{ "model": { "vocab_size": 184704, "hidden_size": 2048, "intermediate_size": 6144, "num_hidden_layers": 28, "num_attention_heads": 16, "num_key_value_heads": 8, "head_dim": 128, "rope_theta": 1000000.0, "rms_norm_eps": 1e-06, "max_position_embeddings": 24576, "latent_dim": 64 }, "tokens": { "eod": 151643, "score_start": 151847, "score_end": 151848, "audio_start": 151851, "audio_end": 151852, "audio_offset": 151853, "audio_codes": 6561 }, "instruction": "Generate music with codec tokens from the given conditions.", "sampling": { "temperature": 0.9, "top_k": 25, "top_p": 0.8, "pause_guard": 40, "guard_distinct": 4, "decoder_steps": 32, "decoder_sampler": "dpm_2" }, "audio": { "sample_rate": 48000, "frame_rate": 25 } }