Yige Xu
|
6cb3713b2c
|
修复针对pytorch1.3.0版本bug的一个bug
|
6 years ago |
Yige Xu
|
bf2920cba9
|
update reproduction/matching/matching_bert.py
|
6 years ago |
xuyige
|
ea0f2f7e00
|
update reproduction/matching to adapt version 0.5.0: 1) move loader codes from DataLoader to PiPe; 2) fix some bugs in matching pipe; 3) delete some expire codes.
|
6 years ago |
ChenXin
|
de17c9a7d3
|
rename base_loader file as data_bundle
|
6 years ago |
xuyige
|
570b214dfb
|
增加fastNLP.embeddings模块并修改对应的现有代码以适配fastNLP.embeddings
|
6 years ago |
xuyige
|
7d07b38e0a
|
fix bug in matching DataLoader
|
6 years ago |
xuyige
|
83f9cb193f
|
update README.md
|
6 years ago |
yh
|
5c80c6f4f5
|
将DataInfo修改为DataBundle
|
6 years ago |
FFTYYY
|
97c7ba313d
|
添加了mwan模型,并稍微修改了matching dataloader
|
6 years ago |
xuyige
|
d70aa96e4c
|
大幅度更新:1、更新requirements;2、将modules.aggregator的内容移至modules.encoder;3、将SQuADMetric重命名为ExtractiveQAMetric;4、更新reproduction的README;5、将reproduction/text_classification的dataloader移动到fastNLP.io.data_loader并做适配性修改
|
6 years ago |
lionel
|
2a1d5dc2a4
|
Add cntn model for matching.
|
6 years ago |
xuyige
|
00cf9820a2
|
fix a bug in matching loader
|
6 years ago |
xuyige
|
8f729b6656
|
merge matching loader to fastNLP package
|
6 years ago |
xuyige
|
25ff06b5b7
|
Update MatchingDataLoader.py
|
6 years ago |
xuyige
|
1bc780a2bd
|
update framework in matching tasks
|
6 years ago |
xuyige
|
ba34a56fa7
|
Create README.md
create reproduction/matching/README.md
|
6 years ago |
xuyige
|
15d9581e6d
|
fix a bug in predictor
|
6 years ago |
xuyige
|
50faa936b4
|
add RTE and QNLI loader
|
6 years ago |
xuyige
|
bc5e071253
|
Delete matching.py
|
6 years ago |
xuyige
|
e0b23b16db
|
update data loader of matching
|
6 years ago |
xuyige
|
3593f0a545
|
fix bugs in matching dataloader
|
6 years ago |
xuyige
|
d1f531c049
|
update matching dataloader in reproduction/matching
|
6 years ago |
xuyige
|
93620e76ed
|
update framework of matching
|
6 years ago |
xuyige
|
39388567ad
|
update matching.py
|
6 years ago |
yh_cc
|
17b5fd0066
|
1. 删除Trainer中对train_data必须为DataSet的assert
2. 删除Trainer的prefetch参数; 在注释中增加num_workers参数
3. Trainer中默认sampler为RandomSampler
|
6 years ago |
xuyige
|
8e82c91751
|
update bert for nli in reproduction/matching
|
6 years ago |
xuyige
|
e643d7aed5
|
update reproduction/README.md
|
6 years ago |
xuyige
|
e05c182b05
|
firstly add matching in reproduction
|
6 years ago |