issues
search
ZhuiyiTechnology
/
t5-pegasus
中文生成式预训练模型
Apache License 2.0
557
stars
84
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
请问一下，dict在哪？
#43
MarsMeng1994
opened
1 year ago
0
tokenizer疑惑
#42
xucx666
opened
1 year ago
0
预训练模型https://open.zhuiyi.ai/releases/nlp/models/zhuiyi/chinese_t5_pegasus_base.zip更新了？
#41
zhaoying9105
opened
1 year ago
0
not Founderror:key encoder/final_layer_norm/scale not found in checkpoin
#40
tanshoudong
opened
1 year ago
2
有计划托管到kaggle上吗？
#39
BrandonStudio
opened
1 year ago
0
请问32g内存的pc机可以运行inference吗
#38
SeekPoint
opened
1 year ago
0
您好，请问有tf2.5.2，或者tf2以上的版本吗？
#37
tyn513
opened
1 year ago
0
关于“以mT5为基础架构和初始权重”的疑问
#36
andysdc
opened
2 years ago
0
您好，请问finetune后存在句子不完整的情况是怎么回事呢？
#35
Karenlyw
opened
2 years ago
0
你好，我将model的encoder和decoder分别保存为onnx，模型输出向量与原始模型输出向量不一致，为啥呢
#34
lazywangyuan
opened
2 years ago
0
请问如何将compute_loss中的tensor(比如y_true)转为array呢
#33
KyrieIrving24
opened
2 years ago
0
sentencepiece.model在哪里获取
#32
tangyaohua
opened
2 years ago
0
请问是只能在summary任务上finetune吗，在其他任务是应该改哪里呢
#31
KyrieIrving24
opened
2 years ago
0
请问有在线推理的示例代码吗
#30
codejeam
closed
2 years ago
0
KeyError: 'mt5.1.1'
#29
codejeam
closed
2 years ago
3
可以加上keras模型参数到checkpoint的转化代码吗？
#28
chenchenhaha
opened
2 years ago
0
bert4keras0.11版本以后需要修改加载方式
#27
lhy2749
opened
2 years ago
0
请问如何在finetune中使用多GPU训练?
#26
02hao09
closed
2 years ago
5
Tokenizer无法判断何时终止
#25
phenixace
opened
2 years ago
1
请问，推理速度如何提升呢？
#24
Dammy159753
opened
2 years ago
1
small版本加载不了
#23
areafather
opened
2 years ago
1
请问生成结果中，经常有最后几个短句循环重复的情况，是为什么？
#22
AltenLi
opened
2 years ago
5
t5-pegasus支持任意长度输入吗
#21
orlando1986
opened
2 years ago
1
请问推理怎么改造成batchsize>1的？
#20
geolvr
opened
3 years ago
3
缺少self.last_token函数
#19
yxwsfz
closed
3 years ago
0
生成样本个数
#18
ZTurboX
opened
3 years ago
1
用bert4keras做ner任务在选择T5模型，在加载该chinese_t5_pegasus_base预训练模型时报错
#17
sxk000
closed
3 years ago
1
小样本微调
#16
WEN-MIN
closed
3 years ago
0
语言支持
#15
jackie930
opened
3 years ago
1
请问能否提供参数配置呢
#14
Laurel-rao
opened
3 years ago
1
DataLossError: Checksum does not match
#13
hxyshare
closed
3 years ago
1
请问一下训练数据只有lcsts和csl吗？
#12
Jackthebighead
opened
3 years ago
1
finetune后保存的best_model.weights有1.2个G，测试时模型占了30个G，怎么会这么大呢……
#11
cxfzzj
opened
3 years ago
4
想改成输入多doc格式
#10
leisurehippo
closed
3 years ago
0
model.load_weights('./best_model.weights')报错'str' object has no attribute 'decode'
#9
nicolas-echo
opened
3 years ago
3
finetune总是OOM，有解决办法么
#8
AnddyWang
opened
3 years ago
3
BERT的结果是怎么得到的？
#7
demeiyan
opened
3 years ago
1
train
#6
mode007
opened
3 years ago
3
运行train.py的时候，最后到model.fit(dataset,steps_per_epoch=1000, ......)时候报错 AttributeError: 'DatasetV1Adapter' object has no attribute 'ndim'，
#5
mode007
closed
3 years ago
0
請問雖然該檔案不開源(f = '/root/data_pretrain/data_shuf.json')，但可否給少量的範例嗎?
#4
guan-yuan
opened
3 years ago
0
为什么你用3090能训练，我用3090都起不了tensorflow1.x的环境
#3
wxtt522
closed
3 years ago
0
推理脚本
#2
jackie930
closed
3 years ago
4
苏神，请问这个f = '/root/data_pretrain/data_shuf.json'文件有开源吗？
#1
PYMAQ
opened
3 years ago
1