issues
search
THUNLP-MT
/
THUMT
An open-source neural machine translation toolkit developed by Tsinghua Natural Language Processing Group
BSD 3-Clause "New" or "Revised" License
703
stars
197
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
how to fine tuning with pre_trained model
#119
WY19940327
opened
2 years ago
0
tensorflow版本target端为什么只在结束加eos,却没有在开始加bos。
#118
LJLQ
opened
2 years ago
0
Code Problem and Potential Solution: Inference with CPU
#117
Beau-xu
opened
2 years ago
0
Add encdec_attention cache to transformer.py to speed up inference.
#116
liushaokong
opened
2 years ago
0
一些疑惑
#115
leileilin
opened
2 years ago
0
请教问题
#114
leileilin
closed
2 years ago
0
报错:TypeError: Expected 'Iterator' as the return annotation for `__iter__` of Dataset, but found thumt.data.iterator.Iterator
#113
leileilin
closed
2 years ago
1
训练时没有生成eval文件夹,也没有在日志中输出验证信息
#112
edwardelric1202
closed
2 years ago
2
训练无响应无报错
#111
leileilin
closed
2 years ago
0
一些疑惑
#110
leileilin
closed
2 years ago
2
希望能出一份中文档
#109
leoFitz1024
closed
6 months ago
0
In dataset Wmt17 zh-en,The result is not good as wmt14 en-de
#108
QiyaoHuang
opened
3 years ago
2
Fix the lacking abstract method in dataset and typing error since pytorch 1.9
#107
aseaday
closed
2 years ago
0
Expected all tensors to be on the same device, but found at least two devices, cuda:0 and cpu!
#106
treeson-li
opened
3 years ago
1
Question about translating with CPU
#105
T2shen
opened
3 years ago
0
TypeError: Can't instantiate abstract class MapDataset with abstract methods _inputs, set_inputs
#104
zhuchenxi
opened
3 years ago
1
pytorch version ? Providing a bool or integral fill value without setting the optional `dtype` or `out` arguments is currently unsupported. In PyTorch 1.7,
#103
anbo724
opened
3 years ago
2
模型训练无法收敛
#102
baoyu-yuan
closed
3 years ago
0
translator.py生成了空的文档,程序无报错
#101
Linxia-MUC
opened
3 years ago
0
what is the hparams_set for benchmark transformer model?
#100
stevensgeek41
closed
3 years ago
1
Update walkthrough.md
#99
stevensgeek41
closed
3 years ago
0
get_relevance出现cast float to string报错
#98
fringe-k
opened
3 years ago
0
about the time for train a model
#97
Rooders
opened
4 years ago
5
training on wmt17 de-en, validation on wmt14 de-en 的bleu值在3w步之后一直维持在0.31附近
#96
wujsAct
closed
4 years ago
3
use cpu to inference
#95
qpzhao
opened
4 years ago
4
Different performance with RNNSearch vs. RNNSearch_LRP
#94
liushengzhong1023
closed
4 years ago
0
使用IWSLT17中-英数据集,在训练过程中BLEU持续升高,没有收敛的迹象,但模型在测试集上的泛化能力很差
#93
edwardelric1202
opened
4 years ago
1
wmt14 en-de
#92
hljjjmssyh
closed
4 years ago
0
如果中断后继续我上个检查点训练
#91
onoff888
opened
4 years ago
1
预训练模型
#90
duguiming111
opened
4 years ago
1
batch_size 10G GPU 单机最大能做到多少
#89
caoyuji1986
opened
4 years ago
1
你好,请问训练出现KeyError: '<unk>'是怎么回事
#88
edwardelric1202
closed
4 years ago
1
Can I output the translation every 1000 step?
#87
yinghy18
closed
4 years ago
1
multiple GPUs training with pytorch
#86
jennifer1995
closed
4 years ago
8
distributed training
#85
shawnkx
opened
4 years ago
0
如何正确的加入预训练的词向量
#84
orangefly0214
opened
4 years ago
2
Can you please share the pretrained model?
#83
DavidDavidsonDK
opened
4 years ago
1
Do you have an instruction manual for the pytorch version?
#82
Felixgithub2017
opened
4 years ago
2
Bugs in bin/scorer.py
#81
zhanghuimeng
closed
4 years ago
1
why you choose 7e-4 instead of 512 ** -0.5 as the learning rate
#80
shawnkx
closed
4 years ago
1
I do not find position_info_type in hyper parameter list in pytorch version thumt
#79
shawnkx
closed
4 years ago
1
MRT tends to deteriorate the performance while fine tuning a pre-trained Transformer.
#78
yongchanghao
closed
4 years ago
2
关于update_cycle
#77
ElliottYan
closed
4 years ago
0
UnicodeDecodeError: 'charmap' codec can't decode byte 0x81 in position 14:
#76
HassanNaeemjutt
closed
5 years ago
1
Update checkpoint_averaging.py
#75
Felixgithub2017
closed
3 years ago
0
checkpoint averaging error.
#74
Felixgithub2017
closed
5 years ago
1
fix typo
#73
alphadl
closed
5 years ago
1
What's the suggested loss_scale value?
#72
Felixgithub2017
closed
4 years ago
1
Has sb. trained the transformer model on WMT14 en-de and test on newstest2014?
#71
minorfox
closed
5 years ago
8
en2zh的实验,在decode阶段,出现空行
#70
wwy510553871
opened
5 years ago
18
Next