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

• e4d0529a3d Removing more bi-directional test NAUSICAA\Julian 2019-04-15 14:41:08 -03:00
• 94be5e6e0f Remove unused tests Piotr Czapla 2019-04-14 19:19:12 +02:00
• 8facc21cfa Add F1 metrics to evaluate Piotr Czapla 2019-04-14 19:13:43 +02:00
• 65397d7ae4 Bi-dir not supported yet NAUSICAA\Julian 2019-04-13 16:38:58 -03:00
• 7e0a8517b6 Remove config params from language_model_learner funcion NAUSICAA\Julian 2019-04-13 16:08:01 -03:00
• edf3b8d76e Adding missing param arch NAUSICAA\Julian 2019-04-13 12:32:19 -03:00
• a948d7df91 Merge branch 'master' of https://github.com/n-waves/ulmfit-multilingual Piotr Czapla 2019-03-26 21:29:33 +01:00
• 0a823be17b Maki ti possible to not load unusp Piotr Czapla 2019-03-26 21:29:23 +01:00
• 24f1d741f0 Add generating of pseudo labels Piotr Czapla 2019-03-26 21:27:56 +01:00
• 66ec30a122 Add more results Piotr Czapla 2019-03-26 21:27:18 +01:00
• debee4887d Train using LASER distribution output distill Marcin 2019-03-19 02:21:23 +01:00
• 7744021014 Merge pull request #40 from n-waves/text_cols Julian Eisenschlos 2019-03-07 13:28:00 -03:00
• 43588b43e1 Move logs text_cols NAUSICAA\Julian 2019-03-07 13:19:40 -03:00
• 227f3b5dff Merge branch 'master' of https://github.com/n-waves/ulmfit-multilingual into text_cols NAUSICAA\Julian 2019-03-07 13:18:52 -03:00
• 837925ff53 Imporved validate_cls & eval to pick the best model based on val accuracy tmp Piotr Czapla 2019-03-03 13:29:49 +01:00
• e793425ef7 Merge pull request #37 from n-waves/qrnn_perf Julian Eisenschlos 2019-03-01 18:06:13 -03:00
• 7b2ac9e94b Add ability to use random-init=True Piotr Czapla 2019-03-01 17:41:07 +01:00
• 11b2b14523 Fix -m ulmfit tar method Piotr Czapla 2019-03-01 17:39:58 +01:00
• 02522fcd44 Rename file NAUSICAA\Julian 2019-02-28 08:17:41 -03:00
• 71e7637dd0 CLS Results NAUSICAA\Julian 2019-02-27 23:44:28 -03:00
• fd022c6826 Merge branch 'master' of https://github.com/n-waves/ulmfit-multilingual into text_cols NAUSICAA\Julian 2019-02-27 23:30:59 -03:00
• 4742a68d28 Some logs from previous experiments Piotr Czapla 2019-02-27 20:31:38 +01:00
• 632ec602f6 update score thai2fit Charin 2019-02-27 15:09:50 +07:00
• f750586114 Remove redundant bptt param Marcin 2019-02-26 18:39:38 +01:00
• c6e0373170 Make models use bptt parameter Piotr Czapla 2019-02-26 18:06:23 +01:00
• 61234cd1dc add BERT link Charin 2019-02-25 20:58:47 +07:00
• c1a1f998c8 add bert Charin 2019-02-25 20:57:57 +07:00
• cb19852a7f Thai Classification Benchmark Charin 2019-02-25 20:52:15 +07:00
• dd74a6d282 Remove previous multi field patch NAUSICAA\Julian 2019-02-24 23:15:20 -03:00
• 06cd4d4d0b Merge branch 'master' of https://github.com/n-waves/ulmfit-multilingual into text_cols NAUSICAA\Julian 2019-02-24 22:58:46 -03:00
• 852336c29b Merge pull request #38 from n-waves/lbl_smoothing Piotr Czapla 2019-02-22 16:57:47 +01:00
• 260faa703c Correct the label smoothing implementation lbl_smoothing Piotr Czapla 2019-02-22 16:55:35 +01:00
• 015f04ec08 Training with noise & label smoothing Piotr Czapla 2019-02-22 12:02:18 +01:00
• c69d31c420 New results lstm 30k 1cyc Piotr Czapla 2019-02-22 12:01:25 +01:00
• c5a93d19fe Changes moved to different branch NAUSICAA\Julian 2019-02-20 17:16:08 -03:00
• 61ae344fda Remove fastai warning NAUSICAA\Julian 2019-02-20 17:02:56 -03:00
• c698c97772 Adding logs NAUSICAA\Julian 2019-02-20 17:02:02 -03:00
• b39fb5390b Adding QRNN benchmark NAUSICAA\Julian 2019-02-20 15:21:58 -03:00
• 4700d95225 Merge branch 'master' of https://github.com/n-waves/ulmfit-multilingual into qrnn_perf NAUSICAA\Julian 2019-02-20 15:16:31 -03:00
• 99d6b22447 Correct noise generation training + convenience functions Piotr Czapla 2019-02-20 10:24:30 +01:00
• c29180a08f Add some results Piotr Czapla 2019-02-20 10:22:44 +01:00
• 13ae29a95d Set mark_fields in True NAUSICAA\Julian 2019-02-18 22:20:24 -03:00
• 5529ec385b Add sentence piece char_coverage 0.99 for non european languages Piotr Czapla 2019-02-18 21:50:46 +01:00
• 7dc7aac327 Merge all columns in classification task into first column Piotr Czapla 2019-02-18 21:50:04 +01:00
• 9fbcf56df3 Add different learning schedules, with default to the old schedule Piotr Czapla 2019-02-18 21:49:16 +01:00
• 458c06f779 Fix df name NAUSICAA\Julian 2019-02-18 17:05:11 -03:00
• dbb929e3ca Adding more text cols to use all CLS data NAUSICAA\Julian 2019-02-18 16:54:00 -03:00
• ce6cc607ae Add saving itos.pkl so that the LM can be used to finetuning Piotr Czapla 2019-02-17 23:18:42 +01:00
• c3276da062 Fixing Imdb loading Piotr Czapla 2019-02-17 23:18:12 +01:00
• 119417fb6e Disable early stopping as it was causing OOMs Piotr Czapla 2019-02-17 23:04:25 +01:00
• 490c792278 Upgrade to the recent the todays version of Fastai Piotr Czapla 2019-02-17 23:03:54 +01:00
• 0dda4b4c2c Zeroshot MLDoc results for ulmfit trained on 10k examples Piotr Czapla 2019-02-17 19:03:02 +01:00
• 1ee1dd950d Merge branch 'master' of https://github.com/n-waves/ulmfit-multilingual into cls cls NAUSICAA\Julian 2019-02-16 16:24:01 -03:00
• b7e3a5e7a4 Update BERT Zero-shot results Julian Eisenschlos 2019-02-16 14:42:40 -03:00
• be0511e42b Adding Bert results Julian Eisenschlos 2019-02-15 15:51:35 -03:00
• 8733487d55 Make the validate vs train decision based on the existance of cls_last.pth istead of a model directory Piotr Czapla 2019-02-15 01:17:53 +01:00
• 5dced1e488 Remove bidir Piotr Czapla 2019-02-15 01:16:37 +01:00
• 5e44e442b2 Merge branch 'master' of https://github.com/n-waves/ulmfit-multilingual Piotr Czapla 2019-02-15 01:11:42 +01:00
• 0e6534ad7b Expose num_lm_epochs in ulmfit eval Piotr Czapla 2019-02-15 01:11:39 +01:00
• 22cb8b1660 Download pretrained models Marcin 2019-02-15 00:00:20 +01:00
• 0f084168c1 Merge branch 'master' of https://github.com/n-waves/ulmfit-multilingual Piotr Czapla 2019-02-14 22:35:29 +01:00
• cd47b3b5dc Fix use_moses=True for mldoc so that it is identical to wiki with uses_moses=False Piotr Czapla 2019-02-14 22:35:20 +01:00
• c28c0fde16 Make ulmfit eval more secure and give more flexibility in dataset_template Piotr Czapla 2019-02-14 22:28:25 +01:00
• 03c4a2cd15 Fixed self attention, local tests passed self-attention-module aayush 2019-02-15 02:10:18 +05:30
• 6fda7f2cda Download preprocessed wikis Marcin 2019-02-14 16:45:51 +01:00
• fdac9f7ccd Save only the best LM model Marcin 2019-02-14 14:01:33 +01:00
• 72e86cefc9 Add QRNN LM results for Italian Marcin 2019-02-14 12:48:04 +01:00
• 0535ef169a QRNN mldoc results for de, en and es Marcin 2019-02-14 11:05:02 +01:00
• b609951561 Fix path of pretrained model Marcin 2019-02-14 00:00:56 +01:00
• 9d893e393b Add QRNN results Marcin 2019-02-14 00:00:43 +01:00
• f95a16cee1 Merge pull request #34 from n-waves/use-configs Marcin Kardas 2019-02-13 22:58:47 +01:00
• 1340f4235c Improve ulmfit eval to allow for zeroshot laser evaluation Piotr Czapla 2019-02-13 15:29:55 +01:00
• b14a393671 Add more results including full zeroshot results Piotr Czapla 2019-02-13 15:29:16 +01:00
• 4215111240 jp -> ja NAUSICAA\Julian 2019-02-13 10:50:01 -03:00
• e657423e7f CLS Download Script NAUSICAA\Julian 2019-02-13 10:41:24 -03:00
• 0e2211dfaf Merge branch 'master' into use-configs Marcin 2019-02-13 14:10:59 +01:00
• 2eee051c67 Expose max length parameter Marcin 2019-02-13 13:40:05 +01:00
• 3586a7dcf9 Merge branch 'pr/29' Piotr Czapla 2019-02-12 20:43:44 +01:00
• 866d0c0bed Addiing XLingual LM polyglot-lm NAUSICAA\Julian 2019-02-12 12:52:12 -03:00
• 4e1b76feee Add MLDoc summary & zeroshot logs Piotr Czapla 2019-02-12 15:01:42 +01:00
• e72cdfb6db Add result logs (it) Piotr Czapla 2019-02-12 15:01:22 +01:00
• e7271f2a29 Add ability to evalulate multiple models at once Piotr Czapla 2019-02-12 15:01:01 +01:00
• a1e7a79b57 Add result logs Piotr Czapla 2019-02-12 15:00:29 +01:00
• 8497cc1e5c Move LM and classifier parameters to configs Marcin 2019-02-11 15:43:43 +01:00
• a630242f97 Expose validate_cls in ulmfit module Piotr Czapla 2019-02-11 11:01:08 +01:00
• c63fe258d2 Fix validataion and add option to add noise to training labels Piotr Czapla 2019-02-11 10:53:25 +01:00
• 22e40d81f3 Additional DE results and first results on FR and JA Piotr Czapla 2019-02-11 10:51:42 +01:00
• e7ec9f0233 fix markdown formatting Piotr Czapla 2019-02-10 10:52:03 +01:00
• 07ab9ccaf8 Add first MLDoc results Piotr Czapla 2019-02-10 09:53:42 +01:00
• 672ef2d59f Remove old result logs Piotr Czapla 2019-02-10 09:53:26 +01:00
• 26736d95de Add code to test & train mldoc classifier Piotr Czapla 2019-02-10 09:52:58 +01:00
• 950daac903 Merge branch 'backwards-lm' Tomasz Pietruszka 2019-01-23 00:08:22 +01:00
• c6cf44a7b9 Fixed test_bilm_classifier_loads_encoder - inconsistent settings, problematic input (too short) Tomasz Pietruszka 2019-01-23 00:05:38 +01:00
• 386fc49431 Adapted test_text_data to batch dimension being the first, and only xxbos token at the start Tomasz Pietruszka 2019-01-22 20:31:30 +01:00
• 0e864fb291 test_end_to_end now working, added instructions to README, sentencepiece dependency Tomasz Pietruszka 2019-01-22 20:10:38 +01:00
• 6fa30829c4 Merge branch 'master' into backwards-lm Tomasz Pietruszka 2019-01-14 17:20:48 +01:00
• dd296f3088 prepare_wiki.sh made executable Tomasz Pietruszka 2019-01-13 17:30:35 +01:00
• 5ba8ae4c59 non-ascii char removed from code. Caused display bugs Tomasz Pietruszka 2019-01-13 17:29:55 +01:00
• b00410e0cb LM save with_opt fix Tomasz Pietruszka 2019-01-13 17:28:39 +01:00
• a1e66c39d4 tokenzier->tokenizer typo Tomasz Pietruszka 2019-01-13 17:27:29 +01:00