13 Commits

Author SHA1 Message Date
root
65e0e2570a 增加了qlora的finetune 2024-07-26 14:57:46 +08:00
刘丹
262840a805 修改了finetune中的默认模型的错误 2024-06-25 10:29:25 +08:00
root
f062357093 add autoawq example 2024-06-24 10:57:19 +08:00
root
b808010417 注释前加了空格,保持代码规范性,刘丹 2024-06-21 16:19:01 +08:00
root
062ea5264a 使用encode方法,代码可读性更强 2024-06-21 16:08:39 +08:00
root
8ae10c60ff 原始代码的usertoken是针对2b的,其他模型会有问题,现在根据不同模型都会调整 2024-06-21 15:31:24 +08:00
winter
8ebe696889 add fine-tuning model settings: bf16 and fp16 2024-03-25 15:49:41 +08:00
Xiang Long
36337f70ea Fix finetune supervised dataset issue 2024-03-16 01:58:08 +08:00
Xiang Long
74ecbcce5e Fix sft_dataset issue and naming error 2024-03-06 17:25:41 +08:00
SillyXu
c47f710b84
Update finetune.py 2024-02-08 10:41:37 +09:00
Xiang Long
24a71e964f Fix fine tune label offset bug 2024-02-01 12:51:30 +08:00
Xiang Long
a250c200ed Add README and update finetune scripts 2024-02-01 02:21:25 +08:00
Xiang Long
fb341897a6 Add fine tune scripts 2024-01-31 23:25:47 +08:00