Commit Graph

Select branches

Hide Pull Requests

BCELoss

bilm

cls

datasets

distill

lbl_smoothing

master

poleval2018

polyglot-lm

pretrain-lm

reproduce-poleval

self-attention-module

text_cols

thai2fit

tmp

ulmfit-multilingual-original-scripts

xnli

Mono Color

• e97085337e Fix dropout and classification accuracy. 0.91 on imdb Piotr Czapla 2018-11-17 16:47:17 +01:00
• d796610452 Removed deprecated default_rules Sebastian 2018-11-17 14:48:06 +00:00
• 7fe059c9a2 Fixed models path for vocabulary Sebastian 2018-11-17 14:47:27 +00:00
• e1418b2114 fix model dir mkdir Piotr Czapla 2018-11-17 00:11:56 +01:00
• 5cc4bfd444 Add loading weights and itos, so that we can extend training Piotr Czapla 2018-11-16 23:48:22 +01:00
• 895a58c88d Merge branch 'master' into bilm Piotr Czapla 2018-11-16 23:39:15 +01:00
• 64ad8d8fd9 Merge pull request #8 from n-waves/sentencepiece Piotr Czapla 2018-11-16 23:35:01 +01:00
• 36b056a465 Fix issues discovered during execution of end-to-end test. Piotr Czapla 2018-11-16 23:31:16 +01:00
• a21f100ab2 Merge branch 'master' into sentencepiece Piotr Czapla 2018-11-16 23:23:49 +01:00
• aa6b59a0b3 Fix vocab size so that it remains 60k as in case of pretrain_lm + add longer lm training Piotr Czapla 2018-11-16 23:19:15 +01:00
• ebbf09775f Merge branch 'master' into bilm Piotr Czapla 2018-11-16 20:09:41 +01:00
• 45389be413 Use drop_mult properly Piotr Czapla 2018-11-16 20:02:32 +01:00
• 23b17da61e Add end to end test and improve the train_clas params Piotr Czapla 2018-11-16 19:51:29 +01:00
• 977a506cfd Fix bug where LM was always intialized as QRNN despite the paramters. Piotr Czapla 2018-11-16 19:50:36 +01:00
• e72b19499c fix indent Aayush 2018-11-16 23:54:48 +05:30
• 59c8852b5d sentencepiece for pretraining modified: fastai_contrib/utils.py modified: ulmfit/pretrain_lm.py aayush 2018-11-16 23:14:24 +05:30
• fb2ed98ac6 Merge branch 'master' into sentencepiece Aayush 2018-11-16 23:09:06 +05:30
• f0e538f621 Add support for QRNN training to train_clas Piotr Czapla 2018-11-16 17:04:52 +01:00
• d4698e095c Change back temporary folder stucture Piotr Czapla 2018-11-16 00:21:14 +01:00
• 363130bcff Update to the docs Piotr Czapla 2018-11-16 00:17:41 +01:00
• 87afa54f63 Merge pull request #9 from n-waves/classifier-fixes Piotr Czapla 2018-11-15 23:48:16 +01:00
• d5d61d31d2 Add directory structure to readme. Piotr Czapla 2018-11-15 23:47:48 +01:00
• 6ba3e68f9e Minor tweaks. Piotr Czapla 2018-11-15 23:41:02 +01:00
• 08d71dd45e Add script that fetches and prepres WT-2 WT-103 Piotr Czapla 2018-11-15 23:40:50 +01:00
• 9a60ef2fbd Add assertions to train_clas and ability to limit the dataset size Piotr Czapla 2018-11-15 23:40:10 +01:00
• d8b95430b8 Chain all scripts needed to prepare a wiki dump, put .unk to new folder Piotr Czapla 2018-11-15 23:39:27 +01:00
• 111fc7e4c3 Add perpare_imdb script Piotr Czapla 2018-11-15 23:37:15 +01:00
• b9c587d9bd Adding SentencePieceTokenizer NAUSICAA\Julian 2018-11-15 19:18:42 -03:00
• 43340d03bc Fix minor typos in train_clas Nirant K 2018-11-15 18:04:15 +00:00
• c9c827c420 Updated method to new fastai v1 version xnli Sebastian 2018-11-15 17:51:44 +00:00
• 1cc49e666f Quick fixes to imports etc. Aayush 2018-11-15 23:13:16 +05:30
• 3e8f9f8b5f [WIP] Sub-word tokenization with sentencepiece Aayush 2018-11-15 23:01:49 +05:30
• 56fc150eb1 Merge xnli Nirant K 2018-11-15 15:53:37 +00:00
• 653f6e96a0 Classifier fails at fit_one_cycle Nirant K 2018-11-15 15:03:52 +00:00
• 6622458fe7 Add aclImdb extractor Nirant K 2018-11-15 15:03:12 +00:00
• 94fa1a3ecb Encapsulated data reading in utils method, removed fastai processing, added doc string Sebastian 2018-11-15 14:36:50 +00:00
• 300d78ee44 Consolidated methods for reading classification data, added method to read XNLI data Sebastian 2018-11-15 14:34:07 +00:00
• d5e6849cd8 Shortened XNLI download script, removed unnecessary parts; extracts now to data folder Sebastian 2018-11-15 14:33:21 +00:00
• 33f9eb2cc7 Reuse RNNCore in implementation of BiLM, add accuracy Piotr Czapla 2018-11-14 21:16:16 +01:00
• 62e6777952 Add download and unzip script Nirant K 2018-11-14 15:53:59 +00:00
• 898c9255e0 Fix minor typos Nirant 2018-11-14 21:07:15 +05:30
• cf7d93c378 Changes to pretrain_lm so that it works with bilm Piotr Czapla 2018-11-14 14:59:23 +01:00
• 0fb8161061 Merge branch 'master' of https://github.com/n-waves/ulmfit-multilingual Nirant K 2018-11-14 13:56:58 +00:00
• 5ba83b1d4e Working version of BILM - probably won't train well yet Piotr Czapla 2018-11-14 14:55:56 +01:00
• 70680c21ce Add better logs, add multi-language flag to pretrain Nirant K 2018-11-14 13:54:31 +00:00
• a85800610b WIP Working Backward LM using our new LangaugeModelLoader Piotr Czapla 2018-11-14 13:22:03 +01:00
• 6ee1a2b27d Add BiLM LanguageModelLoader with tests Piotr Czapla 2018-11-14 12:48:29 +01:00
• 5f845724d0 Add dependencies Nirant 2018-11-14 17:13:40 +05:30
• 820ff231bd Add Download and Process Docs Nirant 2018-11-14 17:04:47 +05:30
• 81519a6066 Merge branch 'master' into bilm Piotr Czapla 2018-11-14 11:41:50 +01:00
• 38ca35c044 Fix pretrain_lm to work with yesterdays changes to fastai text API Piotr Czapla 2018-11-14 11:41:26 +01:00
• cba86ab5d4 Added initial classes and changes for BiLM implementation Sebastian 2018-11-13 17:46:28 +00:00
• e13dbea1e1 Remove missleading docs. Piotr Czapla 2018-11-12 13:14:44 +01:00
• 5ad6c2ace1 Update to new fastai.master with fixed qrnn issue Piotr Czapla 2018-11-12 13:08:00 +01:00
• 40ca6c200e Refactor and temporary fix the pretrain_lm by adding learn.lr_find Piotr Czapla 2018-11-10 00:32:29 +01:00
• eb1da0b630 Updated README.md Piotr Czapla 2018-11-08 21:05:48 +01:00
• 12e52b182e Initial import from fastaiv1 Piotr Czapla 2018-11-08 20:33:10 +01:00
• 2f09109d5d Initial commit Piotr Czapla 2018-11-08 19:52:07 +01:00