Replies: 2 comments 1 reply
-
Can you share the log file and training file please? What do you mean exactly by "the loss is not computed". |
Beta Was this translation helpful? Give feedback.
1 reply
-
I think it may be related to this issue. You may need to rename your labels in your train and test files to REF_energies and REF_forces, and use the flags --energy_key and --forces_key accordingly |
Beta Was this translation helpful? Give feedback.
0 replies
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
-
Dear all
I am trying to fine tune the "large" foundation model for my data. However, when I run "run_train.py" the loss function for the model is not computed.
I use the "foundation" branch and the below arguments:
log_dir = "logs"
model_dir = "."
checkpoints_dir = "checkpoints"
results_dir = "results"
name="MACE_0"
foundation_model="large"
train_file="train.xyz"
valid_fraction=0.05
test_file="test.xyz"
energy_weight=1.0
forces_weight=10.0
E0s="average"
lr=0.01
scaling="rms_forces_scaling"
batch_size=2
max_num_epochs = 150#1500
ema=true
ema_decay=0.99
amsgrad=true
default_dtype="float32"
device="cuda"
seed=3
Do you know why the loss is not calculated and evaluated for each batch?
Beta Was this translation helpful? Give feedback.
All reactions