d69b0b0e90
fixes + clean up + mask is long
2018-11-04 21:26:54 +01:00
3ddff783c1
clean up + mask is long
2018-11-04 21:26:44 +01:00
88c1037991
update requirements
2018-11-04 21:26:18 +01:00
d0cb9fa2a7
clean up model
2018-11-04 21:26:11 +01:00
6cc651778a
update readme
2018-11-04 21:26:03 +01:00
efb44a8310
distributed in extract features
2018-11-04 21:25:55 +01:00
d9d7d1a462
update float()
2018-11-04 21:25:36 +01:00
c6207d85b6
remove old methods
2018-11-04 15:34:00 +01:00
965b2565a0
add distributed training
2018-11-04 15:32:04 +01:00
1ceac85e23
add gradient accumulation
2018-11-04 15:26:14 +01:00
6b0da96b4b
clean up
2018-11-04 15:17:55 +01:00
834b485b2e
logging + update copyright
2018-11-04 12:07:38 +01:00
1701291ef9
multi-gpu cleanup
2018-11-04 11:54:57 +01:00
5ee171689c
what's in loss again
2018-11-04 11:45:44 +01:00
0b7a20c651
add tqdm, clean up logging
2018-11-04 11:07:34 +01:00
d4e3cf3520
add numpy import
2018-11-04 10:54:16 +01:00
cf366417d5
remove run_squad_pytorch
2018-11-04 09:56:00 +01:00
26bdef4321
fixing verbose_argument
2018-11-04 09:53:29 +01:00
d6418c5ef3
tweaking the readme
2018-11-03 23:52:35 +01:00
3b70b270e0
update readme
2018-11-03 23:39:55 +01:00
eaa6db92f1
Merge branch 'master' of https://github.com/huggingface/pytorch-pretrained-BERT
2018-11-03 23:35:16 +01:00
f8276008df
update readme, file names, removing TF code, moving tests
2018-11-03 23:35:14 +01:00
f18ae210e1
fix typo
2018-11-03 22:34:37 +00:00
3c24e4bef1
Multi-Gpu loss - Cleaning
2018-11-03 18:03:17 -04:00
5de1517d6b
WIP modeling_test_pytorch.py
2018-11-03 22:40:50 +01:00
1ba5b58c20
fix typo
2018-11-03 17:10:23 -04:00
5858e8e4dd
Fix both loss and eval metrics -> more coherence on the loss (eval vs train and tf vs pt)
2018-11-03 16:48:24 -04:00
cd09cd5b40
Fix import on initalization
2018-11-03 15:38:30 -04:00
ec66841afa
WIP
2018-11-03 19:12:20 +01:00
139873f6e3
Merge branch 'master' of https://github.com/huggingface/pytorch-pretrained-BERT
2018-11-03 19:06:17 +01:00
04287a4d68
special edition script
2018-11-03 19:06:15 +01:00
a1af5247e1
Add seed in initialization
2018-11-03 14:00:36 -04:00
4faeb38b51
Fix loss loss logging for multi-gpu compatibility
2018-11-03 17:52:51 +00:00
25f73add07
update optimizer run_squad
2018-11-03 17:56:34 +01:00
f514cbbf30
update run_squad with tqdm
2018-11-03 17:52:44 +01:00
cb76c1ddd3
add model.zero_grad()
2018-11-03 17:40:12 +01:00
a4086c5de5
Merge branch 'master' of https://github.com/huggingface/pytorch-pretrained-BERT
2018-11-03 17:38:17 +01:00
088ad45888
fixing optimization
2018-11-03 17:38:15 +01:00
8bd6b235b7
typo on tokenization
2018-11-03 10:27:59 -04:00
2c55568c40
`scatter_` and `scatter`
2018-11-03 10:27:38 -04:00
a6efe1235f
Merge pull request #1 from huggingface/multi-gpu-support
...
Create DataParallel model if several GPUs
2018-11-03 10:10:34 -04:00
5f432480c0
Create DataParallel model if several GPUs
2018-11-03 10:10:01 -04:00
5889765a7c
Update README.md
2018-11-03 09:18:44 -04:00
8c932e37f9
Update the comparison notebook
2018-11-03 09:08:05 -04:00
391a4ec2f3
Small typo in `trange`
...
I seriously don't understand why they defined num_train_epochs as a float in the originial tf code.
I Will change it at the end to avoir merge conflicts for now.
2018-11-03 08:25:15 -04:00
5676d6f799
Remove BERT pretraining files for now
2018-11-03 08:17:22 -04:00
8ec457d370
Fix imports
2018-11-03 12:35:21 +01:00
852e4b3c00
Merge branch 'master' of https://github.com/huggingface/pytorch-pretrained-BERT
2018-11-03 12:23:04 +01:00
0d8d2285ba
fix optimization_test
2018-11-03 12:23:00 +01:00
574e20a92e
Move modeling_test.py to /tensorflow_code
2018-11-03 12:22:06 +01:00
thomwolf