{ | |
"vocab_size": 128000, | |
"dim": 512, | |
"num_layers": 6, | |
"num_heads": 16, | |
"max_recursion": 10, | |
"ffn_expansion": 8, | |
"max_position_embeddings": 1024, | |
"balancing_weight": 0.01, | |
"temperature": 1.0, | |
"window_size": 128, | |
"router_speed_weight": 0.1, | |
"model_speed_weight": 0.5, | |
"energy_weight": 0.001, | |
"d_energy_weight": 0.005, | |
"margin": 10.0, | |
"latent_dim": 256, | |
"model_type": "MoR-GAN", | |
"architecture": "MixtureOfRecursions-GAN", | |
"hidden_act": "gelu" | |
} |