mirror of
https://github.com/deepseek-ai/DeepSeek-Coder.git
synced 2026-11-05 00:57:19 +00:00
[GH-ISSUE #14] 这里微调脚本有什么特别处理吗?能给一下这里微调的的脚本细节吗? #10
Labels
No labels
documentation
ENV
pull-request
research question
WIP
No milestone
No project
No assignees
1 participant
Notifications
Due date
No due date set.
Dependencies
No dependencies set.
Reference
deepseek-ai/DeepSeek-Coder#10
Loading…
Add table
Add a link
Reference in a new issue
No description provided.
Delete branch "%!s()"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Originally created by @wyrcode on GitHub (Nov 6, 2023).
Original GitHub issue: https://github.com/deepseek-ai/DeepSeek-Coder/issues/14
@guoday commented on GitHub (Nov 7, 2023):
微调脚本没什么特别的处理,与大部分微调都一样,只算response的loss
@JianqiaoLu commented on GitHub (Nov 8, 2023):
你们是用的fastchat微调的嘛,好像跑起来之后lr一直是0,虽然我用了xformer,(不过我猜不是这个原因?
@zhangxjohn commented on GitHub (Nov 9, 2023):
我finetune的时候lr也是0, loss 不下降,没有用xformer,同求solution
@JianqiaoLu commented on GitHub (Nov 9, 2023):
有个离谱的事情,我发现你多跑几个epoch有机会出现lr...虽然也不正常。但是至少有了非0的lr. ;)
@zhangxjohn commented on GitHub (Nov 9, 2023):
是非常小趋于0且不变化是吗?依然不会收敛?
@wyrcode commented on GitHub (Nov 10, 2023):
我微调lr和loss是正常的。也是用的和fs差不多的脚本
@JianqiaoLu commented on GitHub (Nov 10, 2023):
心态崩了,为啥我train vicunav1.5正常,但是训练llama2就g了