-
|
Hey Bora Define different model architecturesmodels=("DirectPred" "SupervisedVAE" "GNN") Define different data type combinationsdata_types=("mutation" "mutation,rna" "mutation,cnv") Define different fusion methods (only for models that support it)fusion_methods=("early" "intermediate") Dataset pathDATA_PATH="ccle_vs_gdsc" Iterate over all model and data type combinationsfor model in "${models[@]}"; do done echo "All experiments completed!" I see that it has 500 epoch which I think it's too much how can I reduce n. of epochs ? |
Beta Was this translation helpful? Give feedback.
Replies: 1 comment
-
|
You can decrease run times by setting a higher variance threshold (thus decrease number of considered features). |
Beta Was this translation helpful? Give feedback.
You can decrease run times by setting a higher variance threshold (thus decrease number of considered features).
You can set --early_stop_patience to a lower value.
You can pass a custom configuration file using --config_path argument. See example config file here: https://github.com/BIMSBbioinfo/flexynesis/blob/main/examples/configs/hpo_configuration.yaml