90%|████████▉ | 4235/4722 [41:13:45<4:30:08, 33.28s/it]
 90%|████████▉ | 4236/4722 [41:14:16<4:22:47, 32.44s/it]
 90%|████████▉ | 4237/4722 [41:14:46<4:17:11, 31.82s/it]
 90%|████████▉ | 4238/4722 [41:15:21<4:24:50, 32.83s/it]
 90%|████████▉ | 4239/4722 [41:15:54<4:23:14, 32.70s/it]
 90%|████████▉ | 4240/4722 [41:16:31<4:32:37, 33.94s/it]
                                                        
{'loss': 0.2662, 'learning_rate': 0.0001, 'global_step': 4240, 'epoch': 2.69}
+
 90%|████████▉ | 4240/4722 [41:16:31<4:32:37, 33.94s/it]
 90%|████████▉ | 4241/4722 [41:16:59<4:19:36, 32.38s/it]
 90%|████████▉ | 4242/4722 [41:17:30<4:14:05, 31.76s/it]
 90%|████████▉ | 4243/4722 [41:18:00<4:09:52, 31.30s/it]
 90%|████████▉ | 4244/4722 [41:18:37<4:24:09, 33.16s/it]
 90%|████████▉ | 4245/4722 [41:19:11<4:24:08, 33.22s/it]
 90%|████████▉ | 4246/4722 [41:19:46<4:29:08, 33.92s/it]
 90%|████████▉ | 4247/4722 [41:20:25<4:39:15, 35.27s/it]
 90%|████████▉ | 4248/4722 [41:21:02<4:42:22, 35.74s/it]
 90%|████████▉ | 4249/4722 [41:21:34<4:33:23, 34.68s/it]
 90%|█████████ | 4250/4722 [41:22:13<4:42:55, 35.97s/it]
 90%|█████████ | 4251/4722 [41:22:48<4:41:31, 35.86s/it]
 90%|█████████ | 4252/4722 [41:23:27<4:46:34, 36.58s/it]
 90%|█████████ | 4253/4722 [41:23:58<4:34:35, 35.13s/it]
 90%|█████████ | 4254/4722 [41:24:29<4:23:23, 33.77s/it]
 90%|█████████ | 4255/4722 [41:25:02<4:21:35, 33.61s/it]
 90%|█████████ | 4256/4722 [41:25:37<4:22:42, 33.83s/it]
 90%|█████████ | 4257/4722 [41:26:07<4:14:54, 32.89s/it]
 90%|█████████ | 4258/4722 [41:26:47<4:30:26, 34.97s/it]
 90%|█████████ | 4259/4722 [41:27:25<4:36:10, 35.79s/it]
 90%|█████████ | 4260/4722 [41:27:57<4:27:17, 34.71s/it]
                                                        
{'loss': 0.2026, 'learning_rate': 0.0001, 'global_step': 4260, 'epoch': 2.71}
+
 90%|█████████ | 4260/4722 [41:27:57<4:27:17, 34.71s/it]
 90%|█████████ | 4261/4722 [41:28:30<4:22:08, 34.12s/it]
 90%|█████████ | 4262/4722 [41:28:59<4:09:35, 32.56s/it]
 90%|█████████ | 4263/4722 [41:29:34<4:15:52, 33.45s/it]
 90%|█████████ | 4264/4722 [41:30:01<3:59:07, 31.33s/it]
 90%|█████████ | 4265/4722 [41:30:32<3:58:34, 31.32s/it]
 90%|█████████ | 4266/4722 [41:31:07<4:06:24, 32.42s/it]
 90%|█████████ | 4267/4722 [41:31:43<4:13:32, 33.43s/it]
 90%|█████████ | 4268/4722 [41:32:18<4:18:18, 34.14s/it]
 90%|█████████ | 4269/4722 [41:32:47<4:05:42, 32.54s/it]
 90%|█████████ | 4270/4722 [41:33:19<4:04:31, 32.46s/it]
 90%|█████████ | 4271/4722 [41:33:54<4:09:03, 33.13s/it]
 90%|█████████ | 4272/4722 [41:34:29<4:12:51, 33.71s/it]
 90%|█████████ | 4273/4722 [41:35:02<4:10:09, 33.43s/it]
 91%|█████████ | 4274/4722 [41:35:32<4:02:30, 32.48s/it]
 91%|█████████ | 4275/4722 [41:36:03<3:57:13, 31.84s/it]
 91%|█████████ | 4276/4722 [41:36:34<3:56:17, 31.79s/it]
 91%|█████████ | 4277/4722 [41:37:09<4:02:31, 32.70s/it]
 91%|█████████ | 4278/4722 [41:37:37<3:52:13, 31.38s/it]
 91%|█████████ | 4279/4722 [41:38:10<3:54:45, 31.80s/it]
 91%|█████████ | 4280/4722 [41:38:48<4:08:33, 33.74s/it]
                                                        
{'loss': 0.2575, 'learning_rate': 0.0001, 'global_step': 4280, 'epoch': 2.72}
+
 91%|█████████ | 4280/4722 [41:38:49<4:08:33, 33.74s/it]
 91%|█████████ | 4281/4722 [41:39:19<4:00:17, 32.69s/it]
 91%|█████████ | 4282/4722 [41:39:52<4:00:30, 32.80s/it]
 91%|█████████ | 4283/4722 [41:40:31<4:14:44, 34.82s/it]
 91%|█████████ | 4284/4722 [41:41:07<4:16:23, 35.12s/it]
 91%|█████████ | 4285/4722 [41:41:39<4:08:04, 34.06s/it]
 91%|█████████ | 4286/4722 [41:42:07<3:55:53, 32.46s/it]
 91%|█████████ | 4287/4722 [41:42:41<3:57:48, 32.80s/it]
 91%|█████████ | 4288/4722 [41:43:09<3:47:47, 31.49s/it]
 91%|█████████ | 4289/4722 [41:43:44<3:53:06, 32.30s/it]
 91%|█████████ | 4290/4722 [41:44:14<3:49:07, 31.82s/it]
 91%|█████████ | 4291/4722 [41:44:44<3:44:12, 31.21s/it]
 91%|█████████ | 4292/4722 [41:45:23<3:59:08, 33.37s/it]
 91%|█████████ | 4293/4722 [41:45:50<3:46:21, 31.66s/it]
 91%|█████████ | 4294/4722 [41:46:28<3:59:49, 33.62s/it]
 91%|█████████ | 4295/4722 [41:47:02<3:59:51, 33.70s/it]
 91%|█████████ | 4296/4722 [41:47:34<3:54:01, 32.96s/it]
 91%|█████████ | 4297/4722 [41:48:09<3:58:38, 33.69s/it]
 91%|█████████ | 4298/4722 [41:48:38<3:48:35, 32.35s/it]
 91%|█████████ | 4299/4722 [41:49:07<3:39:47, 31.18s/it]
 91%|█████████ | 4300/4722 [41:49:43<3:50:03, 32.71s/it]
                                                        
{'loss': 0.2477, 'learning_rate': 0.0001, 'global_step': 4300, 'epoch': 2.73}
+
 91%|█████████ | 4300/4722 [41:49:43<3:50:03, 32.71s/it]
 91%|█████████ | 4301/4722 [41:50:18<3:54:08, 33.37s/it]
 91%|█████████ | 4302/4722 [41:50:45<3:41:16, 31.61s/it]
 91%|█████████ | 4303/4722 [41:51:16<3:38:53, 31.34s/it]
 91%|█████████ | 4304/4722 [41:51:45<3:34:01, 30.72s/it]
 91%|█████████ | 4305/4722 [41:52:16<3:33:21, 30.70s/it]
 91%|█████████ | 4306/4722 [41:52:47<3:33:35, 30.81s/it]
 91%|█████████ | 4307/4722 [41:53:23<3:43:11, 32.27s/it]
 91%|█████████ | 4308/4722 [41:53:54<3:39:54, 31.87s/it]
 91%|█████████▏| 4309/4722 [41:54:26<3:39:40, 31.91s/it]
 91%|█████████▏| 4310/4722 [41:55:00<3:43:20, 32.53s/it]
 91%|█████████▏| 4311/4722 [41:55:32<3:42:37, 32.50s/it]
 91%|█████████▏| 4312/4722 [41:56:05<3:43:19, 32.68s/it]
 91%|█████████▏| 4313/4722 [41:56:39<3:45:43, 33.11s/it]
 91%|█████████▏| 4314/4722 [41:57:12<3:44:45, 33.05s/it]
 91%|█████████▏| 4315/4722 [41:57:50<3:54:00, 34.50s/it]
 91%|█████████▏| 4316/4722 [41:58:30<4:05:20, 36.26s/it]
 91%|█████████▏| 4317/4722 [41:58:58<3:47:47, 33.75s/it]
 91%|█████████▏| 4318/4722 [41:59:30<3:43:55, 33.26s/it]
 91%|█████████▏| 4319/4722 [42:00:06<3:48:10, 33.97s/it]
 91%|█████████▏| 4320/4722 [42:00:51<4:09:28, 37.23s/it]
                                                        
{'loss': 0.2333, 'learning_rate': 0.0001, 'global_step': 4320, 'epoch': 2.74}
+
 91%|█████████▏| 4320/4722 [42:00:51<4:09:28, 37.23s/it]
 92%|█████████▏| 4321/4722 [42:01:22<3:56:27, 35.38s/it]
 92%|█████████▏| 4322/4722 [42:01:54<3:49:34, 34.44s/it]
 92%|█████████▏| 4323/4722 [42:02:33<3:57:49, 35.76s/it]
 92%|█████████▏| 4324/4722 [42:03:04<3:47:44, 34.33s/it]
 92%|█████████▏| 4325/4722 [42:03:36<3:42:40, 33.65s/it]
 92%|█████████▏| 4326/4722 [42:04:07<3:36:31, 32.81s/it]
 92%|█████████▏| 4327/4722 [42:04:41<3:38:41, 33.22s/it]
 92%|█████████▏| 4328/4722 [42:05:13<3:36:18, 32.94s/it]
 92%|█████████▏| 4329/4722 [42:05:50<3:43:52, 34.18s/it]
 92%|█████████▏| 4330/4722 [42:06:24<3:42:52, 34.11s/it]
 92%|█████████▏| 4331/4722 [42:06:58<3:40:32, 33.84s/it]
 92%|█████████▏| 4332/4722 [42:07:33<3:41:59, 34.15s/it]
 92%|█████████▏| 4333/4722 [42:08:06<3:40:14, 33.97s/it]
 92%|█████████▏| 4334/4722 [42:08:43<3:45:07, 34.81s/it]
 92%|█████████▏| 4335/4722 [42:09:20<3:48:48, 35.47s/it]
 92%|█████████▏| 4336/4722 [42:09:49<3:36:24, 33.64s/it]
 92%|█████████▏| 4337/4722 [42:10:19<3:28:27, 32.49s/it]
 92%|█████████▏| 4338/4722 [42:10:53<3:30:02, 32.82s/it]
 92%|█████████▏| 4339/4722 [42:11:28<3:35:06, 33.70s/it]
 92%|█████████▏| 4340/4722 [42:12:05<3:41:00, 34.71s/it]
                                                        
{'loss': 0.2689, 'learning_rate': 0.0001, 'global_step': 4340, 'epoch': 2.76}
+
 92%|█████████▏| 4340/4722 [42:12:05<3:41:00, 34.71s/it]
 92%|█████████▏| 4341/4722 [42:12:36<3:32:13, 33.42s/it]
 92%|█████████▏| 4342/4722 [42:13:09<3:30:45, 33.28s/it]
 92%|█████████▏| 4343/4722 [42:13:46<3:37:02, 34.36s/it]
 92%|█████████▏| 4344/4722 [42:14:15<3:26:36, 32.80s/it]
 92%|█████████▏| 4345/4722 [42:14:51<3:32:37, 33.84s/it]
 92%|█████████▏| 4346/4722 [42:15:23<3:28:18, 33.24s/it]
 92%|█████████▏| 4347/4722 [42:15:54<3:22:54, 32.46s/it]
 92%|█████████▏| 4348/4722 [42:16:27<3:24:49, 32.86s/it]
 92%|█████████▏| 4349/4722 [42:16:58<3:19:59, 32.17s/it]
 92%|█████████▏| 4350/4722 [42:17:25<3:09:10, 30.51s/it]
 92%|█████████▏| 4351/4722 [42:17:58<3:13:07, 31.23s/it]
 92%|█████████▏| 4352/4722 [42:18:33<3:19:56, 32.42s/it]
 92%|█████████▏| 4353/4722 [42:19:04<3:17:13, 32.07s/it]
 92%|█████████▏| 4354/4722 [42:19:43<3:29:03, 34.08s/it]
 92%|█████████▏| 4355/4722 [42:20:17<3:28:16, 34.05s/it]
 92%|█████████▏| 4356/4722 [42:20:50<3:25:52, 33.75s/it]
 92%|█████████▏| 4357/4722 [42:21:24<3:25:58, 33.86s/it]
 92%|█████████▏| 4358/4722 [42:22:00<3:29:08, 34.47s/it]
 92%|█████████▏| 4359/4722 [42:22:31<3:22:02, 33.39s/it]
 92%|█████████▏| 4360/4722 [42:23:05<3:23:10, 33.68s/it]
                                                        
{'loss': 0.2046, 'learning_rate': 0.0001, 'global_step': 4360, 'epoch': 2.77}
+
 92%|█████████▏| 4360/4722 [42:23:05<3:23:10, 33.68s/it]
 92%|█████████▏| 4361/4722 [42:23:37<3:18:42, 33.03s/it]
 92%|█████████▏| 4362/4722 [42:24:09<3:17:37, 32.94s/it]
 92%|█████████▏| 4363/4722 [42:24:40<3:13:06, 32.27s/it]
 92%|█████████▏| 4364/4722 [42:25:15<3:18:22, 33.25s/it]
 92%|█████████▏| 4365/4722 [42:25:50<3:19:45, 33.57s/it]
 92%|█████████▏| 4366/4722 [42:26:22<3:16:22, 33.10s/it]
 92%|█████████▏| 4367/4722 [42:26:58<3:21:13, 34.01s/it]
 93%|█████████▎| 4368/4722 [42:27:37<3:29:06, 35.44s/it]
 93%|█████████▎| 4369/4722 [42:28:11<3:25:45, 34.97s/it]
 93%|█████████▎| 4370/4722 [42:28:46<3:26:35, 35.21s/it]
 93%|█████████▎| 4371/4722 [42:29:25<3:32:01, 36.24s/it]
 93%|█████████▎| 4372/4722 [42:29:53<3:17:14, 33.81s/it]
 93%|█████████▎| 4373/4722 [42:30:25<3:12:33, 33.11s/it]
 93%|█████████▎| 4374/4722 [42:31:00<3:15:42, 33.74s/it]
 93%|█████████▎| 4375/4722 [42:31:35<3:17:34, 34.16s/it]
 93%|█████████▎| 4376/4722 [42:32:02<3:04:53, 32.06s/it]
 93%|█████████▎| 4377/4722 [42:32:46<3:24:20, 35.54s/it]
 93%|█████████▎| 4378/4722 [42:33:23<3:26:40, 36.05s/it]
 93%|█████████▎| 4379/4722 [42:33:54<3:17:28, 34.54s/it]
 93%|█████████▎| 4380/4722 [42:34:30<3:19:49, 35.06s/it]
                                                        
{'loss': 0.2546, 'learning_rate': 0.0001, 'global_step': 4380, 'epoch': 2.78}
+
 93%|█████████▎| 4380/4722 [42:34:30<3:19:49, 35.06s/it]
 93%|█████████▎| 4381/4722 [42:35:00<3:10:00, 33.43s/it]
 93%|█████████▎| 4382/4722 [42:35:32<3:07:40, 33.12s/it]
 93%|█████████▎| 4383/4722 [42:36:12<3:17:29, 34.95s/it]
 93%|█████████▎| 4384/4722 [42:36:45<3:14:09, 34.47s/it]
 93%|█████████▎| 4385/4722 [42:37:18<3:10:27, 33.91s/it]
 93%|█████████▎| 4386/4722 [42:37:51<3:08:35, 33.68s/it]
 93%|█████████▎| 4387/4722 [42:38:26<3:10:52, 34.19s/it]
 93%|█████████▎| 4388/4722 [42:39:02<3:13:29, 34.76s/it]
 93%|█████████▎| 4389/4722 [42:39:48<3:30:44, 37.97s/it]
 93%|█████████▎| 4390/4722 [42:40:24<3:28:15, 37.64s/it]
 93%|█████████▎| 4391/4722 [42:40:52<3:10:32, 34.54s/it]
 93%|█████████▎| 4392/4722 [42:41:31<3:17:24, 35.89s/it]
 93%|█████████▎| 4393/4722 [42:42:04<3:13:05, 35.21s/it]
 93%|█████████▎| 4394/4722 [42:42:39<3:11:22, 35.01s/it]
 93%|█████████▎| 4395/4722 [42:43:15<3:12:56, 35.40s/it]
 93%|█████████▎| 4396/4722 [42:43:47<3:05:34, 34.16s/it]
 93%|█████████▎| 4397/4722 [42:44:19<3:02:43, 33.73s/it]
 93%|█████████▎| 4398/4722 [42:44:55<3:06:05, 34.46s/it]
 93%|█████████▎| 4399/4722 [42:45:32<3:08:23, 34.99s/it]
 93%|█████████▎| 4400/4722 [42:46:07<3:07:55, 35.02s/it]
                                                        
{'loss': 0.2691, 'learning_rate': 0.0001, 'global_step': 4400, 'epoch': 2.79}
+
 93%|█████████▎| 4400/4722 [42:46:07<3:07:55, 35.02s/it]
 93%|█████████▎| 4401/4722 [42:46:41<3:06:08, 34.79s/it]
 93%|█████████▎| 4402/4722 [42:47:17<3:06:56, 35.05s/it]
 93%|█████████▎| 4403/4722 [42:47:48<3:00:50, 34.01s/it]
 93%|█████████▎| 4404/4722 [42:48:20<2:57:12, 33.44s/it]
 93%|█████████▎| 4405/4722 [42:48:53<2:56:05, 33.33s/it]
 93%|█████████▎| 4406/4722 [42:49:26<2:54:18, 33.10s/it]
 93%|█████████▎| 4407/4722 [42:49:56<2:48:16, 32.05s/it]
 93%|█████████▎| 4408/4722 [42:50:34<2:57:19, 33.88s/it]
 93%|█████████▎| 4409/4722 [42:51:04<2:50:20, 32.65s/it]
 93%|█████████▎| 4410/4722 [42:51:45<3:03:40, 35.32s/it]
 93%|█████████▎| 4411/4722 [42:52:18<2:58:52, 34.51s/it]
 93%|█████████▎| 4412/4722 [42:52:54<3:01:16, 35.09s/it]
 93%|█████████▎| 4413/4722 [42:53:25<2:54:52, 33.96s/it]
 93%|█████████▎| 4414/4722 [42:54:00<2:54:58, 34.09s/it]
 93%|█████████▎| 4415/4722 [42:54:33<2:53:40, 33.94s/it]
 94%|█████████▎| 4416/4722 [42:55:05<2:49:02, 33.15s/it]
 94%|█████████▎| 4417/4722 [42:55:38<2:47:55, 33.03s/it]
 94%|█████████▎| 4418/4722 [42:56:13<2:50:32, 33.66s/it]
 94%|█████████▎| 4419/4722 [42:56:51<2:56:42, 34.99s/it]
 94%|█████████▎| 4420/4722 [42:57:24<2:53:08, 34.40s/it]
                                                        
{'loss': 0.2444, 'learning_rate': 0.0001, 'global_step': 4420, 'epoch': 2.81}
+
 94%|█████████▎| 4420/4722 [42:57:24<2:53:08, 34.40s/it]
 94%|█████████▎| 4421/4722 [42:57:55<2:47:44, 33.44s/it]
 94%|█████████▎| 4422/4722 [42:58:27<2:44:46, 32.96s/it]
 94%|█████████▎| 4423/4722 [42:59:03<2:48:30, 33.81s/it]
 94%|█████████▎| 4424/4722 [42:59:35<2:46:25, 33.51s/it]
 94%|█████████▎| 4425/4722 [43:00:08<2:44:12, 33.18s/it]
 94%|█████████▎| 4426/4722 [43:00:41<2:43:13, 33.09s/it]
 94%|█████████▍| 4427/4722 [43:01:09<2:35:51, 31.70s/it]
 94%|█████████▍| 4428/4722 [43:01:42<2:37:23, 32.12s/it]
 94%|█████████▍| 4429/4722 [43:02:15<2:37:28, 32.25s/it]
 94%|█████████▍| 4430/4722 [43:02:47<2:37:02, 32.27s/it]
 94%|█████████▍| 4431/4722 [43:03:25<2:44:49, 33.98s/it]
 94%|█████████▍| 4432/4722 [43:03:57<2:40:48, 33.27s/it]
 94%|█████████▍| 4433/4722 [43:04:35<2:47:21, 34.75s/it]
 94%|█████████▍| 4434/4722 [43:05:04<2:38:59, 33.12s/it]
 94%|█████████▍| 4435/4722 [43:05:30<2:27:35, 30.86s/it]
 94%|█████████▍| 4436/4722 [43:05:59<2:24:01, 30.21s/it]
 94%|█████████▍| 4437/4722 [43:06:33<2:29:14, 31.42s/it]
 94%|█████████▍| 4438/4722 [43:07:08<2:34:45, 32.70s/it]
 94%|█████████▍| 4439/4722 [43:07:41<2:33:26, 32.53s/it]
 94%|█████████▍| 4440/4722 [43:08:14<2:34:36, 32.90s/it]
                                                        
{'loss': 0.2667, 'learning_rate': 0.0001, 'global_step': 4440, 'epoch': 2.82}
+
 94%|█████████▍| 4440/4722 [43:08:14<2:34:36, 32.90s/it]
 94%|█████████▍| 4441/4722 [43:08:46<2:32:56, 32.66s/it]
 94%|█████████▍| 4442/4722 [43:09:19<2:32:22, 32.65s/it]
 94%|█████████▍| 4443/4722 [43:09:58<2:40:14, 34.46s/it]
 94%|█████████▍| 4444/4722 [43:10:28<2:34:14, 33.29s/it]
 94%|█████████▍| 4445/4722 [43:11:04<2:36:55, 33.99s/it]
 94%|█████████▍| 4446/4722 [43:11:38<2:36:15, 33.97s/it]
 94%|█████████▍| 4447/4722 [43:12:12<2:35:51, 34.00s/it]
 94%|█████████▍| 4448/4722 [43:12:47<2:36:58, 34.37s/it]
 94%|█████████▍| 4449/4722 [43:13:17<2:30:42, 33.12s/it]
 94%|█████████▍| 4450/4722 [43:13:51<2:30:35, 33.22s/it]
 94%|█████████▍| 4451/4722 [43:14:28<2:35:56, 34.52s/it]
 94%|█████████▍| 4452/4722 [43:14:58<2:28:43, 33.05s/it]
 94%|█████████▍| 4453/4722 [43:15:27<2:23:04, 31.91s/it]
 94%|█████████▍| 4454/4722 [43:16:02<2:25:47, 32.64s/it]
 94%|█████████▍| 4455/4722 [43:16:29<2:18:27, 31.12s/it]
 94%|█████████▍| 4456/4722 [43:17:02<2:20:44, 31.75s/it]
 94%|█████████▍| 4457/4722 [43:17:42<2:31:17, 34.26s/it]
 94%|█████████▍| 4458/4722 [43:18:17<2:31:38, 34.46s/it]
 94%|█████████▍| 4459/4722 [43:18:53<2:32:52, 34.88s/it]
 94%|█████████▍| 4460/4722 [43:19:41<2:48:51, 38.67s/it]
                                                        
{'loss': 0.2469, 'learning_rate': 0.0001, 'global_step': 4460, 'epoch': 2.83}
+
 94%|█████████▍| 4460/4722 [43:19:41<2:48:51, 38.67s/it]
 94%|█████████▍| 4461/4722 [43:20:20<2:49:11, 38.90s/it]
 94%|█████████▍| 4462/4722 [43:20:54<2:42:21, 37.47s/it]
 95%|█████████▍| 4463/4722 [43:21:26<2:34:45, 35.85s/it]
 95%|█████████▍| 4464/4722 [43:21:59<2:29:21, 34.73s/it]
 95%|█████████▍| 4465/4722 [43:22:29<2:23:46, 33.57s/it]
 95%|█████████▍| 4466/4722 [43:22:59<2:18:16, 32.41s/it]
 95%|█████████▍| 4467/4722 [43:23:32<2:18:56, 32.69s/it]
 95%|█████████▍| 4468/4722 [43:24:14<2:29:38, 35.35s/it]
 95%|█████████▍| 4469/4722 [43:24:46<2:25:11, 34.43s/it]
 95%|█████████▍| 4470/4722 [43:25:20<2:23:54, 34.26s/it]
 95%|█████████▍| 4471/4722 [43:25:52<2:20:46, 33.65s/it]
 95%|█████████▍| 4472/4722 [43:26:28<2:23:06, 34.35s/it]
 95%|█████████▍| 4473/4722 [43:27:00<2:19:45, 33.68s/it]
 95%|█████████▍| 4474/4722 [43:27:32<2:16:59, 33.14s/it]
 95%|█████████▍| 4475/4722 [43:28:04<2:14:41, 32.72s/it]
 95%|█████████▍| 4476/4722 [43:28:34<2:11:14, 32.01s/it]
 95%|█████████▍| 4477/4722 [43:29:04<2:07:32, 31.24s/it]
 95%|█████████▍| 4478/4722 [43:29:33<2:04:44, 30.67s/it]
 95%|█████████▍| 4479/4722 [43:30:10<2:11:48, 32.54s/it]
 95%|█████████▍| 4480/4722 [43:30:41<2:09:07, 32.02s/it]
                                                        
{'loss': 0.2684, 'learning_rate': 0.0001, 'global_step': 4480, 'epoch': 2.85}
+
 95%|█████████▍| 4480/4722 [43:30:41<2:09:07, 32.02s/it]
 95%|█████████▍| 4481/4722 [43:31:17<2:12:52, 33.08s/it]
 95%|█████████▍| 4482/4722 [43:31:54<2:17:14, 34.31s/it]
 95%|█████████▍| 4483/4722 [43:32:28<2:16:47, 34.34s/it]
 95%|█████████▍| 4484/4722 [43:33:06<2:20:35, 35.44s/it]
 95%|█████████▍| 4485/4722 [43:33:38<2:15:42, 34.36s/it]
 95%|█████████▌| 4486/4722 [43:34:15<2:17:50, 35.05s/it]
 95%|█████████▌| 4487/4722 [43:34:51<2:18:49, 35.45s/it]
 95%|█████████▌| 4488/4722 [43:35:20<2:11:09, 33.63s/it]
 95%|█████████▌| 4489/4722 [43:35:50<2:06:20, 32.53s/it]
 95%|█████████▌| 4490/4722 [43:36:21<2:03:49, 32.03s/it]
 95%|█████████▌| 4491/4722 [43:37:00<2:11:12, 34.08s/it]
 95%|█████████▌| 4492/4722 [43:37:34<2:10:51, 34.14s/it]
 95%|█████████▌| 4493/4722 [43:38:04<2:04:56, 32.74s/it]
 95%|█████████▌| 4494/4722 [43:38:38<2:05:49, 33.11s/it]
 95%|█████████▌| 4495/4722 [43:39:09<2:02:43, 32.44s/it]
 95%|█████████▌| 4496/4722 [43:39:40<2:01:07, 32.16s/it]
 95%|█████████▌| 4497/4722 [43:40:18<2:07:22, 33.97s/it]
 95%|█████████▌| 4498/4722 [43:40:52<2:06:49, 33.97s/it]
 95%|█████████▌| 4499/4722 [43:41:24<2:03:23, 33.20s/it]
 95%|█████████▌| 4500/4722 [43:42:04<2:10:12, 35.19s/it]
                                                        
{'loss': 0.2077, 'learning_rate': 0.0001, 'global_step': 4500, 'epoch': 2.86}
+
 95%|█████████▌| 4500/4722 [43:42:04<2:10:12, 35.19s/it]
 95%|█████████▌| 4501/4722 [43:42:37<2:08:14, 34.81s/it]
 95%|█████████▌| 4502/4722 [43:43:08<2:02:52, 33.51s/it]
 95%|█████████▌| 4503/4722 [43:43:38<1:58:49, 32.56s/it]
 95%|█████████▌| 4504/4722 [43:44:07<1:53:41, 31.29s/it]
 95%|█████████▌| 4505/4722 [43:44:40<1:55:18, 31.88s/it]
 95%|█████████▌| 4506/4722 [43:45:11<1:54:05, 31.69s/it]
 95%|█████████▌| 4507/4722 [43:45:43<1:54:12, 31.87s/it]
 95%|█████████▌| 4508/4722 [43:46:21<1:59:38, 33.55s/it]
 95%|█████████▌| 4509/4722 [43:46:56<2:00:16, 33.88s/it]
 96%|█████████▌| 4510/4722 [43:47:35<2:05:21, 35.48s/it]
 96%|█████████▌| 4511/4722 [43:48:12<2:06:23, 35.94s/it]
 96%|█████████▌| 4512/4722 [43:48:47<2:04:54, 35.69s/it]
 96%|█████████▌| 4513/4722 [43:49:24<2:05:24, 36.00s/it]
 96%|█████████▌| 4514/4722 [43:49:52<1:56:39, 33.65s/it]
 96%|█████████▌| 4515/4722 [43:50:29<1:59:23, 34.60s/it]
 96%|█████████▌| 4516/4722 [43:51:03<1:58:33, 34.53s/it]
 96%|█████████▌| 4517/4722 [43:51:36<1:56:36, 34.13s/it]
 96%|█████████▌| 4518/4722 [43:52:10<1:56:01, 34.13s/it]
 96%|█████████▌| 4519/4722 [43:52:40<1:51:17, 32.90s/it]
 96%|█████████▌| 4520/4722 [43:53:16<1:53:22, 33.68s/it]
                                                        
{'loss': 0.2208, 'learning_rate': 0.0001, 'global_step': 4520, 'epoch': 2.87}
+
 96%|█████████▌| 4520/4722 [43:53:16<1:53:22, 33.68s/it]
 96%|█████████▌| 4521/4722 [43:53:47<1:50:16, 32.92s/it]
 96%|█████████▌| 4522/4722 [43:54:19<1:48:24, 32.52s/it]
 96%|█████████▌| 4523/4722 [43:54:51<1:47:38, 32.45s/it]
 96%|█████████▌| 4524/4722 [43:55:26<1:49:17, 33.12s/it]
 96%|█████████▌| 4525/4722 [43:55:53<1:43:18, 31.47s/it]
 96%|█████████▌| 4526/4722 [43:56:23<1:41:39, 31.12s/it]
 96%|█████████▌| 4527/4722 [43:56:54<1:40:53, 31.04s/it]
 96%|█████████▌| 4528/4722 [43:57:25<1:39:43, 30.84s/it]
 96%|█████████▌| 4529/4722 [43:57:58<1:41:37, 31.59s/it]
 96%|█████████▌| 4530/4722 [43:58:33<1:44:30, 32.66s/it]
 96%|█████████▌| 4531/4722 [43:59:04<1:42:42, 32.26s/it]
 96%|█████████▌| 4532/4722 [43:59:36<1:41:02, 31.91s/it]
 96%|█████████▌| 4533/4722 [44:00:10<1:42:50, 32.65s/it]
 96%|█████████▌| 4534/4722 [44:00:48<1:47:15, 34.23s/it]
 96%|█████████▌| 4535/4722 [44:01:21<1:46:06, 34.04s/it]
 96%|█████████▌| 4536/4722 [44:01:56<1:45:58, 34.18s/it]
 96%|█████████▌| 4537/4722 [44:02:25<1:40:49, 32.70s/it]
 96%|█████████▌| 4538/4722 [44:02:57<1:39:05, 32.31s/it]
 96%|█████████▌| 4539/4722 [44:03:34<1:43:15, 33.86s/it]
 96%|█████████▌| 4540/4722 [44:04:03<1:37:44, 32.22s/it]
                                                        
{'loss': 0.2401, 'learning_rate': 0.0001, 'global_step': 4540, 'epoch': 2.88}
+
 96%|█████████▌| 4540/4722 [44:04:03<1:37:44, 32.22s/it]
 96%|█████████▌| 4541/4722 [44:04:37<1:39:36, 33.02s/it]
 96%|█████████▌| 4542/4722 [44:05:11<1:39:37, 33.21s/it]
 96%|█████████▌| 4543/4722 [44:05:47<1:41:19, 33.96s/it]
 96%|█████████▌| 4544/4722 [44:06:17<1:37:35, 32.90s/it]
 96%|█████████▋| 4545/4722 [44:06:46<1:33:13, 31.60s/it]
 96%|█████████▋| 4546/4722 [44:07:20<1:34:55, 32.36s/it]
 96%|█████████▋| 4547/4722 [44:07:54<1:35:35, 32.77s/it]
 96%|█████████▋| 4548/4722 [44:08:23<1:32:30, 31.90s/it]
 96%|█████████▋| 4549/4722 [44:08:53<1:29:35, 31.07s/it]
 96%|█████████▋| 4550/4722 [44:09:25<1:30:22, 31.53s/it]
 96%|█████████▋| 4551/4722 [44:09:56<1:28:51, 31.18s/it]
 96%|█████████▋| 4552/4722 [44:10:34<1:34:29, 33.35s/it]
 96%|█████████▋| 4553/4722 [44:11:03<1:29:54, 31.92s/it]
 96%|█████████▋| 4554/4722 [44:11:34<1:29:05, 31.82s/it]
 96%|█████████▋| 4555/4722 [44:12:14<1:35:19, 34.25s/it]
 96%|█████████▋| 4556/4722 [44:12:43<1:30:17, 32.64s/it]
 97%|█████████▋| 4557/4722 [44:13:10<1:24:44, 30.81s/it]
 97%|█████████▋| 4558/4722 [44:13:40<1:23:49, 30.67s/it]
 97%|█████████▋| 4559/4722 [44:14:10<1:22:59, 30.55s/it]
 97%|█████████▋| 4560/4722 [44:14:46<1:26:50, 32.17s/it]
                                                        
{'loss': 0.2773, 'learning_rate': 0.0001, 'global_step': 4560, 'epoch': 2.9}
+
 97%|█████████▋| 4560/4722 [44:14:46<1:26:50, 32.17s/it]
 97%|█████████▋| 4561/4722 [44:15:18<1:26:15, 32.15s/it]
 97%|█████████▋| 4562/4722 [44:15:50<1:25:17, 31.98s/it]
 97%|█████████▋| 4563/4722 [44:16:18<1:22:03, 30.97s/it]
 97%|█████████▋| 4564/4722 [44:16:56<1:26:30, 32.85s/it]
 97%|█████████▋| 4565/4722 [44:17:33<1:29:53, 34.36s/it]
 97%|█████████▋| 4566/4722 [44:18:16<1:35:23, 36.69s/it]
 97%|█████████▋| 4567/4722 [44:18:48<1:31:08, 35.28s/it]
 97%|█████████▋| 4568/4722 [44:19:20<1:28:37, 34.53s/it]
 97%|█████████▋| 4569/4722 [44:19:50<1:24:17, 33.05s/it]
 97%|█████████▋| 4570/4722 [44:20:24<1:24:22, 33.31s/it]
 97%|��████████▋| 4571/4722 [44:20:52<1:20:15, 31.89s/it]
 97%|█████████▋| 4572/4722 [44:21:23<1:18:50, 31.53s/it]
 97%|█████████▋| 4573/4722 [44:21:48<1:13:17, 29.51s/it]
 97%|█████████▋| 4574/4722 [44:22:26<1:19:09, 32.09s/it]
 97%|█████████▋| 4575/4722 [44:23:00<1:19:37, 32.50s/it]
 97%|█████████▋| 4576/4722 [44:23:27<1:15:24, 30.99s/it]
 97%|█████████▋| 4577/4722 [44:24:06<1:21:01, 33.53s/it]
 97%|█████████▋| 4578/4722 [44:24:43<1:22:37, 34.42s/it]
 97%|█████████▋| 4579/4722 [44:25:21<1:24:25, 35.42s/it]
 97%|█████████▋| 4580/4722 [44:25:54<1:22:26, 34.84s/it]
                                                        
{'loss': 0.1959, 'learning_rate': 0.0001, 'global_step': 4580, 'epoch': 2.91}
+
 97%|█████████▋| 4580/4722 [44:25:54<1:22:26, 34.84s/it]
 97%|█████████▋| 4581/4722 [44:26:23<1:17:31, 32.99s/it]
 97%|█████████▋| 4582/4722 [44:27:07<1:25:07, 36.48s/it]
 97%|█████████▋| 4583/4722 [44:27:45<1:25:19, 36.83s/it]
 97%|█████████▋| 4584/4722 [44:28:18<1:21:39, 35.50s/it]
 97%|█████████▋| 4585/4722 [44:28:51<1:19:48, 34.95s/it]
 97%|█████████▋| 4586/4722 [44:29:28<1:20:13, 35.39s/it]
 97%|█████████▋| 4587/4722 [44:29:59<1:17:06, 34.27s/it]
 97%|█████████▋| 4588/4722 [44:30:28<1:12:39, 32.53s/it]
 97%|█████████▋| 4589/4722 [44:30:57<1:09:48, 31.49s/it]
 97%|█████████▋| 4590/4722 [44:31:30<1:10:22, 31.99s/it]
 97%|█████████▋| 4591/4722 [44:32:03<1:10:21, 32.23s/it]
 97%|█████████▋| 4592/4722 [44:32:42<1:14:39, 34.46s/it]
 97%|█████████▋| 4593/4722 [44:33:18<1:14:45, 34.77s/it]
 97%|█████████▋| 4594/4722 [44:33:48<1:10:58, 33.27s/it]
 97%|█████████▋| 4595/4722 [44:34:18<1:08:31, 32.37s/it]
 97%|█████████▋| 4596/4722 [44:34:48<1:06:50, 31.83s/it]
 97%|█████████▋| 4597/4722 [44:35:20<1:06:14, 31.80s/it]
 97%|█████████▋| 4598/4722 [44:35:52<1:05:34, 31.73s/it]
 97%|█████████▋| 4599/4722 [44:36:29<1:08:40, 33.50s/it]
 97%|█████████▋| 4600/4722 [44:37:03<1:07:53, 33.39s/it]
                                                        
{'loss': 0.2761, 'learning_rate': 0.0001, 'global_step': 4600, 'epoch': 2.92}
+
 97%|█████████▋| 4600/4722 [44:37:03<1:07:53, 33.39s/it]
 97%|█████████▋| 4601/4722 [44:37:37<1:07:44, 33.59s/it]
 97%|█████████▋| 4602/4722 [44:38:11<1:07:35, 33.79s/it]
 97%|█████████▋| 4603/4722 [44:38:50<1:10:20, 35.46s/it]
 98%|█████████▊| 4604/4722 [44:39:26<1:09:37, 35.40s/it]
 98%|█████████▊| 4605/4722 [44:40:04<1:10:51, 36.33s/it]
 98%|█████████▊| 4606/4722 [44:40:40<1:10:00, 36.22s/it]
 98%|█████████▊| 4607/4722 [44:41:23<1:13:32, 38.37s/it]
 98%|█████████▊| 4608/4722 [44:41:55<1:08:59, 36.31s/it]
 98%|█████████▊| 4609/4722 [44:42:24<1:04:05, 34.03s/it]
 98%|█████████▊| 4610/4722 [44:42:56<1:02:38, 33.56s/it]
 98%|█████████▊| 4611/4722 [44:43:26<59:51, 32.36s/it]  
 98%|█████████▊| 4612/4722 [44:44:05<1:03:05, 34.41s/it]
 98%|█████████▊| 4613/4722 [44:44:40<1:03:11, 34.78s/it]
 98%|█████████▊| 4614/4722 [44:45:10<59:37, 33.13s/it]  
 98%|█████████▊| 4615/4722 [44:45:47<1:01:32, 34.50s/it]
 98%|█████████▊| 4616/4722 [44:46:30<1:05:13, 36.92s/it]
 98%|█████████▊| 4617/4722 [44:47:10<1:06:16, 37.87s/it]
 98%|█████████▊| 4618/4722 [44:47:41<1:02:00, 35.78s/it]
 98%|█████████▊| 4619/4722 [44:48:14<59:51, 34.87s/it]  
 98%|█████████▊| 4620/4722 [44:48:42<56:02, 32.96s/it]
                                                      
{'loss': 0.2821, 'learning_rate': 0.0001, 'global_step': 4620, 'epoch': 2.93}
+
 98%|█████████▊| 4620/4722 [44:48:42<56:02, 32.96s/it]
 98%|█████████▊| 4621/4722 [44:49:15<55:27, 32.94s/it]
 98%|█████████▊| 4622/4722 [44:49:45<53:31, 32.12s/it]
 98%|█████████▊| 4623/4722 [44:50:24<56:20, 34.15s/it]
 98%|█████████▊| 4624/4722 [44:50:55<53:55, 33.01s/it]
 98%|█████████▊| 4625/4722 [44:51:26<52:33, 32.51s/it]
 98%|█████████▊| 4626/4722 [44:52:02<53:55, 33.70s/it]
 98%|█████████▊| 4627/4722 [44:52:34<52:26, 33.12s/it]
 98%|█████████▊| 4628/4722 [44:53:07<51:54, 33.14s/it]
 98%|█████████▊| 4629/4722 [44:53:34<48:24, 31.23s/it]
 98%|█████████▊| 4630/4722 [44:54:08<49:04, 32.00s/it]
 98%|█████████▊| 4631/4722 [44:54:37<47:15, 31.16s/it]
 98%|█████████▊| 4632/4722 [44:55:15<49:34, 33.05s/it]
 98%|█████████▊| 4633/4722 [44:55:48<49:22, 33.29s/it]
 98%|█████████▊| 4634/4722 [44:56:21<48:25, 33.02s/it]
 98%|█████████▊| 4635/4722 [44:56:53<47:40, 32.88s/it]
 98%|█████████▊| 4636/4722 [44:57:20<44:30, 31.06s/it]
 98%|█████████▊| 4637/4722 [44:57:54<45:15, 31.95s/it]
 98%|█████████▊| 4638/4722 [44:58:21<42:32, 30.38s/it]
 98%|█████████▊| 4639/4722 [44:58:50<41:36, 30.08s/it]
 98%|█████████▊| 4640/4722 [44:59:34<46:32, 34.06s/it]
                                                      
{'loss': 0.2815, 'learning_rate': 0.0001, 'global_step': 4640, 'epoch': 2.95}
+
 98%|█████████▊| 4640/4722 [44:59:34<46:32, 34.06s/it]
 98%|█████████▊| 4641/4722 [45:00:02<43:44, 32.40s/it]
 98%|█████████▊| 4642/4722 [45:00:34<43:02, 32.28s/it]
 98%|█████████▊| 4643/4722 [45:01:04<41:41, 31.67s/it]
 98%|█████████▊| 4644/4722 [45:01:35<40:44, 31.34s/it]
 98%|█████████▊| 4645/4722 [45:02:05<39:38, 30.89s/it]
 98%|█████████▊| 4646/4722 [45:02:39<40:18, 31.82s/it]
 98%|█████████▊| 4647/4722 [45:03:13<40:28, 32.38s/it]
 98%|█████████▊| 4648/4722 [45:03:45<40:01, 32.46s/it]
 98%|█████████▊| 4649/4722 [45:04:17<39:12, 32.22s/it]
 98%|█████████▊| 4650/4722 [45:04:46<37:30, 31.25s/it]
 98%|█████████▊| 4651/4722 [45:05:18<37:24, 31.62s/it]
 99%|█████████▊| 4652/4722 [45:05:50<37:03, 31.76s/it]
 99%|█████████▊| 4653/4722 [45:06:15<34:07, 29.68s/it]
 99%|█████████▊| 4654/4722 [45:06:50<35:23, 31.22s/it]
 99%|█████████▊| 4655/4722 [45:07:31<38:04, 34.10s/it]
 99%|█████████▊| 4656/4722 [45:08:01<36:19, 33.03s/it]
 99%|█████████▊| 4657/4722 [45:08:34<35:46, 33.02s/it]
 99%|█████████▊| 4658/4722 [45:09:07<35:13, 33.02s/it]
 99%|█████████▊| 4659/4722 [45:09:45<36:06, 34.40s/it]
 99%|█████████▊| 4660/4722 [45:10:13<33:42, 32.62s/it]
                                                      
{'loss': 0.1933, 'learning_rate': 0.0001, 'global_step': 4660, 'epoch': 2.96}
+
 99%|█████████▊| 4660/4722 [45:10:13<33:42, 32.62s/it]
 99%|█████████▊| 4661/4722 [45:10:42<31:56, 31.42s/it]
 99%|█████████▊| 4662/4722 [45:11:12<31:03, 31.05s/it]
 99%|█████████▉| 4663/4722 [45:11:51<32:48, 33.36s/it]
 99%|█████████▉| 4664/4722 [45:12:26<32:49, 33.96s/it]
 99%|█████████▉| 4665/4722 [45:12:53<30:12, 31.80s/it]
 99%|█████████▉| 4666/4722 [45:13:24<29:27, 31.57s/it]
 99%|█████████▉| 4667/4722 [45:14:01<30:30, 33.28s/it]
 99%|█████████▉| 4668/4722 [45:14:34<29:43, 33.02s/it]
 99%|█████████▉| 4669/4722 [45:15:09<29:45, 33.70s/it]
 99%|█████████▉| 4670/4722 [45:15:47<30:21, 35.03s/it]
 99%|█████████▉| 4671/4722 [45:16:21<29:20, 34.51s/it]
 99%|█████████▉| 4672/4722 [45:16:59<29:45, 35.70s/it]
 99%|█████████▉| 4673/4722 [45:17:32<28:26, 34.83s/it]
 99%|█████████▉| 4674/4722 [45:18:03<27:03, 33.82s/it]
 99%|█████████▉| 4675/4722 [45:18:35<26:00, 33.21s/it]
 99%|█████████▉| 4676/4722 [45:19:04<24:30, 31.96s/it]
 99%|█████████▉| 4677/4722 [45:19:37<24:06, 32.14s/it]
 99%|█████████▉| 4678/4722 [45:20:12<24:16, 33.09s/it]
 99%|█████████▉| 4679/4722 [45:20:46<23:52, 33.32s/it]
 99%|█████████▉| 4680/4722 [45:21:16<22:34, 32.26s/it]
                                                      
{'loss': 0.2199, 'learning_rate': 0.0001, 'global_step': 4680, 'epoch': 2.97}
+
 99%|█████████▉| 4680/4722 [45:21:16<22:34, 32.26s/it]
 99%|█████████▉| 4681/4722 [45:21:50<22:24, 32.79s/it]
 99%|█████████▉| 4682/4722 [45:22:33<24:02, 36.06s/it]
 99%|█████████▉| 4683/4722 [45:23:13<24:05, 37.06s/it]
 99%|█████████▉| 4684/4722 [45:23:46<22:49, 36.05s/it]
 99%|█████████▉| 4685/4722 [45:24:20<21:48, 35.38s/it]
 99%|█████████▉| 4686/4722 [45:24:53<20:41, 34.49s/it]
 99%|█████████▉| 4687/4722 [45:25:29<20:24, 34.99s/it]
 99%|█████████▉| 4688/4722 [45:25:57<18:37, 32.87s/it]
 99%|████████��▉| 4689/4722 [45:26:30<18:04, 32.87s/it]
 99%|█████████▉| 4690/4722 [45:27:05<17:53, 33.54s/it]
 99%|█████████▉| 4691/4722 [45:27:36<17:01, 32.94s/it]
 99%|█████████▉| 4692/4722 [45:28:08<16:15, 32.53s/it]
 99%|█████████▉| 4693/4722 [45:28:40<15:37, 32.34s/it]
 99%|█████████▉| 4694/4722 [45:29:12<15:04, 32.32s/it]
 99%|█████████▉| 4695/4722 [45:29:44<14:33, 32.35s/it]
 99%|█████████▉| 4696/4722 [45:30:20<14:25, 33.29s/it]
 99%|█████████▉| 4697/4722 [45:30:58<14:25, 34.61s/it]
 99%|█████████▉| 4698/4722 [45:31:31<13:45, 34.38s/it]
100%|█████████▉| 4699/4722 [45:32:04<12:55, 33.72s/it]
100%|█████████▉| 4700/4722 [45:32:35<12:08, 33.12s/it]
                                                      
{'loss': 0.3124, 'learning_rate': 0.0001, 'global_step': 4700, 'epoch': 2.99}
+
100%|█████████▉| 4700/4722 [45:32:35<12:08, 33.12s/it]
100%|█████████▉| 4701/4722 [45:33:12<11:55, 34.09s/it]
100%|█████████▉| 4702/4722 [45:33:50<11:45, 35.25s/it]
100%|█████████▉| 4703/4722 [45:34:21<10:47, 34.06s/it]
100%|█████████▉| 4704/4722 [45:34:55<10:13, 34.10s/it]
100%|█████████▉| 4705/4722 [45:35:24<09:13, 32.56s/it]
100%|█████████▉| 4706/4722 [45:35:52<08:20, 31.25s/it]
100%|█████████▉| 4707/4722 [45:36:25<07:55, 31.69s/it]
100%|█████████▉| 4708/4722 [45:36:59<07:31, 32.24s/it]
100%|█████████▉| 4709/4722 [45:37:33<07:06, 32.79s/it]
100%|█████████▉| 4710/4722 [45:38:05<06:32, 32.74s/it]
100%|█████████▉| 4711/4722 [45:38:33<05:43, 31.23s/it]
100%|█████████▉| 4712/4722 [45:39:04<05:12, 31.28s/it]
100%|█████████▉| 4713/4722 [45:39:36<04:42, 31.39s/it]
100%|█████████▉| 4714/4722 [45:40:09<04:15, 31.96s/it]
100%|█████████▉| 4715/4722 [45:40:38<03:37, 31.09s/it]
100%|█████████▉| 4716/4722 [45:41:13<03:12, 32.04s/it]
100%|█████████▉| 4717/4722 [45:41:49<02:46, 33.36s/it]
100%|█████████▉| 4718/4722 [45:42:21<02:11, 32.83s/it]
100%|█████████▉| 4719/4722 [45:42:57<01:41, 33.99s/it]
100%|█████████▉| 4720/4722 [45:43:32<01:08, 34.30s/it]
                                                      
{'loss': 0.2222, 'learning_rate': 0.0001, 'global_step': 4720, 'epoch': 3.0}
+
100%|█████████▉| 4720/4722 [45:43:32<01:08, 34.30s/it]
100%|█████████▉| 4721/4722 [45:44:02<00:32, 32.77s/it]
100%|██████████| 4722/4722 [45:44:29<00:00, 31.26s/it]
                                                      
{'train_runtime': 164669.988, 'train_samples_per_second': 0.344, 'train_steps_per_second': 0.029, 'train_loss': 0.5372816354023708, 'epoch': 3.0}
+
100%|██████████| 4722/4722 [45:44:29<00:00, 31.26s/it]
100%|██████████| 4722/4722 [45:44:29<00:00, 34.87s/it]
+***** train metrics *****
+  epoch                    =                3.0
+  train_loss               =             0.5373
+  train_runtime            = 1 day, 21:44:29.98
+  train_samples_per_second =              0.344
+  train_steps_per_second   =              0.029