issues
search
wuba
/
qa_match
A simple effective ToolKit for short text matching
Other
328
stars
86
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
更新文档
#23
henryalps
closed
3 years ago
0
关于模型和词表
#22
ljcnh
opened
3 years ago
0
更新文档
#21
henryalps
closed
3 years ago
0
V1.3
#20
lvyuanyuan-ailab
closed
3 years ago
0
TF1.9 显示“cannot find reference 'flags' in '__init__.py'”
#19
StevenIIV
opened
3 years ago
1
词表问题
#18
lushizijizoude
opened
4 years ago
1
session.run这个编码一个句子需要0.14s,慢呢，用的cnn卷积编码，请问你们的速度有多少？貌似网上有提到session.run比较耗时
#17
liuchenbaidu
closed
4 years ago
1
如果只采用字向量，我达到的效果只有76%左右，所以为啥你们可以达到百分之0.83？词向量 是最好 是0.89或者说觉得你们效果好的原因是什么？训练多少epoch?
#16
liuchenbaidu
closed
4 years ago
0
dssm模型占用空间小,语料如果六七十万，模型会到170M多。
#15
liuchenbaidu
closed
4 years ago
3
qa_match/sptm/run_classifier.py的init_checkpoint根本没用到啊
#14
liiitleboy
closed
4 years ago
0
预训练模型
#13
ARSblithe212
opened
4 years ago
0
理解为一个句子分类问题，为什么不用textCNN,简单效果也挺好，而是用比较复杂dssm？
#12
liuchenbaidu
closed
4 years ago
1
如果标准问题非常多，例如十万标准问题，那么每次epoch，验证验证集非常慢。因为是十万类的分类问题。请问你们有用到十万标准问题的训练集吗？这么慢如何处理？
#11
liuchenbaidu
closed
4 years ago
4
预训练模型sptm中的use_queue含义没太懂
#10
wyqnumber
closed
4 years ago
1
'更新v1.1版本'
#9
henryalps
closed
4 years ago
0
models/dssm.py bilstm-dssm,为啥不用cdssm.lstm训练和推理速度都很慢
#8
liuchenbaidu
closed
4 years ago
3
更新V1.1版本
#7
henryalps
closed
4 years ago
0
如何使用公开的模型来测试效果？
#6
threefoldo
closed
4 years ago
1
'更新v1.1版本：完善文档'
#5
henryalps
closed
4 years ago
0
'更新v1.1版本'
#4
henryalps
closed
4 years ago
0
'更新v1.1版本'
#3
henryalps
closed
4 years ago
0
'添加预训练模型源码'
#2
henryalps
closed
4 years ago
0
领域分类问题
#1
zgd716
closed
4 years ago
1