-
Notifications
You must be signed in to change notification settings - Fork 1.4k
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
Loaded weights with error during training #784
Comments
some issue said that's the problem from GPU but is there anyway to avoid or bypass it? How did this problem arise?QAQ |
I encountered the same issue. May I ask if you have resolved it? You mentioned above that it might be a GPU-related problem. Could you please share what GPU you are using? I am currently using an A10 and facing this issue. |
I got same issue while fine tuning.. |
It's normal, our 1.5 model does not have lora weight, but it exists in model_dict, so if you use our model directly to train, it will show this warning, it's fine and will not cause any crash. |
And once you merge lora weight the warning will disappear. |
Self Checks
Cloud or Self Hosted
Self Hosted (Source)
Environment Details
WSL2, python3.10, all envs build just follow the introduction from document.
Steps to Reproduce
The strange warning displayed every time when running the train.py(or webui from manage.py) either 1.5 or 1.4.3 here.
I'm nor sure is it common situation during train process and some friends said that's definitely abnormal.
Fig1 from 1.5 and Fig2 from 1.4.3.
✔️ Expected Behavior
everthing goes well
❌ Actual Behavior
The model file can be created (normally?) and merged.
But the train speed is quite slow and the warning will be displayed again when merging.
Furthermore, the accuracy from Top5 hardly increase slightly and the inferenced voice by this strange model is also weird.
The text was updated successfully, but these errors were encountered: