Dmytro Okhonko 860010e907 Handle 3+ dimensional input in sequence_generator + nits ... Summary: sequence_generator assumes that model input is 2d tensor of longs. But it can be something like 3d tensor of floats and we should be able to handle this as long as first dimension is batch size followed by source lengths. Reviewed By: myleott Differential Revision: D14420044 fbshipit-source-id: bf8b1e42ad1873f7b803c1a377b0af21648db015		2019-03-12 15:12:21 -07:00
..
__init__.py	fairseq-py goes distributed (#106)	2018-02-27 17:09:42 -05:00
test_average_checkpoints.py	Merge internal changes (#136)	2018-04-02 10:13:07 -04:00
test_backtranslation_dataset.py	Modularize generate.py (#351)	2019-02-22 10:08:52 -08:00
test_binaries.py	Handle 3+ dimensional input in sequence_generator + nits	2019-03-12 15:12:21 -07:00
test_character_token_embedder.py	fix tests	2018-09-03 19:15:23 -04:00
test_convtbc.py	Remove more Variable() calls (#198)	2018-06-25 12:23:04 -04:00
test_dictionary.py	Move string line encoding logic from tokenizer to Dictionary (unified diff). (#541)	2019-02-28 09:19:12 -08:00
test_iterators.py	Further generalize EpochBatchIterator and move iterators into new file	2018-09-03 19:15:23 -04:00
test_label_smoothing.py	Add FairseqTask	2018-06-15 13:05:22 -06:00
test_noising.py	Refactor BacktranslationDataset to be more reusable (#354)	2018-11-25 21:26:03 -08:00
test_reproducibility.py	Add Tensorboard support (#530)	2019-02-25 18:40:18 -08:00
test_sequence_generator.py	Modularize generate.py (#351)	2019-02-22 10:08:52 -08:00
test_sequence_scorer.py	Modularize generate.py (#351)	2019-02-22 10:08:52 -08:00
test_token_block_dataset.py	Merge internal changes (#483)	2019-01-30 09:01:10 -08:00
test_train.py	Add code for "Pay Less Attention with Lightweight and Dynamic Convolutions" (#473)	2019-01-25 15:40:26 -08:00
test_utils.py	Remove more Variable() calls (#198)	2018-06-25 12:23:04 -04:00
utils.py	Fix tests + style nits + Python 3.5 compat	2018-11-01 01:28:30 -07:00