Issues: hiyouga/LLaMA-Factory
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.
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
Author
Label
Projects
Milestones
Assignee
Sort
Issues list
Merging a FSDP+qlora checkpoint
pending
This problem is yet to be addressed.
#3573
opened May 4, 2024 by
rodrigo-f-nogueira
How to specify visible gpu in full sft/lora sft?
pending
This problem is yet to be addressed.
#3571
opened May 4, 2024 by
zhaoxu98
1 task done
使用 THUDM/chatglm3-6b“ 和默认数据集训练j时一直提示 "Failed"
pending
This problem is yet to be addressed.
#3569
opened May 4, 2024 by
cfanbo
1 task done
FSDP QLoRa
pending
This problem is yet to be addressed.
#3566
opened May 4, 2024 by
DinhLuan14
1 task done
unsloth/unsloth/Phi-3-mini-4k-instruct-bnb-4bit can't be trained.
pending
This problem is yet to be addressed.
#3565
opened May 3, 2024 by
iwaitu
1 task done
Output difference between LLaMA-Factory and llama.cpp
pending
This problem is yet to be addressed.
#3563
opened May 3, 2024 by
anidh
1 task done
IndexError: too many indices for tensor of dimension 2
#3560
opened May 3, 2024 by
heroding77
1 task done
DPO format - Expected a string, got {}".format(value), got None
pending
This problem is yet to be addressed.
#3555
opened May 3, 2024 by
Katehuuh
1 task done
FSDP QDoRa
pending
This problem is yet to be addressed.
#3550
opened May 2, 2024 by
etemiz
1 task done
How to convert Dolphin-2.9 to LLaMA factory?
solved
This problem has been already solved.
#3535
opened May 1, 2024 by
YixinSong-e
1 task done
多节点sft一直卡在这里,微调llama3 8b
pending
This problem is yet to be addressed.
#3534
opened May 1, 2024 by
gongye19
1 task done
DBRX using more gpu memory than mixtral 8x22B for fsdp+qlora
pending
This problem is yet to be addressed.
#3521
opened Apr 30, 2024 by
mces89
1 task done
Got error when exporting model with quantization
pending
This problem is yet to be addressed.
#3516
opened Apr 29, 2024 by
dickens88
1 task done
model.safetensor size changes in according to different finetuning methods
pending
This problem is yet to be addressed.
#3515
opened Apr 29, 2024 by
hunt-47
CUDA out of memory for fsdp training
pending
This problem is yet to be addressed.
#3494
opened Apr 28, 2024 by
v-yunbin
Llama-3-70B-Instruct使用example中的zero3.config训练,loss很大,输出混乱,有很多重复元素生成。同一套代码llama3 8b的则正常
pending
This problem is yet to be addressed.
#3492
opened Apr 28, 2024 by
fst813
1 task done
Why does it throw the following error when running on the Linux platform? httpx.RemoteProtocolError: Server disconnected without sending a response.
pending
This problem is yet to be addressed.
#3479
opened Apr 27, 2024 by
cuibh11
1 task done
cannot use pure_bf16 with zero3 cpu offload
pending
This problem is yet to be addressed.
#3476
opened Apr 27, 2024 by
mces89
1 task done
[Feature Request] 我们需要更灵活的保存策略?
pending
This problem is yet to be addressed.
#3472
opened Apr 26, 2024 by
marko1616
report to wandb能自动记录本项目里新增的参数么?例如stage、dataset、lora_rank、cutoff_len这些,暂时没看到有上报
enhancement
New feature or request
pending
This problem is yet to be addressed.
#3462
opened Apr 26, 2024 by
onebula
1 task done
deepspeed的bug
pending
This problem is yet to be addressed.
#3461
opened Apr 26, 2024 by
bravelyi
1 task done
Could you please share some tips with your rich experience?
pending
This problem is yet to be addressed.
#3452
opened Apr 25, 2024 by
xiaochengsky
1 task done
SFT zero2 zero3下loss不一致
pending
This problem is yet to be addressed.
#3442
opened Apr 25, 2024 by
wsdmanonymous
1 task done
Langchain didn't work when run src/api_demo.py Meta-Llama-3-8B-Instruct ,but chat.completions.create calling works fine.
pending
This problem is yet to be addressed.
#3421
opened Apr 24, 2024 by
hzgdeerHo
1 task done
量化后的gptq模型,部署成openai后调用报错
pending
This problem is yet to be addressed.
#3408
opened Apr 24, 2024 by
ccp123456789
Previous Next
ProTip!
Follow long discussions with comments:>50.