[BUG] Can't do single task finetuning(not energy head) based on multitask pre-trained model in devel branch now. #3604
Assignees
Labels
Comments
github-merge-queue bot
pushed a commit
that referenced
this issue
Mar 26, 2024
fix #3604 , when doing single-task finetuning from multitask pretrained model and do not define the finetune model branch from command-line or input file.
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Bug summary
I found a bug when doing single task finetuning(not energy head) based on multitask pre-trained model, the reason I have found.
DeePMD-kit Version
devel branch
TensorFlow Version
not related
How did you download the software?
Offline packages
Input Files, Running Commands, Error Log, etc.
Bug in deepmd/pt/utils/finetune.py
change_finetune_model_paramslines 138--150. If I do single task finetuning(not energy head) based on multitask pre-trained model, "model_branch" is "", so finetune_links["Default"] is "". So in deepmd/pt/train/training.py line 551,_model_key_fromis "". So there is an error:Steps to Reproduce
Doing single task finetuning based on multitask pre-trained model. Do not specify a model branch.
Further Information, Files, and Links
No response
The text was updated successfully, but these errors were encountered: