309 Commits

Author	SHA1	Message	Date
Myle Ott	6e4d370af9	More updates for PyTorch (#114)	2018-03-01 14:04:08 -05:00
Myle Ott	9438019ff0	Refactor incremental generation to be more explicit and less magical (#222)	2018-02-27 14:28:24 -08:00
Myle Ott	0d90e35f3b	More unit test fixes	2018-02-27 14:28:24 -08:00
Myle Ott	29c8274128	Fix tests and flake8	2018-02-27 14:28:24 -08:00
Myle Ott	6641520612	fairseq-py goes distributed (#106) ... This PR includes breaking API changes to modularize fairseq-py and adds support for distributed training across multiple nodes. Changes: - c7033ef: add support for distributed training! See updated README for usage. - e016299: modularize fairseq-py, adding support for register_model, register_criterion, register_optimizer, etc. - 154e440: update LSTM implementation to use PackedSequence objects in the encoder, better following best practices and improving perf - 90c2973 and 1da6265: improve unit test coverage	2018-02-27 17:09:42 -05:00
Myle Ott	e1f49695ee	Rename LabelSmoothedCrossEntropy to LabelSmoothedNLLLoss	2017-11-08 08:01:31 -07:00
Myle Ott	6e4b7e22ee	Refactor model definitions ... * Move some functionality out of FConvModel into FairseqModel base class * Move incremental decoding functionality into FairseqIncrementalDecoder module * Refactor positional embeddings to be more specific to FConvModel	2017-11-08 07:59:22 -07:00
Sam Gross	ae0c05d920	Fix call ordering to ATen addmm and sum (#22)	2017-10-11 10:14:19 -04:00
Sergey Edunov	e734b0fa58	Initial commit	2017-09-14 17:22:43 -07:00