- 18 Mar, 2019 2 commits
-
-
lukovnikov authored
-
lukovnikov authored
-
- 13 Mar, 2019 1 commit
-
-
lukovnikov authored
-
- 12 Mar, 2019 4 commits
-
-
lukovnikov authored
-
lukovnikov authored
-
lukovnikov authored
-
lukovnikov authored
-
- 09 Mar, 2019 3 commits
-
-
lukovnikov authored
-
lukovnikov authored
-
lukovnikov authored
-
- 08 Mar, 2019 1 commit
-
-
lukovnikov authored
-
- 06 Mar, 2019 11 commits
-
-
thomwolf authored
-
thomwolf authored
-
thomwolf authored
-
thomwolf authored
-
thomwolf authored
-
Thomas Wolf authored
Processor for SST-2 task
-
Thomas Wolf authored
catch exception if pathlib not install
-
Thomas Wolf authored
Fix top k generation for k != 0
-
Thomas Wolf authored
Allow tokenization of sequences > 512 for caching
-
Thomas Wolf authored
Issue#324: warmup linear fixes
-
Thomas Wolf authored
add BertTokenizer flag to skip basic tokenization
-
- 05 Mar, 2019 3 commits
-
-
Catalin Voss authored
-
Catalin Voss authored
-
John Lehmann authored
-
- 04 Mar, 2019 1 commit
-
-
Aaron Mangum authored
-
- 03 Mar, 2019 2 commits
-
-
Catalin Voss authored
-
Catalin Voss authored
For many applications requiring randomized data access, it's easier to cache the tokenized representations than the words. So why not turn this into a warning?
-
- 28 Feb, 2019 1 commit
-
-
Catalin Voss authored
-
- 27 Feb, 2019 9 commits
-
-
John Hewitt authored
-
lukovnikov authored
-
lukovnikov authored
-
lukovnikov authored
-
lukovnikov authored
-
lukovnikov authored
-
lukovnikov authored
-
lukovnikov authored
-
John Hewitt authored
-
- 26 Feb, 2019 2 commits
-
-
lukovnikov authored
fix for negative learning rate with warmup_linear in BertAdam (happens when t_total is specified incorrectly) + copied BERT optimization warmup functions to OpenAI optimization file + added comments
-
lukovnikov authored
fix for negative learning rate with warmup_linear in BertAdam (happens when t_total is specified incorrectly) + copied BERT optimization warmup functions to OpenAI optimization file + added comments
-
