0%| | 0/5000 [00:00> `use_cache = True` is incompatible with gradient checkpointing. Setting `use_cache = False`... 0%|▍ | 25/5000 [03:38<11:34:32, 8.38s/it] 1%|▊ | 49/5000 [06:58<11:27:58, 8.34s/it] 1%|█▏ | 74/5000 [10:27<11:22:31, 8.31s/it] 2%|█▌ | 99/5000 [13:55<11:20:20, 8.33s/it] 2%|█▉ | 124/5000 [17:24<11:17:05, 8.33s/it] 3%|██▏ | 140/5000 [19:37<11:16:28, 8.35s/it]Traceback (most recent call last): File "run_speech_recognition_seq2seq.py", line 627, in main() File "run_speech_recognition_seq2seq.py", line 577, in main train_result = trainer.train(resume_from_checkpoint=checkpoint) File "/home/sanchit/transformers/src/transformers/trainer.py", line 1774, in train return inner_training_loop( File "/home/sanchit/transformers/src/transformers/trainer.py", line 2088, in _inner_training_loop for step, inputs in enumerate(epoch_iterator): File "/home/sanchit/hf/lib/python3.8/site-packages/accelerate/data_loader.py", line 462, in __iter__ next_batch = next(dataloader_iter) File "/home/sanchit/hf/lib/python3.8/site-packages/torch/utils/data/dataloader.py", line 631, in __next__ data = self._next_data() File "/home/sanchit/hf/lib/python3.8/site-packages/torch/utils/data/dataloader.py", line 675, in _next_data data = self._dataset_fetcher.fetch(index) # may raise StopIteration KeyboardInterrupt