SUMIN/HuggingFace_transformer
1
0
Fork 0
Code Issues Pull Requests Actions 7 Packages Projects Releases Wiki Activity
1,922 Commits 5 Branches 239 Tags
3e9420add1e74fb4e900e3cfee415e77343eae41
Commit Graph

515 Commits

Author SHA1 Message Date
Thomas Wolf
9beaa85b07 Merge pull request #1055 from qipeng/run_squad_fix 
Fix #1015 (tokenizer defaults to use_lower_case=True when loading from trained models)
 2019-08-21 01:20:46 +02:00
Lysandre
2d042274ac Sequence special token handling for BERT and RoBERTa 2019-08-20 14:15:28 -04:00
Peng Qi
3bffd2e8e5 more fixes 2019-08-20 10:59:28 -07:00
Thomas Wolf
3b56427a1e Merge pull request #1040 from FeiWang96/multi_gpu 
Fix bug of multi-gpu training in lm finetuning
 2019-08-20 17:13:44 +02:00
thomwolf
a690edab17 various fix and clean up on run_lm_finetuning 2019-08-20 15:52:12 +02:00
erenup
fc74132598 add best steps to train 2019-08-20 19:06:41 +08:00
Duzeyao
d86b49ac86 swap optimizer.step and scheduler.step 2019-08-20 16:46:34 +08:00
Duzeyao
45ab8bf60e Revert "Update finetune_on_pregenerated.py" 
This reverts commit a1359b970c.
 2019-08-20 16:40:39 +08:00
erenup
97c30b73d5 add test related code 2019-08-20 16:31:04 +08:00
erenup
d5e60e5b7a add test related code 2019-08-20 16:25:50 +08:00
Zeyao Du
a1359b970c Update finetune_on_pregenerated.py 2019-08-20 16:00:07 +08:00
Zeyao Du
28f7ca1f80 swap optimizer.step and scheduler.step 2019-08-20 15:58:42 +08:00
Peng Qi
a368b87791 Fix #1015 2019-08-19 13:07:00 -07:00
Lysandre
f94f1c6016 Distributed training + tokenizer agnostic mask token 2019-08-19 14:58:50 -04:00
Thomas Wolf
5a49b793d9 Merge pull request #1023 from tuvuumass/patch-1 
fix issue #824
 2019-08-19 15:31:46 +02:00
erenup
4270d3da1b fix a bug of evaluating 2019-08-19 16:38:52 +08:00
Chi-Liang Liu
40acf6b52a don't save model without training 2019-08-18 05:02:25 -04:00
erenup
47e9aea0fe add args info to evaluate_result.txt 2019-08-18 17:00:53 +08:00
erenup
5582bc4b23 add multiple choice to robreta and xlnet, test on swag, roberta=0.82.28 
, xlnet=0.80
 2019-08-18 16:01:48 +08:00
wangfei
856a63da4d Fix: save model/model.module 2019-08-18 11:03:47 +08:00
wangfei
1ef41b8337 Revert "Fix: save model/model.module" 
This reverts commit 00e9c4cc96.
 2019-08-18 11:03:12 +08:00
wangfei
00e9c4cc96 Fix: save model/model.module 2019-08-18 11:02:02 +08:00
erenup
e384ae2b9d Merge remote-tracking branch 'huggingface/master' 
merge huggingface/master to update
 2019-08-17 12:05:57 +08:00
Jason Phang
d8923270e6 Correct truncation for RoBERTa in 2-input GLUE 2019-08-16 16:30:38 -04:00
Lysandre
5652f54ac2 Simplified data generator + better perplexity calculator 
GPT-2 now obtains ~20 perplexity on WikiText-2
 2019-08-16 13:49:56 -04:00
LysandreJik
7e7fc53da5 Fixing run_glue example with RoBERTa 2019-08-16 11:53:10 -04:00
LysandreJik
715534800a BERT + RoBERTa masking tokens handling + GPU device update. 2019-08-16 10:10:21 -04:00
LysandreJik
339e556feb CLM for BERT, beginning of CLM fot RoBERTa; still needs a better masking token mechanism. 2019-08-16 10:10:20 -04:00
LysandreJik
5c18825a18 Removed dataset limit 2019-08-16 10:10:20 -04:00
LysandreJik
3e3e145497 Added GPT to the generative fine-tuning. 2019-08-16 10:10:20 -04:00
LysandreJik
47975ed53e Language Modeling fine-tuning using GPT-2. 2019-08-16 10:10:20 -04:00
wangfei
b8ff56896c Fix bug of multi-gpu training in lm finetuning 2019-08-16 12:11:05 +08:00
Rabeeh KARIMI
3d47a7f8ab loads the tokenizer for each checkpoint, to solve the reproducability issue 2019-08-14 10:58:26 +02:00
LysandreJik
39f426be65 Added special tokens <pad> and <mask> to RoBERTa. 2019-08-13 15:19:50 -04:00
Julien Chaumond
baf08ca1d4 [RoBERTa] run_glue: correct pad_token + reorder labels 2019-08-13 12:51:15 -04:00
tuvuumass
ba4bce2581 fix issue #824 2019-08-13 11:26:27 -04:00
Julien Chaumond
912fdff899 [RoBERTa] Update run_glue for RoBERTa 2019-08-12 13:49:50 -04:00
erenup
b219029c45 refactoring old run_swag. This script is mainly refatored from run_squad in pytorch_transformers 2019-08-11 15:20:37 +08:00
Thomas Wolf
b4f9464f90 Merge pull request #960 from ethanjperez/patch-1 
Fixing unused weight_decay argument
 2019-08-07 10:09:55 +02:00
Thomas Wolf
d43dc48b34 Merge branch 'master' into auto_models 2019-08-05 19:17:35 +02:00
thomwolf
70c10caa06 add option mentioned in #940 2019-08-05 17:09:37 +02:00
thomwolf
b90e29d52c working on automodels 2019-08-05 16:06:34 +02:00
Ethan Perez
28ba345ecc Fixing unused weight_decay argument 
Currently the L2 regularization is hard-coded to "0.01", even though there is a --weight_decay flag implemented (that is unused). I'm making this flag control the weight decay used for fine-tuning in this script.
 2019-08-04 12:31:46 -04:00
Thomas Wolf
c054b5ee64 Merge pull request #896 from zijunsun/master 
fix multi-gpu training bug when using fp16
 2019-07-26 19:31:02 +02:00
zijunsun
f0aeb7a814 multi-gpu training also should be after apex fp16（squad） 2019-07-26 15:23:29 +08:00
zijunsun
adb3ef6368 multi-gpu training also should be after apex fp16 2019-07-25 13:09:10 +08:00
Chi-Liang Liu
a7fce6d917 fix squad v1 error (na_prob_file should be None) 2019-07-24 16:11:36 +08:00
thomwolf
6070b55443 fix #868 2019-07-23 17:46:01 +02:00
thomwolf
2c9a3115b7 fix #858 2019-07-23 16:45:55 +02:00
Thomas Wolf
268c6cc160 Merge pull request #845 from rabeehk/master 
fixed version issues in run_openai_gpt
 2019-07-23 15:29:31 +02:00