issues
search
InternLM
/
xtuner
An efficient, flexible and full-featured toolkit for fine-tuning LLM (InternLM2, Llama3, Phi3, Qwen, Mistral, ...)
https://xtuner.readthedocs.io/zh-cn/latest/
Apache License 2.0
3.8k
stars
302
forks
source link
issues
Newest
Newest
Most commented
Recently updated
Oldest
Least commented
Least recently updated
When seq_parallel_world_size is set to a value greater than 1, should use_varlen_attn not be set to true?
#938
Fovercon
opened
3 days ago
0
docker利用xtuner微调时，出错，不知道哪的问题？
#937
159357hou
opened
4 days ago
0
请问目前支持qwen2吗？
#936
Zheng-Jay
opened
6 days ago
1
AttributeError: 'Qwen2FlashAttention2' object has no attribute '_flash_attention_forward'
#935
zhangyuqi-1
opened
6 days ago
1
选择四卡训练卡住
#934
AlittlePIE
opened
1 week ago
1
intern2.5-20B微调 后词表长度不一致
#933
topology1
opened
2 weeks ago
0
使用lengthgroupedsampler代替原本的sampler后卡死
#932
xcy9614
opened
2 weeks ago
0
[Fix] Fix OOM when qlora converting
#931
fanqiNO1
opened
2 weeks ago
0
[Bugs] fix qlora convert bugs
#930
HIT-cwh
closed
1 day ago
0
如何进行val和test？
#929
Diyigelieren
opened
2 weeks ago
0
version `GLIBCXX_3.4.29' not found
#928
amannier
opened
2 weeks ago
0
Failed to inference single image using xtuner chat with llava-llama3-8b model
#927
J0eky
closed
2 weeks ago
1
奖励模型问题
#926
Eren139
opened
2 weeks ago
1
transformers == 4.44.2 xtuner == 0.1.23 训练 qwen2 时报错
#925
thomZ1
opened
3 weeks ago
1
多机多卡训练报错ss1.ss_family == ss2.ss_family. 2 vs 10
#924
sph116
opened
3 weeks ago
0
请问与 llamaFactory 的训练 TGS 对比时的具体实验条件
#923
shihanmax
opened
3 weeks ago
0
报错Cannot find reference 'VarlenAttnArgsToMessageHubHook' in 'init.py'
#922
hutiehua-1
opened
3 weeks ago
1
有个疑问，计算Loss的时候并不是以reward_token_id最终loss计算的，为什么推理的时候可以以reward_token_id为准呢？
#921
woshixiaobai2019
opened
4 weeks ago
6
QwenVL支持
#920
liyan1997
opened
4 weeks ago
0
整合Liger Kernel: 最高效的Triton Training Kernels
#919
ByronHsu
opened
1 month ago
0
一些关于步数统计的疑问
#918
young-chao
opened
1 month ago
0
add rescale sp loss
#917
HIT-cwh
opened
1 month ago
0
reward model训练完如何预测？
#916
tcxia
opened
1 month ago
1
qlora微调的模型是不支持中断后继续训练吗？
#915
deep-practice
opened
1 month ago
2
sharegpt4v数据集map错误
#914
bjzhb666
closed
1 month ago
1
InternVL构造单图多轮对话数据的时候，每轮对话都需要加上<image>标签吗？
#913
deep-practice
opened
1 month ago
1
【应当修改哪个环境变量？】Setting ds_accelerator to cuda (auto detect) df: "/home/guochenchen/.triton/autotune": 没有那个文件或目录
#912
gwoksansan
opened
1 month ago
0
如何修改 master port
#911
AislantVentus
opened
1 month ago
0
train time decrease from 13 hours to 9
#910
mylesgoose
opened
1 month ago
0
LLaVa phi-3 sft 报错 ConnectionResetError: [Errno 104] Connection reset by peer
#909
Yu-Yang-Li
opened
1 month ago
0
微调数据集策略（dataset make confuse）
#908
EasonQYS
opened
1 month ago
0
internvl微调的数据集一条有多个jsonl文件和多个图片该怎么写config
#907
mspythontu
opened
1 month ago
0
[Feature] Support balanced dataset to speed-up VL training
#906
yqyao
opened
1 month ago
2
How to modify the vision encoder of llava-llama3-8b?
#904
Jason8Kang
opened
1 month ago
0
fine-tuning codegeex4
#903
sgjohnson1981
opened
1 month ago
0
训练营3 XTuner运行xtuner train ./internlm2_chat_1_8b_qlora_alpaca_e3_copy.py 报错
#902
Viki-researcher
opened
1 month ago
2
Load failure with the converted finetune InternVL2-2B model
#901
leagend
closed
1 month ago
1
使用 xtuner convert pth_to_hf 会加载模型2次导致显存炸了，怎么解决
#900
c-x-l-w
opened
1 month ago
0
Error when doing sft training according to `https://xtuner.readthedocs.io/en/latest/get_started/quickstart.html#`
#899
YanShuang17
opened
1 month ago
1
如何实现多张卡共同存放单个模型
#898
RyanOvO
opened
1 month ago
1
之前理解错误，麻烦删除
#897
Hellcat1005
closed
1 month ago
0
Packer 好像没有没有分块 attention_mask
#896
WallE-Chang
closed
1 month ago
0
there is no script for gpt fintune
#895
zhenghuawang6
opened
1 month ago
0
长对话的微调训练
#894
RyanOvO
opened
1 month ago
0
怎么自己指定fp16 fp32 bp16？
#893
bjzhb666
closed
1 month ago
1
accumulative_counts起作用了吗？
#892
YixinSong-e
closed
1 month ago
0
多机训练的数据集问题
#891
YixinSong-e
closed
1 month ago
0
internlm2.py Boolean value of Tensor with more than one value is ambiguous
#890
doudoudiule
opened
1 month ago
2
Adjust the order of InternVL dataset log printing
#889
KooSung
opened
1 month ago
1
fix
#888
ArtificialZeng
opened
1 month ago
0
Next