|
allow specifying max_tokens for generation
|
Alexei Baevski
|
|
6 years ago |
|
Support integer learning rates
|
Myle Ott
|
|
6 years ago |
|
implement batching in interactive mode
|
Alexei Baevski
|
|
6 years ago |
|
Fix --prefix-size
|
Myle Ott
|
|
6 years ago |
|
Use eval() to parse args.lr
|
Myle Ott
|
|
6 years ago |
|
No more magical --fp16
|
Myle Ott
|
|
6 years ago |
|
Add FP16 support
|
Myle Ott
|
|
6 years ago |
|
Allow schedule for update-freq
|
Myle Ott
|
|
6 years ago |
|
Improve dataloader speed and deprecate concept of batch_offset (use --sample-without-replacement instead)
|
Myle Ott
|
|
6 years ago |
|
Simulated big batches
|
Sergey Edunov
|
|
6 years ago |
|
Merge internal changes (#136)
|
Myle Ott
|
|
6 years ago |
|
Add support to prefixes (#221)
|
Dario Pavllo
|
|
6 years ago |
|
More unit test fixes
|
Myle Ott
|
|
6 years ago |
|
fairseq-py goes distributed (#106)
|
Myle Ott
|
|
6 years ago |
|
Share input/output embed
|
Sergey Edunov
|
|
6 years ago |
|
Allow --lr to specify a fixed learning rate schedule
|
Myle Ott
|
|
6 years ago |
|
Fallback to `--log-format=simple` for non-TTY terminals
|
Myle Ott
|
|
6 years ago |
|
Add `--log-format` option and JSON logger
|
Myle Ott
|
|
6 years ago |
|
Add LSTM
|
Myle Ott
|
|
6 years ago |
|
Add --max-sentence option for batching based on # sentences
|
Myle Ott
|
|
6 years ago |
|
Refactor generation
|
Louis Martin
|
|
6 years ago |
|
Support different max_source_positions and max_target_positions
|
Myle Ott
|
|
6 years ago |
|
Added -unkpen flag to generate.py following logic of Lua/Torch version
|
Michael Auli
|
|
6 years ago |
|
Add `--curriculum` option
|
Myle Ott
|
|
6 years ago |
|
Fix description for `--sample-without-replacement` option
|
Myle Ott
|
|
6 years ago |
|
Add support for additional optimizers
|
Myle Ott
|
|
6 years ago |
|
Allow --max-len-a to be a float
|
Myle Ott
|
|
6 years ago |
|
Support configurable BPE symbol
|
Myle Ott
|
|
6 years ago |
|
Ignore invalid sentences in test and valid
|
Sergey Edunov
|
|
6 years ago |
|
Don't generate during training, add --quiet to generate.py
|
Sergey Edunov
|
|
6 years ago |