| model: | |
| hnet: | |
| arch_layout: ["c2", ["D12"], "c2"] | |
| d_model: [384, 384] | |
| num_heads: [6, 6] | |
| enc_dec_dim_factor: [4, null] | |
| fp32_router_dechunk: true | |
| training: | |
| multi_budget_training: true | |
| multi_budget_drop_fractions: [0.0, 0.1, 0.2, 0.3, 0.4, 0.5, 0.6] | |
| multi_budget_start_step: 50000 | |
| downsample_factor: 4.0 | |
| ratio_loss_weight: 0.03 | |
| data: | |
| feature_path: "../imagenet/features_train/" | |
| image_size: 256 | |