Commit Graph

15053 Commits

Author SHA1 Message Date
Nikolay Korolev
c8933bb2d9
Delete nonexistent parameter from documentation
Changed documentation of GPT2Model, GPT2LMHeadModel and GPT2DoubleHeadsModel
2019-08-27 12:10:36 +03:00
LysandreJik
e08c01aa1a fix #1102 2019-08-26 18:13:06 -04:00
LysandreJik
84a3a9689d Pytorch Hub & AutoModels 2019-08-26 16:08:43 -04:00
LysandreJik
f68339639a Tests for added AutoModels 2019-08-26 16:02:23 -04:00
LysandreJik
cb60ce59dd Added multiple AutoModel classes: AutoModelWithLMHead, AutoModelForQuestionAnswering and AutoModelForSequenceClassification 2019-08-26 15:44:30 -04:00
LysandreJik
529a16dec6 Generic encoding implementation. 2019-08-26 15:00:43 -04:00
Shijie Wu
f1b018740c Add use_lang_emb to config 2019-08-23 20:33:01 -04:00
Shijie Wu
e85123d398 Add custom tokenizer for zh and ja 2019-08-23 20:27:52 -04:00
thomwolf
06510ccb53 typo 2019-08-23 22:08:10 +02:00
thomwolf
3bcbebd440 max_len_single_sentence & max_len_sentences_pair as attributes so they can be modified 2019-08-23 22:07:26 +02:00
Shijie Wu
436ce07218 Tokenization behave the same as original XLM proprocessing for most languages except zh, ja and th; Change API to allow specifying language in tokenize 2019-08-23 14:40:17 -04:00
thomwolf
ab7bd5ef98 fixing tokenization and training 2019-08-23 17:31:21 +02:00
thomwolf
47d6853439 adding max_lengths for single sentences and sentences pairs 2019-08-23 17:31:11 +02:00
Thomas Wolf
df9d6effae
Merge pull request #1081 from huggingface/fix_distributed_barrier_hang
Fix distributed barrier hang
2019-08-23 16:53:53 +02:00
Thomas Wolf
3f20dd7186
Merge pull request #1075 from abhishekraok/modeling_utils_config_None
reraise EnvironmentError in modeling_utils.py
2019-08-23 12:42:39 +02:00
David Pollack
e13465fb8b change layernorm code to pytorch's native layer norm 2019-08-23 12:12:12 +02:00
Abhishek Rao
c603d099aa reraise EnvironmentError in from_pretrained functions of Model and Tokenizer 2019-08-22 15:25:40 -07:00
LysandreJik
2ba1a14fb0 Decode now calls private property instead of public method 2019-08-22 17:25:55 -04:00
Thomas Wolf
90dcd8c05d
Merge branch 'master' into generative-finetuning 2019-08-22 10:43:30 +02:00
VictorSanh
57272d5ddf fix for glue 2019-08-22 00:25:49 -04:00
VictorSanh
b006a7a12f fix for squad 2019-08-22 00:25:42 -04:00
Abhishek Rao
14eef67eb2 Fix at config rather than model 2019-08-21 15:48:43 -07:00
Abhishek Rao
296df2b18c reraise exception 2019-08-21 15:29:30 -07:00
Lysandre
55f69a11b6 OpenAI GPT tests now extend CommonTests 2019-08-21 18:09:25 -04:00
Lysandre
47267ba556 OpenAI GPT-2 now depends on CommonTests. 2019-08-21 17:50:16 -04:00
Lysandre
034aa0c2d7 Fixed GPT2DoubleHeadsModel example and weight tying 2019-08-21 17:27:38 -04:00
thomwolf
e00b4ff1de fix #1017 2019-08-21 22:22:17 +02:00
Lysandre
814a3f4e01 Removed attention_mask from GPT-2 and GPT documentation. Corrected multiple_choice_labels to actual name mc_labels 2019-08-21 14:11:14 -04:00
Lysandre
2f9397139d Added GPT-2 LARGE to Pre-trained Models documentation 2019-08-21 11:29:37 -04:00
Lysandre
d6bbcbc4cf Added finetuning example to documentation 2019-08-21 11:22:05 -04:00
VictorSanh
6f877d9daf Update dev results on GLUE (bert-base-uncased) w/ median on 5 runs 2019-08-21 03:43:29 +00:00
Thomas Wolf
07681b6b58
Merge pull request #1064 from huggingface/gpt-2-large
Adding gpt-2 large (774M parameters) model
2019-08-21 03:05:56 +02:00
thomwolf
fdc487d8b3 Add max length 2019-08-21 02:35:01 +02:00
thomwolf
aa05dc8935 adding gpt-2 large 2019-08-21 02:29:34 +02:00
Thomas Wolf
e4515faf54
Merge pull request #1057 from huggingface/fixes
Add a few of typos corrections, bugs fixes and small improvements
2019-08-21 01:54:05 +02:00
Thomas Wolf
41789c6c3d
Merge pull request #1059 from GuillemGSubies/master
Better use of spacy tokenizer in open ai and xlm tokenizers
2019-08-21 01:53:48 +02:00
Thomas Wolf
260c86082d
Merge pull request #1027 from samvelyan/iterative_split_on_token
Re-implemented tokenize() iteratively in PreTrainedTokenizer.
2019-08-21 01:46:03 +02:00
Thomas Wolf
d30cbaf5dc
Merge branch 'master' into iterative_split_on_token 2019-08-21 01:33:02 +02:00
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
Thomas Wolf
e753f249e1
Merge pull request #806 from wschin/fix-a-path
Fix a path so that a test can run on Windows
2019-08-21 01:14:40 +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
c3619f5536
Merge pull request #1060 from CrafterKolyan/patch-1
Fix typo. configuratoin -> configuration
2019-08-20 17:39:06 +02:00
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
43489756ad adding proxies options for the from_pretrained methods 2019-08-20 16:59:11 +02:00
thomwolf
a690edab17 various fix and clean up on run_lm_finetuning 2019-08-20 15:52:12 +02:00
Nikolay Korolev
ad6e62cd82
Fix typo. configuratoin -> configuration 2019-08-20 15:43:06 +03:00
Guillem García Subies
388e3251fa
Update tokenization_xlm.py 2019-08-20 14:19:39 +02:00
Guillem García Subies
f5e2ed0fd8
Update tokenization_openai.py 2019-08-20 14:19:25 +02:00
Guillem García Subies
562b998366
Update tokenization_openai.py 2019-08-20 14:10:19 +02:00