diff --git "a/wandb/run-20220504_095140-cwhobv6l/files/output.log" "b/wandb/run-20220504_095140-cwhobv6l/files/output.log"
new file mode 100644--- /dev/null
+++ "b/wandb/run-20220504_095140-cwhobv6l/files/output.log"
@@ -0,0 +1,2987 @@
+wandb: WARNING Config item 'output_dir' was locked by 'sweep' (ignored update).
+wandb: WARNING Config item 'evaluation_strategy' was locked by 'sweep' (ignored update).
+wandb: WARNING Config item 'per_device_train_batch_size' was locked by 'sweep' (ignored update).
+wandb: WARNING Config item 'per_device_eval_batch_size' was locked by 'sweep' (ignored update).
+wandb: WARNING Config item 'gradient_accumulation_steps' was locked by 'sweep' (ignored update).
+wandb: WARNING Config item 'learning_rate' was locked by 'sweep' (ignored update).
+wandb: WARNING Config item 'num_train_epochs' was locked by 'sweep' (ignored update).
+wandb: WARNING Config item 'warmup_steps' was locked by 'sweep' (ignored update).
+wandb: WARNING Config item 'logging_steps' was locked by 'sweep' (ignored update).
+wandb: WARNING Config item 'save_steps' was locked by 'sweep' (ignored update).
+wandb: WARNING Config item 'eval_steps' was locked by 'sweep' (ignored update).
+wandb: WARNING Config item 'metric_for_best_model' was locked by 'sweep' (ignored update).
+wandb: WARNING Config item 'greater_is_better' was locked by 'sweep' (ignored update).
+wandb: WARNING Config item 'generation_max_length' was locked by 'sweep' (ignored update).
+wandb: WARNING Config item 'generation_num_beams' was locked by 'sweep' (ignored update).
+  0%|                                                                                         | 0/19440 [00:00<?, ?it/s]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 11.7968, 'learning_rate': 0.0, 'epoch': 0.0}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|                                                                              | 1/19440 [00:05<27:12:52,  5.04s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 11.1564, 'learning_rate': 0.0, 'epoch': 0.0}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|                                                                              | 2/19440 [00:09<24:40:19,  4.57s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 10.891, 'learning_rate': 0.0, 'epoch': 0.0}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|                                                                              | 3/19440 [00:13<23:26:03,  4.34s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 11.5633, 'learning_rate': 0.0, 'epoch': 0.0}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|                                                                              | 4/19440 [00:17<22:26:41,  4.16s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 11.4986, 'learning_rate': 0.0, 'epoch': 0.0}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|                                                                              | 5/19440 [00:21<21:43:00,  4.02s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 10.5653, 'learning_rate': 0.0, 'epoch': 0.0}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|                                                                              | 6/19440 [00:24<21:06:13,  3.91s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 10.8823, 'learning_rate': 1.805451340454541e-06, 'epoch': 0.0}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|                                                                              | 7/19440 [00:28<20:55:09,  3.88s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 10.8754, 'learning_rate': 3.610902680909082e-06, 'epoch': 0.0}
+  0%|                                                                              | 8/19440 [00:32<20:38:55,  3.83s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|                                                                              | 9/19440 [00:35<20:16:11,  3.76s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 10.6607, 'learning_rate': 5.416354021363623e-06, 'epoch': 0.0}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 10.7778, 'learning_rate': 7.221805361818164e-06, 'epoch': 0.0}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|                                                                             | 10/19440 [00:39<19:51:58,  3.68s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 8.9753, 'learning_rate': 9.027256702272705e-06, 'epoch': 0.0}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|                                                                             | 11/19440 [00:42<19:30:45,  3.62s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|                                                                             | 12/19440 [00:46<19:13:10,  3.56s/it]
+  0%|                                                                             | 12/19440 [00:46<19:13:10,  3.56s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|                                                                             | 13/19440 [00:50<19:38:49,  3.64s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 7.3591, 'learning_rate': 1.2638159383181786e-05, 'epoch': 0.0}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 7.3563, 'learning_rate': 1.4443610723636328e-05, 'epoch': 0.0}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|                                                                             | 14/19440 [00:53<19:16:02,  3.57s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 6.6434, 'learning_rate': 1.6249062064090866e-05, 'epoch': 0.0}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|                                                                             | 15/19440 [00:56<18:50:47,  3.49s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|                                                                             | 16/19440 [01:00<18:32:31,  3.44s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 6.4582, 'learning_rate': 1.805451340454541e-05, 'epoch': 0.0}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 6.016, 'learning_rate': 1.985996474499995e-05, 'epoch': 0.0}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|                                                                             | 17/19440 [01:03<18:15:27,  3.38s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 6.2345, 'learning_rate': 2.1665416085454492e-05, 'epoch': 0.0}
+  0%|                                                                             | 18/19440 [01:06<18:00:39,  3.34s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 5.8016, 'learning_rate': 2.347086742590903e-05, 'epoch': 0.0}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|                                                                             | 19/19440 [01:09<17:47:14,  3.30s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 5.7189, 'learning_rate': 2.5276318766363572e-05, 'epoch': 0.0}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|                                                                             | 20/19440 [01:12<17:36:39,  3.26s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 5.262, 'learning_rate': 2.7081770106818112e-05, 'epoch': 0.0}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|                                                                             | 21/19440 [01:16<17:21:20,  3.22s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 5.4374, 'learning_rate': 2.8887221447272655e-05, 'epoch': 0.0}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|                                                                             | 22/19440 [01:19<17:08:25,  3.18s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|                                                                             | 23/19440 [01:22<16:57:46,  3.14s/it]
+  0%|                                                                             | 23/19440 [01:22<16:57:46,  3.14s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 5.0583, 'learning_rate': 3.249812412818173e-05, 'epoch': 0.0}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|                                                                             | 24/19440 [01:25<16:47:14,  3.11s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 5.3147, 'learning_rate': 3.430357546863628e-05, 'epoch': 0.0}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|                                                                             | 25/19440 [01:28<17:15:22,  3.20s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.6202, 'learning_rate': 3.610902680909082e-05, 'epoch': 0.0}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|                                                                             | 26/19440 [01:31<16:57:58,  3.15s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 5.1504, 'learning_rate': 3.791447814954536e-05, 'epoch': 0.0}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|                                                                             | 27/19440 [01:34<16:44:43,  3.11s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.7351, 'learning_rate': 3.97199294899999e-05, 'epoch': 0.0}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|                                                                             | 28/19440 [01:37<16:33:18,  3.07s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.451, 'learning_rate': 4.152538083045444e-05, 'epoch': 0.0}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|                                                                             | 29/19440 [01:40<16:21:30,  3.03s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 4.8307, 'learning_rate': 4.3330832170908985e-05, 'epoch': 0.0}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|                                                                             | 30/19440 [01:43<16:08:22,  2.99s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|                                                                             | 31/19440 [01:46<15:59:29,  2.97s/it]
+  0%|                                                                             | 31/19440 [01:46<15:59:29,  2.97s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 4.5322, 'learning_rate': 4.694173485181806e-05, 'epoch': 0.0}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▏                                                                            | 32/19440 [01:49<15:51:02,  2.94s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▏                                                                            | 33/19440 [01:52<15:38:41,  2.90s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 4.4712, 'learning_rate': 4.8747186192272604e-05, 'epoch': 0.01}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 4.7247, 'learning_rate': 5.0552637532727144e-05, 'epoch': 0.01}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▏                                                                            | 34/19440 [01:54<15:30:30,  2.88s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.6641, 'learning_rate': 5.235808887318169e-05, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▏                                                                            | 35/19440 [01:57<15:24:34,  2.86s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▏                                                                            | 36/19440 [02:00<15:18:02,  2.84s/it]
+  0%|▏                                                                            | 36/19440 [02:00<15:18:02,  2.84s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 4.658, 'learning_rate': 5.5968991554090764e-05, 'epoch': 0.01}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▏                                                                            | 37/19440 [02:03<15:12:14,  2.82s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▏                                                                            | 38/19440 [02:06<15:44:21,  2.92s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.3339, 'learning_rate': 5.777444289454531e-05, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.3203, 'learning_rate': 5.957989423499985e-05, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▏                                                                            | 39/19440 [02:09<15:29:07,  2.87s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.1477, 'learning_rate': 6.13853455754544e-05, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|��                                                                            | 40/19440 [02:11<15:15:01,  2.83s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▏                                                                            | 41/19440 [02:14<15:00:06,  2.78s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.0639, 'learning_rate': 6.319079691590894e-05, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.1644, 'learning_rate': 6.499624825636346e-05, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▏                                                                            | 42/19440 [02:17<14:44:30,  2.74s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.1302, 'learning_rate': 6.6801699596818e-05, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▏                                                                            | 43/19440 [02:19<14:35:32,  2.71s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▏                                                                            | 44/19440 [02:22<14:21:30,  2.67s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 4.1862, 'learning_rate': 6.860715093727256e-05, 'epoch': 0.01}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 3.6454, 'learning_rate': 7.04126022777271e-05, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▏                                                                            | 45/19440 [02:25<14:10:14,  2.63s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 3.8834, 'learning_rate': 7.221805361818164e-05, 'epoch': 0.01}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▏                                                                            | 46/19440 [02:27<13:55:36,  2.59s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 3.7222, 'learning_rate': 7.402350495863618e-05, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▏                                                                            | 47/19440 [02:30<13:45:08,  2.55s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▏                                                                            | 48/19440 [02:32<13:31:07,  2.51s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 3.3332, 'learning_rate': 7.582895629909072e-05, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▏                                                                            | 49/19440 [02:34<13:18:52,  2.47s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 3.0502, 'learning_rate': 7.763440763954526e-05, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 3.1309, 'learning_rate': 7.94398589799998e-05, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▏                                                                            | 50/19440 [02:37<13:41:24,  2.54s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 6.6939, 'learning_rate': 8.124531032045434e-05, 'epoch': 0.01}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▏                                                                            | 51/19440 [02:42<16:59:16,  3.15s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 5.999, 'learning_rate': 8.305076166090888e-05, 'epoch': 0.01}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▏                                                                            | 52/19440 [02:46<18:38:01,  3.46s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 5.4437, 'learning_rate': 8.485621300136342e-05, 'epoch': 0.01}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▏                                                                            | 53/19440 [02:50<19:28:22,  3.62s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 5.661, 'learning_rate': 8.666166434181797e-05, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▏                                                                            | 54/19440 [02:54<19:49:32,  3.68s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 5.5403, 'learning_rate': 8.846711568227251e-05, 'epoch': 0.01}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▏                                                                            | 55/19440 [02:57<19:54:14,  3.70s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 5.1378, 'learning_rate': 9.027256702272705e-05, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▏                                                                            | 56/19440 [03:01<19:57:09,  3.71s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 5.1641, 'learning_rate': 9.207801836318158e-05, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▏                                                                            | 57/19440 [03:05<20:03:01,  3.72s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▏                                                                            | 58/19440 [03:08<19:56:04,  3.70s/it]
+  0%|▏                                                                            | 58/19440 [03:08<19:56:04,  3.70s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.9691, 'learning_rate': 9.568892104409067e-05, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▏                                                                            | 59/19440 [03:12<19:43:02,  3.66s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 4.9514, 'learning_rate': 9.749437238454521e-05, 'epoch': 0.01}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▏                                                                            | 60/19440 [03:16<19:27:23,  3.61s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 5.1516, 'learning_rate': 9.929982372499975e-05, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▏                                                                            | 61/19440 [03:19<19:13:38,  3.57s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▏                                                                            | 62/19440 [03:22<18:59:12,  3.53s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 5.1644, 'learning_rate': 0.00010110527506545429, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▏                                                                            | 63/19440 [03:26<19:25:01,  3.61s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 4.6898, 'learning_rate': 0.00010291072640590883, 'epoch': 0.01}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 4.8632, 'learning_rate': 0.00010471617774636338, 'epoch': 0.01}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▎                                                                            | 64/19440 [03:30<19:04:42,  3.54s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▎                                                                            | 65/19440 [03:33<18:39:57,  3.47s/it]
+  0%|▎                                                                            | 65/19440 [03:33<18:39:57,  3.47s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▎                                                                            | 66/19440 [03:36<18:25:30,  3.42s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.6448, 'learning_rate': 0.00010832708042727245, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.7384, 'learning_rate': 0.00011013253176772699, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▎                                                                            | 67/19440 [03:40<18:12:00,  3.38s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▎                                                                            | 68/19440 [03:43<17:55:33,  3.33s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.7486, 'learning_rate': 0.00011193798310818153, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▎                                                                            | 69/19440 [03:46<17:58:41,  3.34s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.788, 'learning_rate': 0.00011374343444863608, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 4.8301, 'learning_rate': 0.00011554888578909062, 'epoch': 0.01}
+  0%|▎                                                                            | 70/19440 [03:49<17:40:50,  3.29s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▎                                                                            | 71/19440 [03:52<17:29:12,  3.25s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 4.7484, 'learning_rate': 0.00011735433712954516, 'epoch': 0.01}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.857, 'learning_rate': 0.0001191597884699997, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▎                                                                            | 72/19440 [03:56<17:18:42,  3.22s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▎                                                                            | 73/19440 [03:59<17:04:50,  3.18s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.6385, 'learning_rate': 0.00012096523981045424, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 4.6694, 'learning_rate': 0.0001227706911509088, 'epoch': 0.01}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▎                                                                            | 74/19440 [04:02<16:54:48,  3.14s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▎                                                                            | 75/19440 [04:05<17:23:51,  3.23s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.2003, 'learning_rate': 0.00012457614249136332, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.6907, 'learning_rate': 0.00012638159383181787, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▎                                                                            | 76/19440 [04:08<17:08:38,  3.19s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▎                                                                            | 77/19440 [04:11<16:52:02,  3.14s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.676, 'learning_rate': 0.0001281870451722724, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 4.5547, 'learning_rate': 0.00012999249651272693, 'epoch': 0.01}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▎                                                                            | 78/19440 [04:14<16:37:07,  3.09s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▎                                                                            | 79/19440 [04:17<16:28:17,  3.06s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.4641, 'learning_rate': 0.00013179794785318148, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.5819, 'learning_rate': 0.000133603399193636, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▎                                                                            | 80/19440 [04:20<16:15:48,  3.02s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 4.3248, 'learning_rate': 0.00013540885053409056, 'epoch': 0.01}
+  0%|▎                                                                            | 81/19440 [04:23<16:05:54,  2.99s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.3149, 'learning_rate': 0.00013721430187454511, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▎                                                                            | 82/19440 [04:26<16:00:31,  2.98s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▎                                                                            | 83/19440 [04:29<15:52:18,  2.95s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 4.3363, 'learning_rate': 0.00013901975321499964, 'epoch': 0.01}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.4412, 'learning_rate': 0.0001408252045554542, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▎                                                                            | 84/19440 [04:32<15:46:38,  2.93s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▎                                                                            | 85/19440 [04:35<15:38:53,  2.91s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.661, 'learning_rate': 0.00014263065589590872, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.3019, 'learning_rate': 0.00014443610723636327, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▎                                                                            | 86/19440 [04:37<15:31:19,  2.89s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▎                                                                            | 87/19440 [04:40<15:21:47,  2.86s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.1839, 'learning_rate': 0.00014624155857681783, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.2393, 'learning_rate': 0.00014804700991727235, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▎                                                                            | 88/19440 [04:43<15:50:39,  2.95s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▎                                                                            | 89/19440 [04:46<15:37:56,  2.91s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 4.3483, 'learning_rate': 0.0001498524612577269, 'epoch': 0.01}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▎                                                                            | 90/19440 [04:49<15:19:14,  2.85s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.2097, 'learning_rate': 0.00015165791259818143, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 3.9999, 'learning_rate': 0.000153463363938636, 'epoch': 0.01}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▎                                                                            | 91/19440 [04:52<15:06:21,  2.81s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▎                                                                            | 92/19440 [04:54<14:54:32,  2.77s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 3.8905, 'learning_rate': 0.0001552688152790905, 'epoch': 0.01}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▎                                                                            | 93/19440 [04:57<14:39:59,  2.73s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 3.9887, 'learning_rate': 0.00015707426661954504, 'epoch': 0.01}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.229, 'learning_rate': 0.0001588797179599996, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▎                                                                            | 94/19440 [05:00<14:27:08,  2.69s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.4799, 'learning_rate': 0.00016068516930045412, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▍                                                                            | 95/19440 [05:02<14:14:19,  2.65s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▍                                                                            | 96/19440 [05:05<14:00:03,  2.61s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 3.9251, 'learning_rate': 0.00016249062064090867, 'epoch': 0.01}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  0%|▍                                                                            | 97/19440 [05:07<13:45:56,  2.56s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 3.8613, 'learning_rate': 0.00016429607198136323, 'epoch': 0.01}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 3.7416, 'learning_rate': 0.00016610152332181775, 'epoch': 0.02}
+  1%|▍                                                                            | 98/19440 [05:10<13:33:44,  2.52s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 3.5651, 'learning_rate': 0.0001679069746622723, 'epoch': 0.02}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▍                                                                            | 99/19440 [05:12<13:16:07,  2.47s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▍                                                                           | 100/19440 [05:15<13:32:59,  2.52s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 3.2072, 'learning_rate': 0.00016971242600272683, 'epoch': 0.02}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▍                                                                           | 101/19440 [05:19<17:00:07,  3.16s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 6.576, 'learning_rate': 0.00017151787734318139, 'epoch': 0.02}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▍                                                                           | 102/19440 [05:23<18:34:52,  3.46s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 5.8767, 'learning_rate': 0.00017332332868363594, 'epoch': 0.02}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▍                                                                           | 103/19440 [05:27<19:18:31,  3.59s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 5.463, 'learning_rate': 0.00017512878002409047, 'epoch': 0.02}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▍                                                                           | 104/19440 [05:31<19:42:12,  3.67s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 5.3542, 'learning_rate': 0.00017693423136454502, 'epoch': 0.02}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▍                                                                           | 105/19440 [05:35<19:50:47,  3.70s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.9525, 'learning_rate': 0.00017873968270499955, 'epoch': 0.02}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 5.2095, 'learning_rate': 0.0001805451340454541, 'epoch': 0.02}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▍                                                                           | 106/19440 [05:39<19:58:21,  3.72s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 4.8454, 'learning_rate': 0.00018235058538590865, 'epoch': 0.02}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▍                                                                           | 107/19440 [05:42<20:03:52,  3.74s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▍                                                                           | 108/19440 [05:46<19:48:23,  3.69s/it]
+  1%|▍                                                                           | 108/19440 [05:46<19:48:23,  3.69s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▍                                                                           | 109/19440 [05:50<19:33:01,  3.64s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 4.571, 'learning_rate': 0.0001859614880668177, 'epoch': 0.02}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▍                                                                           | 110/19440 [05:53<19:20:36,  3.60s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.931, 'learning_rate': 0.00018776693940727223, 'epoch': 0.02}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.7981, 'learning_rate': 0.00018957239074772678, 'epoch': 0.02}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▍                                                                           | 111/19440 [05:56<19:06:58,  3.56s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▍                                                                           | 112/19440 [06:00<18:51:06,  3.51s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.7499, 'learning_rate': 0.00019137784208818134, 'epoch': 0.02}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▍                                                                           | 113/19440 [06:04<19:18:58,  3.60s/it]
+  1%|▍                                                                           | 113/19440 [06:04<19:18:58,  3.60s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▍                                                                           | 114/19440 [06:07<18:59:15,  3.54s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.9919, 'learning_rate': 0.00019498874476909042, 'epoch': 0.02}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 4.7554, 'learning_rate': 0.00019679419610954494, 'epoch': 0.02}
+  1%|▍                                                                           | 115/19440 [06:10<18:36:10,  3.47s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▍                                                                           | 116/19440 [06:14<18:18:32,  3.41s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.3994, 'learning_rate': 0.0001985996474499995, 'epoch': 0.02}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 4.526, 'learning_rate': 0.00020040509879045405, 'epoch': 0.02}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▍                                                                           | 117/19440 [06:17<18:05:21,  3.37s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▍                                                                           | 118/19440 [06:20<17:48:08,  3.32s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 4.3614, 'learning_rate': 0.00020221055013090858, 'epoch': 0.02}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▍                                                                           | 119/19440 [06:23<17:35:01,  3.28s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 4.8438, 'learning_rate': 0.00020401600147136313, 'epoch': 0.02}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▍                                                                           | 120/19440 [06:26<17:24:09,  3.24s/it]
+  1%|▍                                                                           | 120/19440 [06:26<17:24:09,  3.24s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▍                                                                           | 121/19440 [06:30<17:15:30,  3.22s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.7043, 'learning_rate': 0.0002076269041522722, 'epoch': 0.02}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.8447, 'learning_rate': 0.00020943235549272676, 'epoch': 0.02}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▍                                                                           | 122/19440 [06:33<17:22:49,  3.24s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▍                                                                           | 123/19440 [06:36<17:09:50,  3.20s/it]
+  1%|▍                                                                           | 123/19440 [06:36<17:09:50,  3.20s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.7529, 'learning_rate': 0.00021304325817363582, 'epoch': 0.02}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▍                                                                           | 124/19440 [06:39<16:54:54,  3.15s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.6383, 'learning_rate': 0.00021484870951409034, 'epoch': 0.02}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▍                                                                           | 125/19440 [06:43<17:22:14,  3.24s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▍                                                                           | 126/19440 [06:46<17:06:02,  3.19s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.7783, 'learning_rate': 0.0002166541608545449, 'epoch': 0.02}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.4605, 'learning_rate': 0.00021845961219499945, 'epoch': 0.02}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▍                                                                           | 127/19440 [06:49<16:43:18,  3.12s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▌                                                                           | 128/19440 [06:51<16:27:28,  3.07s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 4.4511, 'learning_rate': 0.00022026506353545398, 'epoch': 0.02}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.5805, 'learning_rate': 0.00022207051487590853, 'epoch': 0.02}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▌                                                                           | 129/19440 [06:54<16:14:02,  3.03s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▌                                                                           | 130/19440 [06:57<16:05:33,  3.00s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.5971, 'learning_rate': 0.00022387596621636306, 'epoch': 0.02}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▌                                                                           | 131/19440 [07:00<15:52:31,  2.96s/it]
+  1%|▌                                                                           | 131/19440 [07:00<15:52:31,  2.96s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 4.4013, 'learning_rate': 0.00022748686889727216, 'epoch': 0.02}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▌                                                                           | 132/19440 [07:03<15:42:26,  2.93s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▌                                                                           | 133/19440 [07:06<15:34:20,  2.90s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 4.7983, 'learning_rate': 0.0002292923202377267, 'epoch': 0.02}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 4.3568, 'learning_rate': 0.00023109777157818124, 'epoch': 0.02}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▌                                                                           | 134/19440 [07:09<15:24:25,  2.87s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▌                                                                           | 135/19440 [07:12<15:21:37,  2.86s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.3551, 'learning_rate': 0.00023290322291863577, 'epoch': 0.02}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▌                                                                           | 136/19440 [07:14<15:14:36,  2.84s/it]
+  1%|▌                                                                           | 136/19440 [07:14<15:14:36,  2.84s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 4.6872, 'learning_rate': 0.00023651412559954488, 'epoch': 0.02}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▌                                                                           | 137/19440 [07:17<15:09:08,  2.83s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▌                                                                           | 138/19440 [07:20<15:39:28,  2.92s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.4209, 'learning_rate': 0.0002383195769399994, 'epoch': 0.02}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.5009, 'learning_rate': 0.00024012502828045396, 'epoch': 0.02}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▌                                                                           | 139/19440 [07:23<15:27:43,  2.88s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▌                                                                           | 140/19440 [07:26<15:11:38,  2.83s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 3.8601, 'learning_rate': 0.00024193047962090848, 'epoch': 0.02}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▌                                                                           | 141/19440 [07:28<14:57:29,  2.79s/it]
+  1%|▌                                                                           | 141/19440 [07:28<14:57:29,  2.79s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.2856, 'learning_rate': 0.0002455413823018176, 'epoch': 0.02}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▌                                                                           | 142/19440 [07:31<14:43:41,  2.75s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▌                                                                           | 143/19440 [07:34<14:29:19,  2.70s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.1366, 'learning_rate': 0.00024734683364227214, 'epoch': 0.02}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▌                                                                           | 144/19440 [07:36<14:15:47,  2.66s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 3.9762, 'learning_rate': 0.00024915228498272664, 'epoch': 0.02}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.3886, 'learning_rate': 0.0002509577363231812, 'epoch': 0.02}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▌                                                                           | 145/19440 [07:39<14:02:30,  2.62s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 4.5055, 'learning_rate': 0.00025276318766363575, 'epoch': 0.02}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▌                                                                           | 146/19440 [07:41<13:53:31,  2.59s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▌                                                                           | 147/19440 [07:44<13:41:48,  2.56s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 3.9612, 'learning_rate': 0.00025456863900409025, 'epoch': 0.02}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▌                                                                           | 148/19440 [07:46<13:29:21,  2.52s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.1039, 'learning_rate': 0.0002563740903445448, 'epoch': 0.02}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▌                                                                           | 149/19440 [07:49<13:17:47,  2.48s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 3.8241, 'learning_rate': 0.0002581795416849993, 'epoch': 0.02}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 3.989, 'learning_rate': 0.00025998499302545385, 'epoch': 0.02}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▌                                                                           | 150/19440 [07:51<13:37:18,  2.54s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▌                                                                           | 151/19440 [07:56<16:55:30,  3.16s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 6.3059, 'learning_rate': 0.0002617904443659084, 'epoch': 0.02}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▌                                                                           | 152/19440 [08:00<18:29:23,  3.45s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 6.203, 'learning_rate': 0.00026359589570636296, 'epoch': 0.02}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▌                                                                           | 153/19440 [08:04<19:13:15,  3.59s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 5.9501, 'learning_rate': 0.0002654013470468175, 'epoch': 0.02}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▌                                                                           | 154/19440 [08:08<19:41:22,  3.68s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 5.5182, 'learning_rate': 0.000267206798387272, 'epoch': 0.02}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 5.8331, 'learning_rate': 0.00026901224972772657, 'epoch': 0.02}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▌                                                                           | 155/19440 [08:12<19:55:25,  3.72s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 5.2629, 'learning_rate': 0.0002708177010681811, 'epoch': 0.02}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▌                                                                           | 156/19440 [08:15<20:03:25,  3.74s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▌                                                                           | 157/19440 [08:19<19:54:07,  3.72s/it]
+  1%|▌                                                                           | 157/19440 [08:19<19:54:07,  3.72s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▌                                                                           | 158/19440 [08:23<19:47:23,  3.69s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 5.2855, 'learning_rate': 0.00027442860374909023, 'epoch': 0.02}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▌                                                                           | 159/19440 [08:26<19:39:10,  3.67s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 5.1616, 'learning_rate': 0.00027623405508954473, 'epoch': 0.02}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▋                                                                           | 160/19440 [08:30<19:29:00,  3.64s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 5.4447, 'learning_rate': 0.0002780395064299993, 'epoch': 0.02}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 5.435, 'learning_rate': 0.00027984495777045383, 'epoch': 0.02}
+  1%|▋                                                                           | 161/19440 [08:33<19:13:27,  3.59s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▋                                                                           | 162/19440 [08:37<18:59:37,  3.55s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 5.4416, 'learning_rate': 0.0002816504091109084, 'epoch': 0.02}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▋                                                                           | 163/19440 [08:41<19:35:56,  3.66s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 5.1677, 'learning_rate': 0.00028345586045136294, 'epoch': 0.03}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▋                                                                           | 164/19440 [08:44<19:21:47,  3.62s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.9068, 'learning_rate': 0.00028526131179181744, 'epoch': 0.03}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▋                                                                           | 165/19440 [08:48<19:22:34,  3.62s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 5.0924, 'learning_rate': 0.000287066763132272, 'epoch': 0.03}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+{'loss': 4.9907, 'learning_rate': 0.00028887221447272655, 'epoch': 0.03}
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▋                                                                           | 166/19440 [08:52<19:19:25,  3.61s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▋                                                                           | 167/19440 [08:55<19:10:30,  3.58s/it]
+  1%|▋                                                                           | 167/19440 [08:55<19:10:30,  3.58s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▋                                                                           | 168/19440 [08:59<18:59:36,  3.55s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 4.8957, 'learning_rate': 0.00029248311715363565, 'epoch': 0.03}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▋                                                                           | 169/19440 [09:02<18:57:15,  3.54s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 5.0207, 'learning_rate': 0.00029428856849409015, 'epoch': 0.03}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 5.2025, 'learning_rate': 0.0002960940198345447, 'epoch': 0.03}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▋                                                                           | 170/19440 [09:05<18:39:10,  3.48s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▋                                                                           | 171/19440 [09:09<18:21:38,  3.43s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 5.3462, 'learning_rate': 0.00029789947117499926, 'epoch': 0.03}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+{'loss': 6.3801, 'learning_rate': 0.0002997049225154538, 'epoch': 0.03}
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▋                                                                           | 172/19440 [09:12<17:45:40,  3.32s/it]`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+`use_cache=True` is incompatible with gradient checkpointing. Setting `use_cache=False`...
+Could not estimate the number of tokens of the input, floating-point operations will not be computed
+  1%|▋                                                                           | 173/19440 [09:15<17:14:47,  3.22s/it]Traceback (most recent call last):
+  File "/home/sanchit_huggingface_co/xtreme_s_xlsr_2_bart_covost2_fr_en_2/run_xtreme_s.py", line 947, in <module>
+    main()
+  File "/home/sanchit_huggingface_co/xtreme_s_xlsr_2_bart_covost2_fr_en_2/run_xtreme_s.py", line 874, in main
+    train_result = trainer.train(resume_from_checkpoint=checkpoint)
+  File "/home/sanchit_huggingface_co/transformers/src/transformers/trainer.py", line 1524, in train
+    self._maybe_log_save_evaluate(tr_loss, model, trial, epoch, ignore_keys_for_eval)
+  File "/home/sanchit_huggingface_co/transformers/src/transformers/trainer.py", line 1647, in _maybe_log_save_evaluate
+    self.log(logs)
+  File "/home/sanchit_huggingface_co/transformers/src/transformers/trainer.py", line 1960, in log
+    self.control = self.callback_handler.on_log(self.args, self.state, self.control, logs)
+  File "/home/sanchit_huggingface_co/transformers/src/transformers/trainer_callback.py", line 381, in on_log
+    return self.call_event("on_log", args, state, control, logs=logs)
+  File "/home/sanchit_huggingface_co/transformers/src/transformers/trainer_callback.py", line 388, in call_event
+    result = getattr(callback, event)(
+  File "/home/sanchit_huggingface_co/transformers/src/transformers/integrations.py", line 658, in on_log
+    self._wandb.log({**logs, "train/global_step": state.global_step})
+  File "/home/sanchit_huggingface_co/gcp/lib/python3.9/site-packages/wandb/sdk/wandb_run.py", line 1349, in log
+    self.history._row_add(data)
+  File "/home/sanchit_huggingface_co/gcp/lib/python3.9/site-packages/wandb/sdk/wandb_history.py", line 44, in _row_add
+    self._flush()
+  File "/home/sanchit_huggingface_co/gcp/lib/python3.9/site-packages/wandb/sdk/wandb_history.py", line 59, in _flush
+    self._callback(row=self._data, step=self._step)
+  File "/home/sanchit_huggingface_co/gcp/lib/python3.9/site-packages/wandb/sdk/wandb_run.py", line 1027, in _history_callback
+    self._backend.interface.publish_history(
+  File "/home/sanchit_huggingface_co/gcp/lib/python3.9/site-packages/wandb/sdk/interface/interface.py", line 506, in publish_history
+    self._publish_history(history)
+  File "/home/sanchit_huggingface_co/gcp/lib/python3.9/site-packages/wandb/sdk/interface/interface_shared.py", line 59, in _publish_history
+    self._publish(rec)
+  File "/home/sanchit_huggingface_co/gcp/lib/python3.9/site-packages/wandb/sdk/interface/interface_queue.py", line 49, in _publish
+    raise Exception("The wandb backend process has shutdown")
+Exception: The wandb backend process has shutdown
\ No newline at end of file