microsoft / MASS

MASS: Masked Sequence to Sequence Pre-training for Language Generation
https://arxiv.org/pdf/1905.02450.pdf
Other
1.12k stars 206 forks source link

Fine-tuning Error #83

Open jimmyhsia opened 4 years ago

jimmyhsia commented 4 years ago

I am trying the script in Running , and it get error like this -- Process 3 terminated with the following error: Traceback (most recent call last): File "/home/jimmyhsia0522/.local/lib/python3.7/site-packages/torch/multiprocessing/spawn.py", line 19, in _wrap fn(i, *args) File "/home/jimmyhsia0522/.local/lib/python3.7/site-packages/fairseq_cli/train.py", line 284, in distributed_main main(args, init_distributed=True) File "/home/jimmyhsia0522/.local/lib/python3.7/site-packages/fairseq_cli/train.py", line 68, in main extra_state, epoch_itr = checkpoint_utils.load_checkpoint(args, trainer) File "/home/jimmyhsia0522/.local/lib/python3.7/site-packages/fairseq/checkpoint_utils.py", line 109, in load_checkpoint reset_meters=args.reset_meters, File "/home/jimmyhsia0522/.local/lib/python3.7/site-packages/fairseq/trainer.py", line 173, in load_checkpoint 'Criterion does not match; please reset the optimizer (--reset-optimizer).' AssertionError: Criterion does not match; please reset the optimizer (--reset-optimizer). And i check fairseq code in trainer.py error line and here is my optimizer criterion_name: MaskedLmLoss self.criterion.class.name: LabelSmoothedCrossEntropyCriterion i have no ideal anyone help? thank

StillKeepTry commented 4 years ago

Can you give me more details about your problem?