- 14 Jun, 2019 5 commits
-
-
Thomas Wolf authored
Fixing issue "Training beyond specified 't_total' steps with schedule 'warmup_linear'" reported in #556
-
Thomas Wolf authored
GPT-2 (medium size model, special_tokens, fine-tuning, attention) + repo code coverage metric
-
Thomas Wolf authored
-
thomwolf authored
-
thomwolf authored
-
- 12 Jun, 2019 1 commit
-
-
Thomas Wolf authored
[hotfix] Fix frozen pooler parameters in SWAG example.
-
- 11 Jun, 2019 2 commits
-
-
Meet Pragnesh Shah authored
-
Thomas Wolf authored
apply Whole Word Masking technique
-
- 10 Jun, 2019 1 commit
-
-
jeonsworld authored
apply Whole Word Masking technique. referred to [create_pretraining_data.py](https://github.com/google-research/bert/blob/master/create_pretraining_data.py)
-
- 06 Jun, 2019 6 commits
-
-
VictorSanh authored
-
VictorSanh authored
-
VictorSanh authored
-
VictorSanh authored
-
VictorSanh authored
-
VictorSanh authored
-
- 03 Jun, 2019 3 commits
-
-
thomwolf authored
-
VictorSanh authored
This reverts commit de5e5682.
-
VictorSanh authored
-
- 31 May, 2019 8 commits
-
-
Thomas Wolf authored
Add GPT* compatibility to torchhub
-
VictorSanh authored
-
VictorSanh authored
-
VictorSanh authored
-
VictorSanh authored
-
VictorSanh authored
-
VictorSanh authored
-
VictorSanh authored
-
- 30 May, 2019 3 commits
-
-
VictorSanh authored
-
Victor SANH authored
-
VictorSanh authored
-
- 13 May, 2019 1 commit
-
-
samuelbroscheit authored
-
- 11 May, 2019 2 commits
-
-
samuel.broscheit authored
-
https://github.com/huggingface/pytorch-pretrained-BERT/issues/556samuel.broscheit authored
Reason for issue was that optimzation steps where computed from example size, which is different from actual size of dataloader when an example is chunked into multiple instances. Solution in this pull request is to compute num_optimization_steps directly from len(data_loader).
-
- 10 May, 2019 1 commit
-
-
Thomas Wolf authored
Updating learning rate with special warm up in examples
-
- 09 May, 2019 3 commits
-
-
burcturkoglu authored
-
-
burcturkoglu authored
-
- 08 May, 2019 4 commits
-
-
thomwolf authored
-
thomwolf authored
-
Thomas Wolf authored
Make the epsilon of LayerNorm configurable.
-
Thomas Wolf authored
move pytroch_pretrained_bert cache folder under same path as torch
-