| { | |
| "model_type": "sls", | |
| "architectures": [ | |
| "SLS" | |
| ], | |
| "d_model": 1024, | |
| "nhead": 8, | |
| "num_layers": 8, | |
| "dim_feedforward": 2048, | |
| "use_flash": false, | |
| "num_tokens": 448, | |
| "noise": 1.0, | |
| "l2norm": false, | |
| "cls_mode": "last", | |
| "kl_loss": 0.2, | |
| "kl_target_scale": 1.0 | |
| } | |