issues
search
wangyuxinwhy
/
uniem
unified embedding model
Apache License 2.0
814
stars
61
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
如何评测自己的模型
#83
TWJxi
opened
1 year ago
4
bump version
#82
wangyuxinwhy
closed
1 year ago
0
fix bug
#81
wangyuxinwhy
closed
1 year ago
0
生成embedding的速度比text2vec慢很多
#80
ralgond
opened
1 year ago
1
create_uniem_embedder是否是从头训练?
#79
gaoxiao
opened
1 year ago
3
Fix openai eval bug
#78
wangyuxinwhy
closed
1 year ago
0
怎么保存checkpoint呢
#77
gctian
closed
1 year ago
12
在T2Ranking 1W测试集上复现text-embedding-ada-002结果失败
#76
oasis-0927
closed
1 year ago
2
请问怎么指定特定的卡来训练
#75
ericg108
closed
1 year ago
12
相似度过高
#74
KekeWa
opened
1 year ago
3
关于训练，模型容易跑飞的情况
#73
qianzhang2018
closed
1 year ago
4
使用m3e模型去做相似度计算时，采用什么阈值来判断合适与否。
#72
caijijuhe
opened
1 year ago
3
请问如何微调无监督文本
#71
nieallen
opened
1 year ago
1
m3e distributed learning
#70
wangyuxinwhy
closed
1 year ago
0
你好，没有找到hfl/chinese-roberta-wwm-ext的small版，只找到了base和large，请问train_m3e.py是在这个roberta-wwm基础上训练得到的吗？
#69
huangjiaheng
opened
1 year ago
3
单机多卡跑train_m3e.py报错，
#68
heavenhellchen
opened
1 year ago
1
请问作者是否有兴趣做一个中文数据集的在线benchmark
#67
hjq133
closed
1 year ago
3
add accelerator config
#66
wangyuxinwhy
closed
1 year ago
0
关于使用accelerate启动多卡的问题
#65
qianzhang2018
closed
1 year ago
10
batch size 小的时候使用 `SigmoidContrastLoss` 会好一点，大的时候 `SoftmaxContrastLoss` 好一点？
#64
NLPJCL
opened
1 year ago
2
m3e支持交互式模型嘛？
#63
NLPJCL
closed
1 year ago
4
update accelerate
#62
wangyuxinwhy
closed
1 year ago
0
关于微调环境
#61
Galaxy-Ding
closed
1 year ago
7
sentenceTransformer在encode方法上面有什么优化吗
#60
CopyNinja1999
opened
1 year ago
3
关于py3.10
#59
char-con
closed
1 year ago
1
用正负样本三元组进行微调，遇到无法微调的问题
#58
KekeWa
opened
1 year ago
37
加载预训练模型的问题，是因为transformer版本问题吗
#57
KekeWa
closed
1 year ago
1
关于最长token数量的问题，question about maximum token size
#56
DarrenIm
closed
1 year ago
1
关于T2ranking的评测代码，没有在uniem里边找到。
#55
xjtulixiangyang
closed
1 year ago
1
T2ranking 取了前1万个doc具体的是怎么理解？
#54
xjtulixiangyang
closed
1 year ago
2
检索排序的指标问题
#53
Wenze7
closed
1 year ago
1
能否在华为npu 910a上微调和推理？
#52
tomjamescn
closed
1 year ago
1
onnx支持
#51
tomjamescn
closed
1 year ago
1
max_length 和 position embedding 相关
#50
hjq133
closed
1 year ago
3
T2Ranking检索任务的ground Truth（qrels）的处理是不是应该把 0 1 去掉呀
#49
Wenze7
closed
1 year ago
5
Add epoch end callback
#48
wangyuxinwhy
closed
1 year ago
0
🐞fix:兼容 torch1
#47
wangyuxinwhy
closed
1 year ago
0
📃docs:fix docs
#46
wangyuxinwhy
closed
1 year ago
0
Do M3E support multilingual？
#45
chaochaoSZ
closed
1 year ago
3
typo in source code
#44
habaneraa
closed
1 year ago
2
📃docs:fxi readme
#43
wangyuxinwhy
closed
1 year ago
0
Finetuner model class
#42
wangyuxinwhy
closed
1 year ago
0
很棒的效果！请教一下，我在复现的时候发现80G A100放不下80的batch_size,是怎么做到80的？另外train_m3e.py这个脚本采用的是PairInBatch，没有用到add_swap_loss。process_zh_datasets.py处理的数据也是Pair格式
#41
ARSblithe212
closed
1 year ago
11
文本分类中的acc是验证集上的结果还是测试集上的结果呢？
#40
graciechen
closed
1 year ago
3
关于数据集 是否已经添加过了？
#39
Galaxy-Ding
closed
1 year ago
10
关于评测测试集
#38
Nipi64310
closed
1 year ago
6
✨feat:minimax embedding
#37
wangyuxinwhy
closed
1 year ago
0
embedder 引入了 新的参数 model_class，但没有正确的在Finetuner中传递。
#36
FFengIll
closed
1 year ago
5
复现M3E-Base
#35
hjq133
closed
1 year ago
4
finetune时报错 argument after ** must be a mapping, not NoneType
#34
yhygta
closed
1 year ago
11
Previous
Next