Register
Login
Resources
Docs Blog Datasets Glossary Case Studies Tutorials & Webinars
Product
Data Engine LLMs Platform Enterprise
Pricing Explore
Connect to our Discord channel
Commit History
Message Author SHA1 Date
allow specifying max_tokens for generation   Alexei Baevski 6 years ago
Save and restore wall time in checkpoints   Myle Ott 6 years ago
Simplify train.py (merge with singleprocess_train.py)   Myle Ott 6 years ago
Refactor incremental generation to be more explicit and less magical (#222)   Myle Ott 6 years ago
More unit test fixes   Myle Ott 6 years ago
Fix tests and flake8   Myle Ott 6 years ago
fairseq-py goes distributed (#106)   Myle Ott 6 years ago
Fix max_positions calculation in train.py   Myle Ott 6 years ago
Report log likelihood for label smoothing   Sergey Edunov 6 years ago
Add --max-sentences-valid to train.py   Myle Ott 6 years ago
Streamline data formatting utils   Myle Ott 6 years ago
Output number of model parameters in train.py   Myle Ott 6 years ago
Save number of GPUs in args (and checkpoints)   Myle Ott 6 years ago
Fixed 2 typos (#75)   toothlessdragon 6 years ago
Fallback to `--log-format=simple` for non-TTY terminals   Myle Ott 6 years ago
Fixes for `--log-format`   Myle Ott 6 years ago
Fix max_positions_valid in train.py   Myle Ott 6 years ago
Add `--log-format` option and JSON logger   Myle Ott 6 years ago
Replace unk with original string   Louis Martin 6 years ago
Loop over evaluation dataloader in descending order   Myle Ott 6 years ago
Add --max-sentence option for batching based on # sentences   Myle Ott 6 years ago
Improvements to data loader   Myle Ott 6 years ago
Fix seed so that data is properly shuffled between epochs   Myle Ott 6 years ago
Support different max_source_positions and max_target_positions   Myle Ott 6 years ago
Add `--curriculum` option   Myle Ott 6 years ago
Set seed after each epoch to improve consistency when resuming   Myle Ott 6 years ago
Prevent math overflow when loss is too high   Louis Martin 6 years ago
Simplify deps of build_model to only depend on dict (instead of dataset)   Myle Ott 6 years ago
Refactor model saving/loading to be more reusable   Myle Ott 6 years ago
Better logging from criterions   Myle Ott 6 years ago