{ "vocab_size": 256, "sequence_length": 1024, "local_dimension": 384, "local_layers": 4, "local_heads": 6, "local_intermediate_size": 1536, "dimension": 1024, "heads": 16, "intermediate_size": 12032, "route_paths": 3, "active_layers": 8, "max_patches": 192, "patch_rate": 0.125, "entropy_threshold": 4.125, "rope_theta": 10000.0, "patch_aux_weight": 0.25, "model_type": "cascade_byte_lm", "architectures": [ "CascadeForCausalLM" ] }