| { | |
| "d_model": 408, | |
| "nhead": 8, | |
| "num_layers": 4, | |
| "dim_feedforward": 2048, | |
| "out_dim": 512, | |
| "dropout": 0.2, | |
| "activation": "gelu", | |
| "lr0": 0.00005, | |
| "warmup_steps": 1000, | |
| "projection_hidden_dim": 1024, | |
| "max_length": 512, | |
| "batch_size": 128 | |
| } | |
| { | |
| "d_model": 408, | |
| "nhead": 8, | |
| "num_layers": 4, | |
| "dim_feedforward": 2048, | |
| "out_dim": 512, | |
| "dropout": 0.2, | |
| "activation": "gelu", | |
| "lr0": 0.00005, | |
| "warmup_steps": 1000, | |
| "projection_hidden_dim": 1024, | |
| "max_length": 512, | |
| "batch_size": 128 | |
| } | |