{ "hidden_size": 768, "intermediate_size": 2112, "num_hidden_layers": 16, "num_attention_heads": 12, "num_key_value_heads": 6, "head_dim": 64, "vocab_size": 32000, "mask_vocab_size": 32001, "max_position_embeddings": 5120, "rope_theta": 10000.0, "rms_norm_eps": 1e-06, "hidden_act": "silu", "timestep_emb_hidden": 768, "mask_token_id": 32000, "pad_token_id": 1, "mask_ratio_min": 0.0, "mask_ratio_max": 1.0, "model_type": "metadiffusion", "architectures": [ "MetaDiffusionForCausalLM" ] }