- 30 May, 2019 1 commit
-
-
Victor SANH authored
-
- 05 May, 2019 1 commit
-
-
huntzhan authored
-
- 01 May, 2019 1 commit
-
-
Ben Mann authored
-
- 27 Apr, 2019 1 commit
-
-
Ailing Zhang authored
-
- 25 Apr, 2019 2 commits
-
-
thomwolf authored
-
lukovnikov authored
- added some images for illustration - updated comments in optimization
-
- 24 Apr, 2019 1 commit
-
-
thomwolf authored
-
- 21 Apr, 2019 2 commits
-
-
lukovnikov authored
-
lukovnikov authored
- removed unused plotting code - using ABC for LRSchedule - added some schedule object init tests
-
- 17 Apr, 2019 8 commits
- 16 Apr, 2019 3 commits
-
-
Abhi Sharma authored
This fix is in reference to issue #382. GPT2 can now be trained in mixed precision, which I've confirmed with testing. I also tested unconditional generation on multiple seeds before and after changing 1e10 to 1e4 and there was no difference. Please let me know if there is anything else I can do to make this pull request better. Thanks for all your work!
-
thomwolf authored
-
thomwolf authored
-
- 15 Apr, 2019 10 commits
- 12 Apr, 2019 2 commits
-
-
Martin Boyanov authored
-
thomwolf authored
-
- 11 Apr, 2019 5 commits
- 03 Apr, 2019 3 commits
-
-
lukovnikov authored
-
lukovnikov authored
-
lukovnikov authored
-
