| { | |
| "aggregator_dropout": 0.1, | |
| "aggregator_type": "multibranch", | |
| "calibrator_method": "spline", | |
| "d_feedforward": 512, | |
| "d_model": 256, | |
| "encoder_dropout": 0.1, | |
| "encoder_pooling": "cls", | |
| "encoder_type": "transformer", | |
| "head_dropout": 0.1, | |
| "head_hidden_dim": 128, | |
| "max_seq_len": 128, | |
| "n_attention_branches": 3, | |
| "n_difficulty_classes": 5, | |
| "n_encoder_layers": 4, | |
| "n_heads": 4, | |
| "star_ranges": { | |
| "0": [ | |
| 1, | |
| 5 | |
| ], | |
| "1": [ | |
| 1, | |
| 7 | |
| ], | |
| "2": [ | |
| 1, | |
| 8 | |
| ], | |
| "3": [ | |
| 1, | |
| 10 | |
| ], | |
| "4": [ | |
| 1, | |
| 10 | |
| ] | |
| }, | |
| "stochastic_mask_prob": 0.3, | |
| "top_k_ratio": 0.1 | |
| } |