| # Model hyperparameters | |
| learning_rate: 0.001 # The learning rate for the optimizer | |
| batch_size: 2 # Number of samples per batch | |
| test_batch_size: 8 # Number of samples per batch | |
| epochs: 40 # Total training epochs | |
| optimizer: "ranger" # Optimization algorithm | |
| dropout: 0.05 # Dropout regularization rate | |
| weight_decay: 0.0001 | |
| k: 5 | |
| ninp: 256 | |
| nlayers: 9 | |
| nclass: 2 | |
| ntoken: 5 #AUGC + padding/N token | |
| nhead: 8 | |
| use_bpp: False | |
| bpp_file_folder: "../../input/bpp_files/" | |
| gradient_accumulation_steps: 2 | |
| use_triangular_attention: false | |
| pairwise_dimension: 64 | |
| use_grad_checkpoint: true | |
| # Other configurations | |
| fold: 0 | |
| nfolds: 6 | |
| input_dir: "../../input/" | |
| gpu_id: "0" | |