issues
search
google-research
/
electra
ELECTRA: Pre-training Text Encoders as Discriminators Rather Than Generators
Apache License 2.0
2.31k
stars
351
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
Fix F1Scorer in finetune/classification/classification_metrics
#92
jgkimi
closed
3 years ago
1
Fix F1Scorer of classification_metrics
#91
jgkimi
closed
3 years ago
4
UnicodeDecodeError: 'utf-8' codec can't decode byte 0xd7 in position 0: invalid continuation byte (while running build_openwebtext_pretraining_dataset.py )
#90
elyorman
closed
3 years ago
0
Question: Same Batchsize on different TPU sizes
#89
PhilipMay
closed
3 years ago
2
Add toggle to turn off `strip_accents`.
#88
PhilipMay
closed
3 years ago
13
Sampling step?
#87
anshulsamar
opened
3 years ago
2
How can I make ELECTRA pretraining/dataloading use only one gpu ?
#86
richarddwang
closed
3 years ago
3
Finetuning loss doesn't converge when using loading weights
#85
smeaktrobush
closed
3 years ago
0
Improve Description of `--blanks-separate-docs`.
#84
PhilipMay
opened
3 years ago
0
max_predictions_per_seq and TPU training configuration
#83
Mistobaan
opened
3 years ago
0
When will the Chinese model be released?
#82
zsweet
opened
3 years ago
1
Metrics definition
#81
IssaIssa1
opened
4 years ago
0
build_pretraining_dataset.py
#80
shinhyeokoh
closed
3 years ago
1
Checkpoint
#79
Zhengxian-Fan
closed
4 years ago
0
"Device or resource busy" for mounted paths
#78
emirkin
opened
4 years ago
0
Request: pypi package for ELECTRA
#77
mgroovyank
opened
4 years ago
1
keyerror:loss.
#76
fenfaqingnian
closed
3 years ago
2
Failed to convert object of type <class 'dict'> to Tensor
#75
lizaigaoge550
closed
4 years ago
1
add code for continuing pre-training from an ELECTRA checkpoint
#74
tuvuumass
opened
4 years ago
3
The difference of reproduced results on electra_small_owt
#73
zheyuye
opened
4 years ago
5
Data loss: truncated record at 10035180
#72
jjkim-zz
opened
4 years ago
1
mask prob in large model
#71
santaonchair
closed
4 years ago
1
NaN loss during training (again)
#70
gchlodzinski
opened
4 years ago
3
Could you share GLUE dev set results for BERT-small, ELECTRA-small and ELECTRA-small++?
#69
stevezheng23
opened
4 years ago
1
Is `google/electra-small-generator` small or small++ ?
#68
richarddwang
closed
4 years ago
1
Error when pretraining on TPU: `Malformed device specification`
#67
danyaljj
closed
4 years ago
2
Training Electra on 2 phases like Bert
#66
agemagician
closed
4 years ago
3
Calculating ELECTRA infer FLOPs
#65
asharma20
closed
3 years ago
1
Question about layerwise learning rate decay
#64
TianyuZhuuu
closed
4 years ago
2
problem encountered in reproducing Electra-Large
#63
spectrometerH
closed
4 years ago
1
How to configure tensorflow_gpu 1.15?
#62
MarkClemens301
closed
3 years ago
0
Fix deprecated keyword argument in dropout layer.
#61
jarednielsen
opened
4 years ago
0
ignore PAD during dynamic masking
#60
ccchang0111
opened
4 years ago
2
Should dynamic masking also ignore ['PAD']
#59
ccchang0111
closed
3 years ago
2
[How to create vocab.txt file]
#58
Vietdung113
opened
4 years ago
4
Token-masking method: whole words or sub-words?
#57
cbaziotis
closed
4 years ago
2
RFC: List of community provided models
#56
stefan-it
opened
4 years ago
0
modified tfrecords_path split by / to accomodate windows path as well…
#55
prakashr85
opened
4 years ago
0
Issue with loading weights for eval
#54
asharma20
closed
4 years ago
2
[WIP] Define finetuning tasks in command-line hparams
#53
mapmeld
opened
4 years ago
0
BasicTokenizer: _run_strip_accents
#52
Vodolazskyi
closed
4 years ago
1
The implementation of layerwise learning rate decay
#51
importpandas
closed
4 years ago
2
KeyError: '[SEP]'
#50
elyesmanai
closed
4 years ago
5
problem on electra's pretraining method
#49
real-brilliant
closed
4 years ago
1
Low usage of gpu
#48
amy-hyunji
closed
4 years ago
2
Add keep_checkpoint_max parameter
#47
stefan-it
closed
4 years ago
0
Build Dataset Issue
#46
qute012
closed
4 years ago
1
'adam_m not found in checkpoint ' when further pretraining
#45
DayuanJiang
closed
4 years ago
6
`num_train_steps` for further pretraining
#44
DayuanJiang
closed
4 years ago
1
Format of corpus
#43
mahnerak
closed
4 years ago
4
Previous
Next