{
    "model_type": "diffusion_cond",
    "sample_size": 524288,
    "sample_rate": 44100,
    "audio_channels": 2,
    "model": {
        "pretransform": {
            "type": "autoencoder",
            "iterate_batch": false,
            "model_half": true,
            "config": {
                "encoder": {
                    "type": "oobleck",
                    "requires_grad": false,
                    "config": {
                        "in_channels": 2,
                        "channels": 128,
                        "c_mults": [1, 2, 4, 8, 16],
                        "strides": [2, 4, 4, 8, 8],
                        "latent_dim": 128,
                        "use_snake": true
                    }
                },
                "decoder": {
                    "type": "oobleck",
                    "config": {
                        "out_channels": 2,
                        "channels": 128,
                        "c_mults": [1, 2, 4, 8, 16],
                        "strides": [2, 4, 4, 8, 8],
                        "latent_dim": 64,
                        "use_snake": true,
                        "final_tanh": false
                    }
                },
                "bottleneck": {
                    "type": "vae"
                },
                "latent_dim": 64,
                "downsampling_ratio": 2048,
                "io_channels": 2
            }
        },
        "conditioning": {
            "configs": [
                {
                    "id": "prompt",
                    "type": "t5",
                    "config": {
                        "t5_model_name": "t5-base",
                        "max_length": 64
                    }
                },
                {
                    "id": "seconds_total",
                    "type": "number",
                    "config": {
                        "min_val": 0,
                        "max_val": 256
                    }
                }
            ],
            "cond_dim": 768
        },
        "diffusion": {
            "cross_attention_cond_ids": ["prompt", "seconds_total"],
            "global_cond_ids": ["seconds_total"],
            "diffusion_objective": "rf_denoiser",
            "type": "dit",
            "config": {
                "io_channels": 64,
                "embed_dim": 1024,
                "depth": 16,
                "num_heads": 8,
                "cond_token_dim": 768,
                "global_cond_dim": 768,
                "transformer_type": "continuous_transformer",
                "attn_kwargs": {
                    "qk_norm": "ln"
                }
            }
        },
        "io_channels": 64
    },
    "training": {
        "use_ema": true,
        "pre_encoded": false,
        "log_loss_info": false,
        "timestep_sampler": "trunc_logit_normal",
        "clip_grad_norm": 1.0,
        "cfg_dropout_prob": 0.0,
        "arc": {
            "ode_warmup": {
                "warmup_steps": 0,
                "refresh_rate": 10,
                "sampling_steps": 25,
                "cfg": 4
            },
            "noise_dist": {
                "generator": "logsnr_uniform",
                "discriminator": "logit_normal"
            },
            "use_model_as_discriminator": true,
            "discriminator_base_ckpt": "/path/to/base/rf/model",
            "discriminator": {
                "type": "convnext",
                "dit_hidden_layer": 12,
                "weights": {
                    "generator": 1.0,
                    "discriminator": 1.0
                },
                "loss_type": "relativistic",
                "config": {
                    "channels": 128,
                    "strides": [1, 2, 2],
                    "c_mults": [1, 2, 4],
                    "num_blocks": [2, 2, 2]
                },
                "contrastive": true,
                "include_grad_penalties": true
            }
        },
        "optimizer_configs": {
            "diffusion": {
                "optimizer": {
                    "type": "AdamW",
                    "config": {
                        "lr": 1e-6,
                        "betas": [0.9, 0.95],
                        "eps": 1e-8,
                        "weight_decay": 0.01,
                        "foreach": true
                    }
                }
            },
            "discriminator": {
                "optimizer": {
                    "type": "AdamW",
                    "config": {
                        "lr": 1e-6,
                        "betas": [0.9, 0.95],
                        "eps": 1e-8,
                        "weight_decay": 0.01,
                        "foreach": true
                    }
                }
            }
        },
        "demo": {
            "demo_every": 250,
            "demo_steps": 8,
            "num_demos": 8,
            "demo_cond": [
                {"prompt": "drum breaks 174 BPM", "seconds_total": 6},
                {"prompt": "People talking in a crowded cafe", "seconds_total": 10},
                {"prompt": "A short, beautiful piano riff in C minor", "seconds_total": 6},
                {"prompt": "Tight Snare Drum", "seconds_total": 1},
                {"prompt": "A dog barking next to a waterfall", "seconds_total": 6},
                {"prompt": "Glitchy bass design", "seconds_total": 4},
                {"prompt": "Synth pluck arp with reverb and delay, 128 BPM", "seconds_total": 6},
                {"prompt": "Birds singing in the forest", "seconds_total": 10}
            ],
            "demo_cfg_scales": [1]
        }
    }
}