{ "d": 512, "vocab_size": 49152, "num_heads": 8, "head_dim": 64, "mlp_hidden_dim": 1408, "diff_blocks_num": 3, "layers_per_block": 6, "rope_theta": 50000.0, "max_position_embeddings": 4096, "mask_token_id": 49151, "max_insert": 4, "rms_eps": 1e-06, "attn_backend": "sdpa" }