https://github.com/allenai/document-qa

sort by:
Revision Author Date Message Commit Date
2f9fa68 Delete out-dated DotMerge layergst 29 May 2018, 20:16:19 UTC
48c598d Updates train_bidaf.py 14 May 2018, 23:34:43 UTC
21fc045 Merge pull request #32 from RachelKer/master dev and train files in config.py 14 May 2018, 23:06:45 UTC
79acd1f dev and train files in config.py Uses the dev and train files in config.py instead of data/squad by default when building a dataset. 11 May 2018, 10:04:39 UTC
35c9b86 Fix SQuAD ablate script if not in paragraph model 22 February 2018, 22:10:20 UTC
c6d7864 Minor README fixes 10 December 2017, 00:47:00 UTC
05c6e23 Adds TriviaQA Wiki processing This adds the ablate script, ensure the evaluation scripts works with this dataset, and unifies the ranked evluation scripts into "ranked_scores.py" 26 November 2017, 00:52:52 UTC
b868ff7 Merge branch 'master' of https://github.com/allenai/document-qa 26 November 2017, 00:48:16 UTC
a672534 Adds Apache 2.0 LICENSE 22 November 2017, 20:49:04 UTC
902f902 Fix server bugs 22 November 2017, 18:26:32 UTC
a3352bf Removes outdated method in `Preprocessor`" 22 November 2017, 10:44:41 UTC
d3feaba Updates server Use Bing v7.0 Properly cache Sessions Log in a way more compatable with Sanic, and log to a file Dismiss errors when a new search is done 22 November 2017, 10:32:35 UTC
7f13a4e Support bing api v7 and span/confidence output for QaSystem 21 November 2017, 11:17:52 UTC
01bedd4 Improves doc 21 November 2017, 11:16:31 UTC
d14f37b avoids length-zero spans 21 November 2017, 11:13:32 UTC
7fd6c7e Removes console.log 21 November 2017, 10:24:37 UTC
91c8e29 Updates boilerpipe.jare to properly exit on timeouts 21 November 2017, 10:24:12 UTC
e7b494e Minor doc improvements 21 November 2017, 09:00:22 UTC
9272ac8 Minor doc fixes in docqa/evaluator.py 19 November 2017, 08:36:10 UTC
8c8639b Add server code 19 November 2017, 08:08:24 UTC
8519d8d Require pandas >= 0.18.0 19 November 2017, 04:05:18 UTC
8df62b5 Parameterizes deliminator in 19 November 2017, 04:04:22 UTC
6ca7cac Merge pull request #11 from rajasagashe/fix-print-bug correctly prints directory name in value error 17 November 2017, 05:21:55 UTC
6d94433 correctly prints directory name in value error 17 November 2017, 05:10:14 UTC
37fe1ff Adds run_on_user_documents.py example 15 November 2017, 19:56:44 UTC
a3af4a4 Renames variables in TextPreprocessor 15 November 2017, 19:52:37 UTC
d4831e5 Adds document splitter 15 November 2017, 19:51:51 UTC
026eb57 Mildly improves interface 15 November 2017, 19:51:14 UTC
3072ae0 Fixes switched em/f1 + comments `MultiParagraphSpanEvaluator` 14 November 2017, 18:22:10 UTC
3420f18 Optimizes imports in ablate_triviaqa.py 13 November 2017, 23:03:19 UTC
04f40c9 Removes incorrect best_weight setting in ablate_triviaqa 13 November 2017, 23:02:53 UTC
30d67f0 Removes old print statement 13 November 2017, 22:53:15 UTC
7cf2f91 Bugfix in ablate_unfiltered 13 November 2017, 22:53:00 UTC
6bcba14 Fixes sorting for eval printing in trivia_full_document_eval 13 November 2017, 22:52:45 UTC
a0c132e Links pre-trained models in README 11 November 2017, 22:07:48 UTC
f35c30f Improves convert_to_cpu.py 11 November 2017, 22:05:18 UTC
b9258b6 Fixes arguement validation bug 11 November 2017, 06:10:47 UTC
b540601 Fixes .all_docs for questions from TriviaQA wiki 10 November 2017, 22:22:57 UTC
453316d Adds backward-compatible derserialization support for CharWordEmbedder 10 November 2017, 22:11:27 UTC
29322a5 Adds TriviaQaWikiDataset class 10 November 2017, 22:10:01 UTC
950e656 Fixes pickle error 10 November 2017, 22:09:29 UTC
56ddc67 Sort correcty in triviaqa_full_document_eval.py 10 November 2017, 22:08:56 UTC
5dd7ce9 Comment alignment_bias 10 November 2017, 07:59:19 UTC
51359e6 Sort questions in squad_eval 10 November 2017, 07:59:01 UTC
994af87 Adds ELMo SQuAD model 10 November 2017, 07:48:25 UTC
dab7c34 Improves evaluation scripts No longer requires ema flag, bugfix squad_full_document, and print table-format output for the document scripts. 09 November 2017, 09:12:55 UTC
616a0cd Removes uneeded import 09 November 2017, 07:02:58 UTC
b447274 Formats doc string 09 November 2017, 07:02:22 UTC
495b6b8 Removes some defunct analysis scripts 07 November 2017, 06:26:21 UTC
3d2f6f3 Fixes README typo + improve ablate scripts help output 06 November 2017, 16:35:12 UTC
18edd07 Fixes more README links 03 November 2017, 08:32:27 UTC
46888bb Fix README link 02 November 2017, 17:59:31 UTC
ea5e550 Minor fix to configurations 01 November 2017, 06:24:28 UTC
7d52c92 Updates README with a link to the paper 31 October 2017, 00:16:41 UTC
3d34a55 Adds a few more links to README 30 October 2017, 17:35:51 UTC
2fcc1f1 Escape tildes in README 30 October 2017, 16:52:22 UTC
22f36a1 Clean scripts and a few comments, finish workable README 30 October 2017, 16:52:17 UTC
2130443 Rename trivia_qa to triviaqa + work on README 30 October 2017, 16:51:15 UTC
85d83fc Removes tensorflow from requirements since it should be installed seperately 30 October 2017, 16:51:15 UTC
7d4715d Remove commented out hacks 30 October 2017, 16:51:15 UTC
4d1a060 Minor cleanup + work on README 30 October 2017, 16:51:15 UTC
359c30e Refactor into 'docqa' and minor tweaks 30 October 2017, 16:42:09 UTC
6354d08 Cleanup scripts 24 September 2017, 05:08:22 UTC
f8d7e89 Lots of clean + tweaks to training + some analysis/evaluation stuff 24 September 2017, 05:03:26 UTC
0ab3309 Cleanup .gitignore 24 September 2017, 04:59:24 UTC
e897ab9 Delete experimental 24 September 2017, 04:57:42 UTC
e9ac976 Delete old files 24 September 2017, 04:56:05 UTC
3ec7a08 Shift more dead code to experimental, reboot placeholder imp just to see if it matter on SQuAD 14 September 2017, 02:36:57 UTC
c06fc84 Cleanup, some new training/visualizing code 13 September 2017, 05:30:31 UTC
e9aa686 Some re-organizing + add word-len masks 08 September 2017, 17:02:36 UTC
a9502ab Improving 'inverse' tokenization supprot + a few new approaches to confidence 31 August 2017, 17:10:43 UTC
866b632 Cleanup and put experimental stuff in its own directory 31 August 2017, 17:09:30 UTC
178bac7 Chunk of work on modelling fir TriviaQa, doc level SQuAD, and in-memory multi-paragraph training data 26 August 2017, 20:55:32 UTC
3822fd8 Clean up name shuffeling 21 August 2017, 04:21:07 UTC
fa03815 Adds niket/aristo dat + cuddn work 21 August 2017, 00:32:12 UTC
6c430ca Adds training with text-derived spans 11 August 2017, 21:25:02 UTC
d0b69cb Fix unittests + minor layer fixes 11 August 2017, 00:26:44 UTC
ee4ab20 Minor fixes 11 August 2017, 00:21:58 UTC
8acb8fa Minor cleanup and rename runner->trainer 10 August 2017, 23:08:05 UTC
b3d6637 Dump of work so far for on triviaqa and squad 10 August 2017, 22:52:22 UTC
back to top