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

• f0b19e1da6 spelling master wassname 2019-11-28 16:35:03 +08:00
• a3216db803 fix sklearn warning on f1 score BCELoss wassname 2019-11-27 11:08:32 +08:00
• 8d9f6cb60c misc wassname 2019-11-26 22:37:36 +08:00
• 67da707648 tidy and metrics wassname 2019-11-26 21:03:17 +08:00
• 6deed0107b remove binary databunch wassname 2019-11-26 15:49:19 +08:00
• 762167f5f3 draft funtions and metrics wassname 2019-11-26 15:47:42 +08:00
• 46ec76ec78 multiclass (but single classes don't work) wassname 2019-11-26 12:19:32 +08:00
• 7557805043 made a binary classification version making binary category label list wassname 2019-11-26 10:16:23 +08:00
• 6f1d1232b4 Add pwc badges Piotr Czapla 2019-11-12 12:50:43 +01:00
• f4ece1c083 Update README.md with an example how to train your own language model pretrain-lm Piotr Czapla 2019-11-11 09:55:48 +01:00
• 8f23d396b9 Make form_pretrained accept a path as well as a pretarined model name Piotr Czapla 2019-11-11 09:51:11 +01:00
• 9eac0bb42e Add command line convenience function & ability limit data set Piotr Czapla 2019-11-11 09:50:29 +01:00
• d838727d51 Merge pull request #55 from n-waves/multifit Piotr Czapla 2019-11-07 09:56:48 +01:00
• 7d6cacb91d Add the missing LICENSE file. Piotr Czapla 2019-11-07 09:51:37 +01:00
• 26d681754f Fix spelling errors in README.md Piotr Czapla 2019-11-07 09:29:52 +01:00
• c009b8e0ba Rename ulmfit to multfit to make room for future directions of the repo. Piotr Czapla 2019-11-07 09:29:36 +01:00
• f29f453fa8 Merge branch 'master' into multifit Piotr Czapla 2019-11-06 11:36:54 +01:00
• 59e9966fdc Update to README.md with resutls and links to example notebooks Piotr Czapla 2019-11-06 11:32:11 +01:00
• 400e54dec1 Compatiblity with fastai 1.0.47 Piotr Czapla 2019-11-06 10:45:48 +01:00
• d85bd94241 README + requirements Piotr Czapla 2019-11-06 10:45:18 +01:00
• 30993f81a3 Fix command line training & rename _learner to get_learner Piotr Czapla 2019-11-06 10:44:29 +01:00
• 8f46753195 Fix the weight decay - issue that was causing lower results on DE. Piotr Czapla 2019-11-02 09:35:13 +01:00
• a47b20fcf5 Add option to turn off label smoothing eps normalisation & weight convertion New weights -> Old format Piotr Czapla 2019-11-02 09:34:40 +01:00
• a4b06149b1 Expose method to get configured text processor Piotr Czapla 2019-11-02 09:32:53 +01:00
• e009a02aab Update the paper_version configuration to match more closely original experiements Piotr Czapla 2019-10-21 10:18:52 +02:00
• 3c2364e3d8 Clean up bugs caused by the new tokenizer class Piotr Czapla 2019-10-21 10:04:59 +02:00
• c9cdf09d5b Create LICENCE ulmfit-multilingual-original-scripts Julian Eisenschlos 2019-10-15 19:59:50 +02:00
• 031e0c18eb Add diff printing during load_ Piotr Czapla 2019-10-15 16:55:15 +02:00
• 70c74a1cc5 Refactor tokenization Piotr Czapla 2019-10-15 04:17:54 +02:00
• 77a2780a6b Clean up datasets Piotr Czapla 2019-10-15 00:57:40 +02:00
• ad50c73377 Make ULMFiT name consistant. Piotr Czapla 2019-10-15 00:12:31 +02:00
• 43f5c1282d Add sotabench scripts Piotr Czapla 2019-10-15 00:09:18 +02:00
• f846bf8a4b Add OpenFIle preproc to SentencePiece preproc Piotr Czapla 2019-10-14 17:20:26 +02:00
• b362f2a7cb from_pretrained Piotr Czapla 2019-10-14 17:18:28 +02:00
• 5a3fbcecb5 Ability to use fastai databunch directly Piotr Czapla 2019-10-14 17:18:13 +02:00
• 6336f5ebb8 Make it possible to create databunch out of dataframes Piotr Czapla 2019-10-14 13:14:07 +02:00
• 91dbd9bc84 Clean up configurations Piotr Czapla 2019-10-14 13:13:09 +02:00
• 7847331751 Remove cupy from requirements.txt Piotr Czapla 2019-10-14 13:11:52 +02:00
• 230d31362e Merge pull request #27 from n-waves/polyglot-lm Piotr Czapla 2019-10-08 22:16:54 +02:00
• 96a870b9b9 Merge pull request #49 from cahya-wirawan/tokens_min Piotr Czapla 2019-10-08 21:46:33 +02:00
• 1fe2bd46a6 finall clean ups Piotr Czapla 2019-10-07 14:50:03 +02:00
• 66341801c0 added question of minimal tokens/article in to prepare_wiki.sh changed 'fname' to 'file' in load_data function to match the current fastai version Cahya Wirawan 2019-09-11 12:49:58 +02:00
• 3fe6c19af2 fix spelling eeror in finetune_lm Piotr Czapla 2019-09-08 08:43:15 +02:00
• cac4df27bd Add paper version configuration and orignal ulmfit Piotr Czapla 2019-09-07 22:37:26 +02:00
• c390b4ed35 Add code to JA-multifit Piotr Czapla 2019-09-07 21:47:43 +02:00
• dd473aa98d Add repoduction of Multifit result for JA using newest hyper params Piotr Czapla 2019-09-07 21:18:07 +02:00
• 1ff9e01766 Remove unused code Piotr Czapla 2019-09-07 21:17:32 +02:00
• ee7c23a3be Remove old multifit logs Piotr Czapla 2019-09-07 19:12:44 +02:00
• 02ee52d0ef Update README.md Piotr Czapla 2019-09-07 19:11:52 +02:00
• 26e54a9c7d Use the pervious datset_path from finetuning for classsificator training Piotr Czapla 2019-09-07 19:11:29 +02:00
• 2cdf380adf New bolerplate code Piotr Czapla 2019-09-07 16:19:09 +02:00
• 8162cc5648 Clean up the old multfit training code Piotr Czapla 2019-09-07 15:22:01 +02:00
• 14cda72d76 remaining exeperiments and logs done for poleval 19 reproduce-poleval Piotr Czapla 2019-08-29 15:51:59 +02:00
• 2fe5c8a588 Update to fastai v1.0.57 - use new sentence piece implementaiton & sizes of hidden layers Piotr Czapla 2019-08-29 15:51:17 +02:00
• 68d6b1c829 fix bug in the hack for poleval reddit Piotr Czapla 2019-06-10 20:31:04 +02:00
• a7ac4f5170 missing file Piotr Czapla 2019-06-10 20:14:57 +02:00
• 7d59c4ecf2 Sentence piece has new option -fix that changes coverage to 99.95% Piotr Czapla 2019-06-10 20:14:11 +02:00
• ce0a29f385 Make it possible to set ftseed in poleval19_init Piotr Czapla 2019-06-06 17:45:13 +02:00
• 854391a130 Merge branch 'reproduce-poleval' of https://github.com/n-waves/ulmfit-multilingual into reproduce-poleval Piotr Czapla 2019-05-19 00:17:27 +02:00
• 38e131f802 Make the ulmfit compatible with old fastai (ulmfit_multilingual) Piotr Czapla 2019-05-19 00:17:03 +02:00
• 7326df2c98 Train sentencepiece on unsup Marcin 2019-05-19 00:11:59 +02:00
• 6f4db9de1a Fix ensemble output (by removing index column) Piotr Czapla 2019-05-18 22:50:55 +02:00
• 06c32ae2dc Add ability to create folders when saving ensemble output Piotr Czapla 2019-05-18 21:59:12 +02:00
• f38f8c0670 fix issue in the databunch name generation Piotr Czapla 2019-05-18 21:48:32 +02:00
• 8927e45fb2 Add ensemble commnad Piotr Czapla 2019-05-18 21:48:08 +02:00
• 15c2d0663a Save predictions for a proper dataset Marcin 2019-05-18 03:49:44 +02:00
• 7a33ea5d1f Add no-test + ability to start training from the classifcation data set without wiki Piotr Czapla 2019-05-18 13:10:29 +02:00
• 75b934bb83 Merge branch 'reproduce-poleval' of https://github.com/n-waves/ulmfit-multilingual into reproduce-poleval Piotr Czapla 2019-05-15 18:00:45 +02:00
• 7c9129670e Poleval 19 ablation experiments Piotr Czapla 2019-05-15 18:00:40 +02:00
• 30f5a7901c Save predictions Marcin 2019-05-14 15:34:26 +02:00
• 7955db37ad Merge branch 'reproduce-poleval' of https://github.com/n-waves/ulmfit-multilingual into reproduce-poleval Piotr Czapla 2019-05-14 13:27:03 +02:00
• ad3c11edb8 Add ability to turnoff weighted CrossEntropy Piotr Czapla 2019-05-14 13:26:57 +02:00
• 6dc4a5f102 Add Kappa and Mathew score calcualtion + ls command to main Piotr Czapla 2019-05-14 11:53:16 +02:00
• 746c6ca185 Don't reset lmseed Marcin 2019-05-14 06:57:02 +02:00
• fe31778d70 Add option to remove duplicates Marcin 2019-05-14 02:30:25 +02:00
• 84405b6e6c Allow bwd in model name Marcin 2019-05-13 17:51:04 +02:00
• b065e36f1b Fix double name exception Piotr Czapla 2019-05-13 20:02:48 +02:00
• 9101a0313c Let poleval19_full accept name Piotr Czapla 2019-05-13 19:53:45 +02:00
• 29b600b8b5 Add ability to check different model save Piotr Czapla 2019-05-13 17:47:57 +02:00
• b55715e4f6 Fix ftseed used in pretrain_lm so ulmfit lm works again Piotr Czapla 2019-05-13 17:47:28 +02:00
• 6ef75e4dd7 Fix issue in poleval_eval + add valid metrics Piotr Czapla 2019-05-13 09:17:59 +02:00
• 1c3044f5df fix issue with cls_best add lmseed to poleval Piotr Czapla 2019-05-13 03:44:35 +02:00
• 53eadfc232 Add poleval19_full and fix name inference from paths Piotr Czapla 2019-05-13 03:25:12 +02:00
• d11deac01a Add poleval commands + saving evaluation metrics to cache Piotr Czapla 2019-05-13 02:51:01 +02:00
• c42bd8d9ed Update to eval to support seeds values Piotr Czapla 2019-05-13 00:07:46 +02:00
• 724233585e Make it compatible with fastai 51 Piotr Czapla 2019-05-12 22:19:02 +02:00
• 50b73510de Add workaround for metrics init Marcin 2019-05-12 21:43:54 +02:00
• 6aff0a9e22 Add more seeds Marcin 2019-05-12 21:29:32 +02:00
• 38e422ff0e fix dropout poleval2018 Ubuntu 2019-05-11 22:00:34 +00:00
• 19507f1d38 seed + new tokens for poleval2018 Ubuntu 2019-05-11 19:51:27 +00:00
• 3125ce7d8b Fix seeds Marcin 2019-05-11 16:59:59 +02:00
• 8f2b7edee4 Merge pull request #45 from cahya-wirawan/tokens_min Piotr Czapla 2019-05-07 00:09:31 +02:00
• 27a720b105 Make decoder bias optional Marcin 2019-05-01 20:57:20 +02:00
• 616bec18dc Added an option to set the minimal limit of tokens per article Cahya Wirawan 2019-04-30 11:48:14 +02:00
• f5306b4064 Add a few examples of commands avaliable in ulmfit-multilingual Piotr Czapla 2019-04-20 14:34:26 +02:00
• f9032ad846 Update to README.md Piotr Czapla 2019-04-20 14:28:45 +02:00
• a312572004 Merge branch 'master' of https://github.com/n-waves/ulmfit-multilingual Piotr Czapla 2019-04-19 10:34:18 +02:00
• 9f494b3ec6 Remove old bilm code that wasn't working Piotr Czapla 2019-04-19 10:32:35 +02:00
• ba186113dc Merge pull request #43 from n-waves/fix-tests Julian Eisenschlos 2019-04-15 15:14:40 -03:00
• 2b7ec26f8b Lower LM accuracy test threshold NAUSICAA\Julian 2019-04-15 15:00:46 -03:00