Commits · 189c1b91a63f3653934f9796e9e28f4686a4f914 · chenpangpang / transformers

15 Dec, 2020 1 commit
- native amp leak fix landed in 1.7.1 (#9115) · 14c79c3e
  Stas Bekman authored Dec 15, 2020
```
update README with good news that the leak fix has been applied to pytorch-1.7.1.
```
  14c79c3e
11 Dec, 2020 1 commit

Sylvain Gugger authored Dec 11, 2020



* Reorganize example folder

* Continue reorganization

* Change requirements for tests

* Final cleanup

* Finish regroup with tests all passing

* Copyright

* Requirements and readme

* Make a full link for the documentation

* Address review comments

* Apply suggestions from code review
Co-authored-by: Lysandre Debut <lysandre@huggingface.co>

* Add symlink

* Reorg again

* Apply suggestions from code review
Co-authored-by: Thomas Wolf <thomwolf@users.noreply.github.com>

* Adapt title

* Update to new strucutre

* Remove test

* Update READMEs
Co-authored-by: Lysandre Debut <lysandre@huggingface.co>
Co-authored-by: Thomas Wolf <thomwolf@users.noreply.github.com>

783d7d26

04 Dec, 2020 2 commits

[seq2seq] document the caveat of leaky native amp (#8930) · df311a5c

Stas Bekman authored Dec 04, 2020



* document the caveat of leaky native amp

* Update examples/seq2seq/README.md
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>

df311a5c

[s2s finetune_trainer] add instructions for distributed training (#8884) · 4c3d98dd
Stas Bekman authored Dec 03, 2020

4c3d98dd

26 Nov, 2020 1 commit
- potpurri of small fixes (#8807) · ddf3c646
  Stas Bekman authored Nov 26, 2020
  
  ddf3c646
11 Nov, 2020 1 commit
- [s2s] distill t5-large -> t5-small (#8376) · 81ebd706
  Sumithra Bhakthavatsalam authored Nov 11, 2020
```
Co-authored-by: Sam Shleifer <sshleifer@gmail.com>
```
  81ebd706
05 Nov, 2020 1 commit
- Docs bart training ref (#8330) · 17450397
  Leandro von Werra authored Nov 05, 2020
```
Co-authored-by: Sam Shleifer <sshleifer@gmail.com>
```
  17450397
29 Oct, 2020 1 commit
- [s2s] distillBART docs for paper replication (#8150) · 49e4fece
  Sam Shleifer authored Oct 29, 2020
  
  49e4fece
20 Oct, 2020 1 commit
- [s2s] create doc for pegasus/fsmt replication (#7934) · 0e24e4c1
  Stas Bekman authored Oct 20, 2020
  
  0e24e4c1
11 Oct, 2020 1 commit
- [examples] bump pl=0.9.0 (#7053) · 827c5194
  Sam Shleifer authored Oct 11, 2020
  
  827c5194
09 Oct, 2020 1 commit
- [s2s] Switch README urls to cdn (#7670) · 297233fa
  Sam Shleifer authored Oct 08, 2020
  
  297233fa
27 Sep, 2020 1 commit

[s2s] add create student script (#7290) · eab5f596

Suraj Patil authored Sep 28, 2020


Co-authored-by: Suraj Patil <surajp815@gmail.com>
Co-authored-by: Sam Shleifer <sshleifer@gmail.com>

eab5f596

25 Sep, 2020 1 commit
- doc changes (#7385) · 415071b4
  Suraj Patil authored Sep 25, 2020
  
  415071b4
24 Sep, 2020 3 commits
- Seq2SeqTrainer (#6769) · 9e68d075
  Suraj Patil authored Sep 25, 2020
```
Co-authored-by: Sam Shleifer <sshleifer@gmail.com>
```
  9e68d075
- [s2s] distributed eval allows num_return_sequences > 1 (#7254) · d9d0f114
  Sam Shleifer authored Sep 24, 2020
  
  d9d0f114
- [seq2seq] make it easier to run the scripts (#7274) · eadd870b
  Stas Bekman authored Sep 24, 2020
  
  eadd870b
22 Sep, 2020 1 commit
- [s2s] add supported architecures to MD (#7252) · 25b0463d
  Sam Shleifer authored Sep 22, 2020
  
  25b0463d
17 Sep, 2020 1 commit
- [s2s] dynamic batch size with --max_tokens_per_batch (#7030) · a5638b2b
  Sam Shleifer authored Sep 17, 2020
  
  a5638b2b
16 Sep, 2020 2 commits
- [s2s] distributed eval cleanup (#7186) · 0203ad43
  Sam Shleifer authored Sep 16, 2020
  
  0203ad43
- [s2s run_eval] new features (#7109) · fdaf8ab3
  Stas Bekman authored Sep 16, 2020
```
Co-authored-by: Sam Shleifer <sshleifer@gmail.com>
```
  fdaf8ab3
03 Sep, 2020 1 commit
- tweak tar command in readme (#6919) · b8e4906c
  brett koonce authored Sep 03, 2020
  
  b8e4906c
30 Aug, 2020 1 commit
- [s2s README] Add more dataset download instructions (#6737) · dfa10a41
  Sam Shleifer authored Aug 30, 2020
  
  dfa10a41
28 Aug, 2020 1 commit

prepare_seq2seq_batch makes labels/ decoder_input_ids made later. (#6654) · 9336086a

Sam Shleifer authored Aug 28, 2020

* broken test

* batch parity

* tests pass

* boom boom

* boom boom

* split out bart tokenizer tests

* fix tests

* boom boom

* Fixed dataset bug

* Fix marian

* Undo extra

* Get marian working

* Fix t5 tok tests

* Test passing

* Cleanup

* better assert msg

* require torch

* Fix mbart tests

* undo extra decoder_attn_mask change

* Fix import

* pegasus tokenizer can ignore src_lang kwargs

* unused kwarg test cov

* boom boom

* add todo for pegasus issue

* cover one word translation edge case

* Cleanup

* doc

9336086a

17 Aug, 2020 1 commit
- [s2s] docs, document desired filenames nicely (#6525) · 72add6c9
  Sam Shleifer authored Aug 16, 2020
  
  72add6c9
11 Aug, 2020 1 commit
- rename prepare_translation_batch -> prepare_seq2seq_batch (#6103) · be1520d3
  Sam Shleifer authored Aug 11, 2020
  
  be1520d3
06 Aug, 2020 1 commit
- [s2s]Use prepare_translation_batch for Marian finetuning (#6293) · 2804fff8
  Sam Shleifer authored Aug 06, 2020
```
Co-authored-by: Sylvain Gugger <35901082+sgugger@users.noreply.github.com>
```
  2804fff8
03 Aug, 2020 1 commit
- [s2s] Document better mbart finetuning command (#6229) · 57eb1cb6
  Sam Shleifer authored Aug 03, 2020
```
* Document better MT command

* improve multigpu command
```
  57eb1cb6
01 Aug, 2020 1 commit
- [s2s] clean up + doc (#6184) · d8dbf3b7
  Stas Bekman authored Aug 01, 2020
```
Co-authored-by: Sam Shleifer <sshleifer@gmail.com>
```
  d8dbf3b7
30 Jul, 2020 1 commit
- [s2s] add support for overriding config params (#6149) · 3212b885
  Stas Bekman authored Jul 29, 2020
  
  3212b885
28 Jul, 2020 3 commits
- link to README.md (#6068) · f0c70085
  Stas Bekman authored Jul 28, 2020
```
* add a link to README.md

* Update README.md
```
  f0c70085
- MBART: support summarization tasks where max_src_len > max_tgt_len (#6003) · 3c7fbf35
  Sam Shleifer authored Jul 28, 2020
```
* MBART: support summarization tasks

* fix test

* Style

* add tokenizer test
```
  3c7fbf35
- [s2s] Don't mention packed data in README (#6079) · 7a68d401
  Sam Shleifer authored Jul 27, 2020
  
  7a68d401
27 Jul, 2020 1 commit
- [s2s] dont document packing because it hurts performance (#6077) · 1e00ef68
  Sam Shleifer authored Jul 27, 2020
  
  1e00ef68
22 Jul, 2020 1 commit
- [test] partial coverage for train_mbart_enro_cc25.sh (#5976) · c3206eef
  Sam Shleifer authored Jul 22, 2020
  
  c3206eef
21 Jul, 2020 2 commits
- [examples/seq2seq]: add --label_smoothing option (#5919) · 5b193b39
  Sam Shleifer authored Jul 21, 2020
  
  5b193b39
- typos in seq2seq/readme (#5937) · ccbf74a6
  Aditya Soni authored Jul 21, 2020
  
  ccbf74a6
18 Jul, 2020 4 commits
- Seq2SeqDataset uses linecache to save memory by @Pradhy729 (#5792) · 09a2f406
  Sam Shleifer authored Jul 18, 2020
```
Co-authored-by: Pradhy729 <49659913+Pradhy729@users.noreply.github.com>
```
  09a2f406
- [seq2seq] distillation.py accepts trainer arguments (#5865) · dad5e12e
  Sam Shleifer authored Jul 18, 2020
  
  dad5e12e
- [seq2seq] MAX_LEN env var for MT commands (#5837) · ba240018
  Sam Shleifer authored Jul 17, 2020
  
  ba240018
- Lightning Updates for v0.8.5 (#5798) · 529850ae
  Nathan Raw authored Jul 17, 2020
```
Co-authored-by: Sam Shleifer <sshleifer@gmail.com>
```
  529850ae