Floating point exception
#38
by
zhongqy
- opened
when I use QWQ to interface，some error happened：
Loading checkpoint shards: 100%|████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████████| 17/17 [00:17<00:00, 1.05s/it]
Floating point exception
environment：
cuda：12.1
pytorch：2.4.0
How to solve it？
hello，I even get the promble, and then I change model loading type to torch.float32，it can be work.
hello，I even get the promble, and then I change model loading type to torch.float32，it can be work.
In my trying, it's work for other local model but not on QWQ, so i try to find other way. Still thank you!