Skip to content

Issues: hiyouga/LLaMA-Factory

🚨FAQs | 常见问题🚨
#4614 opened Jun 28, 2024 by hiyouga
Open
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

256 Open 5,093 Closed
256 Open 5,093 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

Qwen2-VL-7B 图文微调数据与纯文本微调数据混合训练,loss跌为0 pending This problem is yet to be addressed
#6159 opened Nov 27, 2024 by VincentVanNF
1 task done
两台机器全参数微调Qwen2.5-14B-Instruct挂起不动 pending This problem is yet to be addressed
#6143 opened Nov 26, 2024 by zhaoxjmail
1 task done
1
qwen2_vl 用 lora 训练慢，有什么工具可以排查慢的原因，有 torch.profiler 可以用吗？ pending This problem is yet to be addressed
#6134 opened Nov 25, 2024 by enerai
1 task done
pt模式下，关于样本数量、训练步数的不一致的现象 pending This problem is yet to be addressed
#6133 opened Nov 25, 2024 by kascas
1 task done
2
httpx.ConnectError: [Errno 111] Connection refused pending This problem is yet to be addressed
#6119 opened Nov 23, 2024 by cuiMidAutumn
1 task done
/root/LLaMA-Factory/src/llamafactory/launcher.py FAILED pending This problem is yet to be addressed
#6118 opened Nov 23, 2024 by Evi233
1 task done
1
多机训练的训练速度和单机一样 pending This problem is yet to be addressed
#6111 opened Nov 22, 2024 by Wiselnn570
1 task done
1
再次求助。。wandb断点延伸曲线 pending This problem is yet to be addressed
#6110 opened Nov 22, 2024 by Saberlve
1 task done
求助：模型qwen2.5-7b-instruct全量sft的时候，训练过程中loss突然变为0。 pending This problem is yet to be addressed
#6109 opened Nov 22, 2024 by Chtholly1
1 task done
1
使用API调用奖励模型得到的奖励值与PPO过程中的奖励值差距巨大/The disparity between the reward values obtained from calling the reward model using the API and the reward values from the PPO process is huge pending This problem is yet to be addressed
#6100 opened Nov 21, 2024 by LuRenjias
1 task done
mllm 数据格式如果存在问题，如何设置忽略该样本 pending This problem is yet to be addressed
#6096 opened Nov 21, 2024 by DietDietDiet
1 task done
LoRA微调Qwen2-VL-2B时，Loss一直为0，grad_norm为nan pending This problem is yet to be addressed
#6092 opened Nov 20, 2024 by Tian-ye1214
1 task done
1
pip install -e ".[torch,metrics]"时，出现 ERROR: No matching distribution found for setuptools>=61.0 pending This problem is yet to be addressed
#6089 opened Nov 20, 2024 by sun1092469590
1 task done
2
BAdam算法finetune的迭代轮次和论文不是很符合 pending This problem is yet to be addressed
#6088 opened Nov 20, 2024 by PhzCode
1 task done
4
训练参数以及训练时间疑问求解 pending This problem is yet to be addressed
#6087 opened Nov 20, 2024 by Beyond0831
1 task done
Maybe memory leak leak occurs after evaluation when using enable_liger_kernel. pending This problem is yet to be addressed
#6085 opened Nov 20, 2024 by upskyy
1 task done
关于 llamafactory-cli train 和 torchrun 训练耗费时间以及效果均不同的疑惑 pending This problem is yet to be addressed
#6080 opened Nov 19, 2024 by Maydaytyh
1 task done
昇腾910B3 两机16卡 lora sft Qwen2-72b报OOM npu This problem is related to NPU devices pending This problem is yet to be addressed
#6074 opened Nov 19, 2024 by hangxu124
1 task done
RuntimeError: weight lm_head.weight does not exist，我用Qwen2进行dpo微调后，再调用模型报错 pending This problem is yet to be addressed
#6073 opened Nov 19, 2024 by dahaogewsh
1
无法在生成的 generated_predictions.jsonl 中保留额外字段并丢失 <image> 标记 pending This problem is yet to be addressed
#6070 opened Nov 19, 2024 by enerai
1 task done
function call 模型能支持流式输出吗？ pending This problem is yet to be addressed
#6063 opened Nov 18, 2024 by SafeCool
1 task done
无缘无故被kill掉了，大佬能帮忙看看吗 pending This problem is yet to be addressed
#6058 opened Nov 18, 2024 by 1615070057
1 task done
GTPQ量化Qwen2.5-32B-Instruct LORA微调后的版本报错：torch._C._LinAlgError: linalg.cholesky: The factorization could not be completed because the input is not positive-definite (the leading minor of order 25942 is not positive-definite). pending This problem is yet to be addressed
#6057 opened Nov 18, 2024 by camposs1979
1 task done
使用llamafactory-cli api启动llama3.1-70b-Instruct模型，在dify中应用，模型失去function call能力 pending This problem is yet to be addressed
#6056 opened Nov 17, 2024 by limusen75
1 task done
1
量化qwen2.5-32b时出错，但7b没问题 pending This problem is yet to be addressed
#6048 opened Nov 16, 2024 by czhcc
1 task done
1
ProTip! Updated in the last three days: updated:>2024-11-24.