Skip to content

Issues: QwenLM/Qwen2.5

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

46 Open 842 Closed
46 Open 842 Closed
Author
Filter by author
Loading
Label
Filter by label
Loading
Use alt + click/return to exclude labels
or ⇧ + click/return for logical OR
Projects
Filter by project
Loading
Milestones
Filter by milestone
Loading
Assignee
Filter by who’s assigned
Sort
Sort by
Newest Oldest Most commented Least commented Recently updated Least recently updated Best match
Most reactions
👍 👎 😄 🎉 😕 ❤️ 🚀 👀

Issues list

[Badcase]: Qwen 2.5:72b-instruct 在百炼平台或其他线上平台的表现比本地部署的 Ollama 模型更好，当输入字数过多时，ollama本地部署的模型会出现答非所问的情况。请问如何解决这个问题？
#1137 opened Dec 19, 2024 by thhbbx
4 tasks done
2
[Badcase]: Tool calling response inconsistent with tool definition
#1136 opened Dec 19, 2024 by ATAKing1023
4 tasks done
@tuhahaha
[Bug]: 无法提取知识图谱的实体和关系
#1135 opened Dec 18, 2024 by LuWei6896
4 tasks done
1
[Badcase]: qwen2-72B在H800 fp8量化下，输出json会有概率多一个转义符
#1132 opened Dec 17, 2024 by alanshao2
4 tasks done
1
[Badcase]: Qwen2.5对于一些提示词的输出全都是感叹号
#1128 opened Dec 12, 2024 by Ivy233
4 tasks done
1
[QwQ-Preview]: 微调之后合并lora权重之后的回答会泄露Human和Assistant的内容不知道为什么
#1126 opened Dec 9, 2024 by 123yxh
@yubowen-ph
[QwQ-Preview]: LiveCodeBench复现
#1121 opened Dec 5, 2024 by mattgithub1919
@huybery
[Bug]: Poor support for Hebrew
#1114 opened Nov 30, 2024 by shohamyamin
4 tasks done
@yangapku
1
您好，为什么QwQ-32B-Preview模型测试GPQA数据集时，模型会回答中文，而且模型回答好像停止不了，一直回答到token限制为止
#1111 opened Nov 28, 2024 by 13416157913
4 tasks done
@liudayiheng
[Badcase]: Twenty-four game, how to get 24 from 23, 3, 11, 16 Not possible with the given constraints
#1109 opened Nov 28, 2024 by novohool
4 tasks done
@liudayiheng
[Bug]: Qwen2.5-32b-int4用vllm跑 好像只会生成感叹号 duplicate This issue or pull request already exists help wanted Extra attention is needed
#1103 opened Nov 26, 2024 by ciaoyizhen
4 tasks
4
[Badcase]: Qwen2.5-7B-Instruct 中德、中意翻译 不遵循指令、code switch enhancement New feature or request
#1097 opened Nov 23, 2024 by pio57019
4 tasks done
@yangapku
[Badcase]: Generate json format error
#1095 opened Nov 20, 2024 by jxrjlxc02
4 tasks done
2
[Bug]: 使用vllm进行推理时，设置parallel_tool_calls似乎不生效。我想实现单工具调用，应该怎么设置呢
#1092 opened Nov 19, 2024 by RayneSun
4 tasks done
5
[Bug]: vllm infer Qwen2.5-32B-Instruct-AWQ with 2 * Nvidia-L20, output repeat !!!! duplicate This issue or pull request already exists help wanted Extra attention is needed
#1090 opened Nov 18, 2024 by RoyaltyLJW
4 tasks done
17
[Badcase]: 抢占式实例部署qwen2.5-72B成功，调用失败
#1088 opened Nov 18, 2024 by ZX1998-12
4 tasks done
1
您好，请问一下在Qwen2.5的发布榜单中Arena-Hard评测中，裁判模型是使用哪个？基线模型是使用哪个？
#1086 opened Nov 18, 2024 by 13416157913
4 tasks done
@hzhwcmhf
[Badcase]: 使用Reasoning数据微调2.0成功，但是2.5失败
#1085 opened Nov 18, 2024 by UESTCthb
4 tasks done
1
[Badcase]: 使用llamafactory量化gptq时，使用2个GPU时出错
#1084 opened Nov 17, 2024 by czhcc
4 tasks done
1
您好，请问一下为什么Qwen2.5官方发布的榜单中72B和32B的LiveCodeBench分数跟LiveCodeBench作者公开72B和32B的分数差异这么大？
#1079 opened Nov 15, 2024 by 13416157913
4 tasks done
@huybery
3
[Bug]: qwen2.5 tools输出格式问题 enhancement New feature or request
#1072 opened Nov 12, 2024 by kwjlhh
4 tasks done
@tuhahaha
6
[Bug]: transformers[4.46.2] error in multi-gpu when training. inactive
#1070 opened Nov 11, 2024 by tanypullin
4 tasks done
2
[Bug]: The final reason why you will get a model that cannot stop generation when you fine-tune the Qwen2.5-7b-base use Lora and a non-<|endoftext|> token as eos_token. inactive
#1064 opened Nov 8, 2024 by hxs91
4 tasks done
10
[Bug]: 使用Qwen2.5-72B-Instruct-AWQ和Qwen2.5-32B-Instruct-AWQ官方给的演示例子，运行的时候提示：returned non-zero exit status 1 inactive
#1059 opened Nov 7, 2024 by SuSuStarSmile
4 tasks done
2
A simplified version of the inference code ?
#1057 opened Nov 6, 2024 by weizhenhuan
5 tasks done
2
ProTip! Find all open issues with in progress development work with linked:pr.