| 7087807 | Lukasz Kaiser | 29 June 2017, 23:20:30 UTC | Merge pull request #73 from rsepassi/push v1.0.9 | 29 June 2017, 23:20:30 UTC |
| e4fe66c | Ryan Sepassi | 29 June 2017, 20:10:03 UTC | Tweak TF_CONFIG script and bump version to 1.0.9 PiperOrigin-RevId: 160563166 | 29 June 2017, 20:18:08 UTC |
| 22ca232 | Ashish Vaswani | 29 June 2017, 19:13:33 UTC | Self-attention feed forward layer. Replaces the feed-fwd layer with a layer that does self attention across channel depth. PiperOrigin-RevId: 160556355 | 29 June 2017, 20:18:08 UTC |
| 75f398d | Lukasz Kaiser | 29 June 2017, 19:07:36 UTC | Correct decoding for class labels, add --local_eval_frequency. PiperOrigin-RevId: 160555605 | 29 June 2017, 20:18:00 UTC |
| f61ce53 | Lukasz Kaiser | 28 June 2017, 00:48:11 UTC | Corrections to make BLEU and bluenet run, debugging sharding on 1 GPU. PiperOrigin-RevId: 160352874 | 29 June 2017, 20:14:11 UTC |
| 877ba58 | Lukasz Kaiser | 27 June 2017, 22:59:36 UTC | Small training cleanups and bluenet work. PiperOrigin-RevId: 160339931 | 29 June 2017, 20:14:08 UTC |
| a83ef29 | Niki Parmar | 27 June 2017, 21:07:55 UTC | Change blue metric name, better docs PiperOrigin-RevId: 160323679 | 29 June 2017, 20:14:04 UTC |
| 85158fe | Ryan Sepassi | 27 June 2017, 02:16:40 UTC | Bump to v1.0.8 PiperOrigin-RevId: 160228099 | 29 June 2017, 20:14:01 UTC |
| a2a6178 | Lukasz Kaiser | 29 June 2017, 18:32:22 UTC | Merge pull request #57 from ReDeiPirati/algorithmic_reverse_nlplike_2 algorithmic_reverse_nlplike generator | 29 June 2017, 18:32:22 UTC |
| 06df1d4 | Lukasz Kaiser | 29 June 2017, 18:30:35 UTC | Merge pull request #69 from yynil/master I'm afraid I need to ask for pull request again for the "Out of index" Error when doing Class_Label_Modality | 29 June 2017, 18:30:35 UTC |
| af235c1 | Lukasz Kaiser | 29 June 2017, 17:16:28 UTC | Merge pull request #66 from vthorsteinsson/fix-tokens Unicode in SubwordTextEncoder | 29 June 2017, 17:16:28 UTC |
| c4c768f | Lukasz Kaiser | 29 June 2017, 16:47:52 UTC | Merge pull request #59 from pltrdy/ptb Penn TreeBank dataset for language modeling | 29 June 2017, 16:47:52 UTC |
| 9512f08 | pltrdy | 29 June 2017, 15:45:22 UTC | PTB: Fixes and removing <unk> token for out of vocabulary We no longer support a 'unk' parameter for TokenTextEncoder. Fixing some vocabulary issues | 29 June 2017, 15:45:22 UTC |
| bc75385 | vthorsteinsson | 29 June 2017, 14:28:19 UTC | Merge remote-tracking branch 'upstream/master' into fix-tokens | 29 June 2017, 14:28:19 UTC |
| c6211f1 | vthorsteinsson | 29 June 2017, 12:51:28 UTC | Separated alphabet generation; added bisect() | 29 June 2017, 12:51:28 UTC |
| bd300de | yueyu.lin | 29 June 2017, 03:37:03 UTC | Two fix: 1. for beam search use the shape.ndims to avoid out of index error 2. for greedy search, still use the shape.ndims to avoid the out of index error. Before that I misuse the slice operation:-( | 29 June 2017, 03:37:03 UTC |
| 450f339 | Lukasz Kaiser | 28 June 2017, 18:57:09 UTC | Merge pull request #63 from yynil/master Ask for a pull request to avoid "index out ouf bound" when doing prediction | 28 June 2017, 18:57:09 UTC |
| 31f5dfa | ReDeiPirati | 28 June 2017, 13:24:19 UTC | Clear comment, add sanity check, change algorithmic_reverse_nlplike max_length and add __pycache__ entry in .gitignore | 28 June 2017, 13:24:19 UTC |
| eddb835 | vthorsteinsson | 28 June 2017, 11:53:20 UTC | Merge remote-tracking branch 'upstream/master' into fix-tokens | 28 June 2017, 11:53:20 UTC |
| 5563a06 | yueyu.lin | 28 June 2017, 09:41:59 UTC | Add comments that will use a new email name | 28 June 2017, 09:41:59 UTC |
| ccbe006 | yueyu.lin | 28 June 2017, 03:57:30 UTC | Avoid "index out of bound" when decoding | 28 June 2017, 03:57:30 UTC |
| 0cf8783 | vthorsteinsson | 27 June 2017, 16:37:31 UTC | Subtoken handling enhancements | 27 June 2017, 16:37:31 UTC |
| 576bccb | vthorsteinsson | 27 June 2017, 15:42:56 UTC | Merge remote-tracking branch 'upstream/master' into fix-tokens | 27 June 2017, 15:42:56 UTC |
| 9bf7c94 | pltrdy | 27 June 2017, 14:44:33 UTC | PTB dataset | 27 June 2017, 14:47:49 UTC |
| e6e6ba2 | pltrdy | 27 June 2017, 14:34:57 UTC | Unknown token for TokenTextEncoder | 27 June 2017, 14:46:16 UTC |
| 759789b | ReDeiPirati | 27 June 2017, 13:18:55 UTC | Add algorithmic_reverse_nlplike to PROBLEM_HPARAMS_MAP | 27 June 2017, 13:18:55 UTC |
| 79f309a | ReDeiPirati | 27 June 2017, 11:25:19 UTC | Add new generator: algorithmic_reverse_nlplike that generates samples following Zipf's LAw | 27 June 2017, 11:25:19 UTC |
| 38b9c11 | ReDeiPirati | 27 June 2017, 09:56:09 UTC | Update | 27 June 2017, 09:56:09 UTC |
| 3b2fb4a | K1ngP1r@t& | 27 June 2017, 06:12:20 UTC | Merge pull request #5 from tensorflow/master Update to v1.0.8 | 27 June 2017, 06:12:20 UTC |
| de38b16 | Lukasz Kaiser | 27 June 2017, 02:18:37 UTC | Merge pull request #54 from rsepassi/push Push v1.0.8 | 27 June 2017, 02:18:37 UTC |
| 8226f15 | Ryan Sepassi | 27 June 2017, 00:48:35 UTC | Bump to v1.0.8 | 27 June 2017, 00:48:35 UTC |
| f5e371f | Lukasz Kaiser | 26 June 2017, 23:17:03 UTC | internal merge from github PiperOrigin-RevId: 160210398 | 27 June 2017, 00:45:56 UTC |
| c820307 | Lukasz Kaiser | 26 June 2017, 22:43:31 UTC | internal merge PiperOrigin-RevId: 160206168 | 27 June 2017, 00:45:51 UTC |
| 14dafcd | Noam Shazeer | 26 June 2017, 22:28:35 UTC | Make execution mode available to models and modalities by putting it in the PiperOrigin-RevId: 160204189 | 27 June 2017, 00:45:49 UTC |
| cb4d676 | Lukasz Kaiser | 26 June 2017, 18:43:09 UTC | internal merge from github PiperOrigin-RevId: 160172524 | 27 June 2017, 00:45:44 UTC |
| b9f4754 | Ryan Sepassi | 26 June 2017, 18:41:23 UTC | Fix open-source distributed training PiperOrigin-RevId: 160172257 | 27 June 2017, 00:45:42 UTC |
| 8595cc6 | vthorsteinsson | 26 June 2017, 23:35:39 UTC | Tokenizer and subword token handling in Unicode | 26 June 2017, 23:35:39 UTC |
| 49f5afc | Lukasz Kaiser | 26 June 2017, 23:04:54 UTC | Merge pull request #46 from vthorsteinsson/fix-char Use ByteTextEncoding() for encoding strings in character based models | 26 June 2017, 23:04:54 UTC |
| 1811dbe | Lukasz Kaiser | 26 June 2017, 20:56:36 UTC | Merge pull request #52 from ip01/master due to https://github.com/tensorflow/tensor2tensor/issues/47 | 26 June 2017, 20:56:36 UTC |
| 14796f2 | K1ngP1r@t& | 26 June 2017, 08:39:13 UTC | Merge pull request #4 from tensorflow/master Update | 26 June 2017, 08:39:13 UTC |
| 685a37f | Igor Pechersky | 26 June 2017, 06:48:04 UTC | due to https://github.com/tensorflow/tensor2tensor/issues/47 | 26 June 2017, 06:48:04 UTC |
| 0e84cc9 | vthorsteinsson | 26 June 2017, 00:59:23 UTC | Changed parameter name to character_vocab | 26 June 2017, 00:59:23 UTC |
| a43c4c7 | vthorsteinsson | 26 June 2017, 00:57:18 UTC | Made token_vocab required in character_generator(); updated wmt_test.py | 26 June 2017, 00:57:18 UTC |
| 98f45d1 | vthorsteinsson | 26 June 2017, 00:25:21 UTC | Merge remote-tracking branch 'upstream/master' | 26 June 2017, 00:25:21 UTC |
| f6f85aa | vthorsteinsson | 26 June 2017, 00:23:30 UTC | Merge remote-tracking branch 'upstream/master' into fix-char | 26 June 2017, 00:23:30 UTC |
| 2e85c3f | Lukasz Kaiser | 25 June 2017, 22:18:38 UTC | Merge pull request #42 from vthorsteinsson/fix-next Use iterator.__next__() for Python3 | 25 June 2017, 22:18:38 UTC |
| eee5b1c | vthorsteinsson | 25 June 2017, 15:40:32 UTC | Use ByteTextEncoding() for encoding source and target strings in character based models | 25 June 2017, 15:40:32 UTC |
| b4c0f79 | vthorsteinsson | 25 June 2017, 14:49:35 UTC | Merge remote-tracking branch 'upstream/master' into fix-next | 25 June 2017, 14:49:35 UTC |
| 99f7bef | vthorsteinsson | 25 June 2017, 14:46:20 UTC | Merge remote-tracking branch 'upstream/master' | 25 June 2017, 14:46:20 UTC |
| 38a3cf9 | Lukasz Kaiser | 25 June 2017, 03:21:08 UTC | Merge pull request #41 from vthorsteinsson/fix-import Remove urllib import (Python3 fix) | 25 June 2017, 03:21:08 UTC |
| 5f490a1 | vthorsteinsson | 25 June 2017, 01:12:34 UTC | Import urllib fix | 25 June 2017, 01:12:34 UTC |
| 7ea979c | vthorsteinsson | 25 June 2017, 01:09:18 UTC | Merge fix-next | 25 June 2017, 01:09:18 UTC |
| 792860b | vthorsteinsson | 25 June 2017, 01:00:42 UTC | File mode change | 25 June 2017, 01:00:42 UTC |
| f38b052 | vthorsteinsson | 25 June 2017, 00:59:31 UTC | Use iter.__next__ for Python3 | 25 June 2017, 00:59:31 UTC |
| d8aab8a | vthorsteinsson | 25 June 2017, 00:55:00 UTC | File attribute | 25 June 2017, 00:55:00 UTC |
| a0ffe45 | vthorsteinsson | 25 June 2017, 00:54:05 UTC | Reverted change | 25 June 2017, 00:54:05 UTC |
| 4f1d5e2 | vthorsteinsson | 25 June 2017, 00:52:39 UTC | Fixed file permissions | 25 June 2017, 00:52:39 UTC |
| bbff1ec | vthorsteinsson | 25 June 2017, 00:22:48 UTC | Fixed import | 25 June 2017, 00:22:48 UTC |
| bf7825b | Lukasz Kaiser | 24 June 2017, 22:32:24 UTC | Merge pull request #35 from jekbradbury/patch-1 fix broken reference in inference | 24 June 2017, 22:32:24 UTC |
| fc8bfc3 | Lukasz Kaiser | 24 June 2017, 22:32:07 UTC | Merge pull request #37 from rmccorm4/master update README: 'data-set' inconsistent with 'dataset' everywhere else… | 24 June 2017, 22:32:07 UTC |
| 9fbfa3d | vthorsteinsson | 24 June 2017, 14:26:43 UTC | Fixed file modes | 24 June 2017, 14:26:43 UTC |
| 9d913d7 | vthorsteinsson | 24 June 2017, 11:15:26 UTC | Merge branch 'fix/urllib' | 24 June 2017, 11:15:26 UTC |
| 6e9dedb | vthorsteinsson | 24 June 2017, 10:45:43 UTC | Sync with upstream | 24 June 2017, 11:09:58 UTC |
| 5840a71 | vthorsteinsson | 24 June 2017, 10:45:43 UTC | Sync with upstream | 24 June 2017, 10:45:43 UTC |
| dd24a5b | vthorsteinsson | 24 June 2017, 10:30:19 UTC | Sync with upstream | 24 June 2017, 10:30:19 UTC |
| 9905796 | Ryan McCormick | 24 June 2017, 06:01:15 UTC | fix 'artificats' to 'artifacts' | 24 June 2017, 06:01:15 UTC |
| 9e2eae1 | Ryan McCormick | 24 June 2017, 05:46:10 UTC | update README: 'data-set' inconsistent with 'dataset' everywhere else in the file. | 24 June 2017, 05:46:10 UTC |
| 3884fa0 | jekbradbury | 24 June 2017, 02:17:12 UTC | fix broken reference in inference The class `modality.ClassLabelModality` does not appear to exist (any more?); this change makes inference work for me. | 24 June 2017, 02:17:12 UTC |
| d029c45 | Lukasz Kaiser | 24 June 2017, 01:22:07 UTC | Merge pull request #34 from rsepassi/push v1.0.7 | 24 June 2017, 01:22:07 UTC |
| d578f52 | Ryan Sepassi | 24 June 2017, 01:06:47 UTC | internal merge PiperOrigin-RevId: 160018490 | 24 June 2017, 01:07:57 UTC |
| b53d6df | Ryan Sepassi | 24 June 2017, 01:00:22 UTC | Bump version to 1.0.7 PiperOrigin-RevId: 160018021 | 24 June 2017, 01:07:55 UTC |
| c2ce7a6 | Lukasz Kaiser | 24 June 2017, 00:37:10 UTC | Play with CIFAR models and shake-shake a little. PiperOrigin-RevId: 160016542 | 24 June 2017, 01:07:53 UTC |
| 9594213 | Ryan Sepassi | 23 June 2017, 18:15:13 UTC | gitignore update PiperOrigin-RevId: 159970261 | 24 June 2017, 01:07:51 UTC |
| b09291d | Ryan Sepassi | 23 June 2017, 18:14:41 UTC | Bump version to 1.0.6 PiperOrigin-RevId: 159970178 | 24 June 2017, 01:07:50 UTC |
| 3410bea | Vilhjalmur Thorsteinsson | 24 June 2017, 00:42:00 UTC | Python3 compatibility; better Unicode support (#22) * Python3 compatibility fixes * Removed print statements * Fixes suggested by @rsepassi * Made Python3 ByteTextEncoder compatible with Python2 * Python3 compatibility fixes | 24 June 2017, 00:42:00 UTC |
| 265ed09 | vthorsteinsson | 23 June 2017, 21:18:30 UTC | Merge branch 'master' of https://github.com/vthorsteinsson/tensor2tensor | 23 June 2017, 21:18:30 UTC |
| 848d074 | vthorsteinsson | 22 June 2017, 10:32:12 UTC | Python3 compatibility fixes | 23 June 2017, 21:15:47 UTC |
| 522d51c | vthorsteinsson | 23 June 2017, 20:57:06 UTC | Made Python3 ByteTextEncoder compatible with Python2 | 23 June 2017, 20:57:06 UTC |
| 3b244fa | K1ngP1r@t& | 23 June 2017, 08:16:56 UTC | Merge pull request #3 from tensorflow/master Update | 23 June 2017, 08:16:56 UTC |
| 204b359 | Lukasz Kaiser | 23 June 2017, 03:18:29 UTC | Merge pull request #29 from rsepassi/push Push 1.0.6 | 23 June 2017, 03:18:29 UTC |
| 2f4d5b7 | Ryan Sepassi | 23 June 2017, 02:36:01 UTC | Bump version to 1.0.6 | 23 June 2017, 02:36:01 UTC |
| 854480d | Lukasz Kaiser | 23 June 2017, 01:43:42 UTC | internal merge PiperOrigin-RevId: 159906200 | 23 June 2017, 02:29:04 UTC |
| e3e2772 | Ryan Sepassi | 23 June 2017, 00:33:22 UTC | add readme toc, improve data downloads PiperOrigin-RevId: 159900539 | 23 June 2017, 02:29:04 UTC |
| e586398 | Noam Shazeer | 22 June 2017, 04:17:58 UTC | Cap number of parallel data readers at the number of data shards. Otherwise, insufficient shuffling can lead to multiple instances of the same example in a batch, and worse quality. This is a problem in particular for single-sharded datasets. | 23 June 2017, 02:28:51 UTC |
| 0e7d44e | Ashish Vaswani | 22 June 2017, 00:36:21 UTC | fixed a small bug in transformer_parsing_base PiperOrigi-RevId: 159769630 | 23 June 2017, 02:28:31 UTC |
| 9d04261 | Lukasz Kaiser | 23 June 2017, 00:03:04 UTC | Merge pull request #27 from stefan-it/show-download-progress data-generators: show nice progress bar for download process | 23 June 2017, 00:03:04 UTC |
| 6ad92c3 | Stefan Schweter | 22 June 2017, 21:58:29 UTC | data-generators: show nice progress bar for download process | 22 June 2017, 21:58:29 UTC |
| ccf4986 | vthorsteinsson | 22 June 2017, 20:40:58 UTC | Fixes suggested by @rsepassi | 22 June 2017, 20:40:58 UTC |
| eee7bbf | Ryan Sepassi | 22 June 2017, 19:01:47 UTC | Merge pull request #26 from adgitate1/patch-1 Fix typo in README.md | 22 June 2017, 19:01:47 UTC |
| 69cf727 | adgitate1 | 22 June 2017, 18:15:51 UTC | Update README.md | 22 June 2017, 18:15:51 UTC |
| b3f8c3d | vthorsteinsson | 22 June 2017, 10:43:23 UTC | Removed print statements | 22 June 2017, 10:43:23 UTC |
| 12c6af2 | vthorsteinsson | 22 June 2017, 10:35:15 UTC | Merge remote-tracking branch 'upstream/master' | 22 June 2017, 10:35:15 UTC |
| 6baa074 | vthorsteinsson | 22 June 2017, 10:32:12 UTC | Python3 compatibility fixes | 22 June 2017, 10:32:12 UTC |
| 834c617 | K1ngP1r@t& | 22 June 2017, 08:23:08 UTC | Merge pull request #2 from tensorflow/master Update | 22 June 2017, 08:23:08 UTC |
| 8ec4233 | Lukasz Kaiser | 22 June 2017, 00:39:35 UTC | Merge pull request #13 from rsepassi/push v1.0.5 | 22 June 2017, 00:39:35 UTC |
| 8195f34 | Ryan Sepassi | 22 June 2017, 00:13:48 UTC | Remove tensorflow dependency from setup.py to enable cpu installs and bump version to 1.0.5 PiperOrigin-RevId: 159767222 | 22 June 2017, 00:26:06 UTC |
| 0fad290 | Noam Shazeer | 21 June 2017, 22:36:00 UTC | updated image transformer. now combines channels to have only 1024 positions for rev-cifar instead of 3072. PiperOrigin-RevId: 159754350 | 22 June 2017, 00:25:50 UTC |
| a8463f5 | Noam Shazeer | 21 June 2017, 19:44:28 UTC | Added parameter-attention option to transformer model. PiperOrigin-RevId: 159729158 | 21 June 2017, 21:42:55 UTC |
| b368f7c | Ryan Sepassi | 21 June 2017, 19:44:24 UTC | Add gitter links PiperOrigin-RevId: 159729147 | 21 June 2017, 21:42:54 UTC |
| 01787ca | Ryan Sepassi | 21 June 2017, 19:22:25 UTC | Add leakr dictionary to prevent internal stuff leaking out PiperOrigin-RevId: 159726434 | 21 June 2017, 21:42:53 UTC |
