| { | |
| "model_type": "dit_expert", | |
| "expert_id": 2, | |
| "architecture": "dit", | |
| "hidden_size": 1152, | |
| "num_layers": 28, | |
| "num_heads": 16, | |
| "patch_size": 2, | |
| "in_channels": 4, | |
| "out_channels": 4, | |
| "mlp_ratio": 4.0, | |
| "use_text_conditioning": true, | |
| "use_class_conditioning": false, | |
| "num_classes": 1000, | |
| "text_embed_dim": 768, | |
| "use_dit_time_embed": true, | |
| "use_adaln_single": true, | |
| "cfg_dropout_prob": 0.1 | |
| } |