{ "model_type": "diffusion_cond", "sample_size": 441000, "sample_rate": 44100, "audio_channels": 2, "model": { "pretransform": { "type": "autoencoder", "iterate_batch": 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": "google/flan-t5-large", "max_length": 128 } }, { "id": "seconds_start", "type": "number", "config": { "min_val": 0, "max_val": 512 } }, { "id": "seconds_total", "type": "number", "config": { "min_val": 0, "max_val": 512 } } ], "cond_dim": 768 }, "diffusion": { "cross_attention_cond_ids": ["prompt", "seconds_start", "seconds_total"], "global_cond_ids": ["seconds_start", "seconds_total"], "type": "dit", "config": { "io_channels": 64, "embed_dim": 1536, "depth": 24, "num_heads": 24, "cond_token_dim": 768, "global_cond_dim": 1536, "project_cond_tokens": false, "transformer_type": "continuous_transformer" } }, "io_channels": 64 }, "training": { "use_ema": true, "log_loss_info": false, "optimizer_configs": { "diffusion": { "optimizer": { "type": "AdamW", "config": { "lr": 5e-5, "betas": [0.9, 0.999], "weight_decay": 1e-3 } }, "scheduler": { "type": "InverseLR", "config": { "inv_gamma": 1000000, "power": 0.5, "warmup": 0.99 } } } }, "demo": { "demo_every": 2000, "demo_steps": 250, "num_demos": 3, "demo_cond": [ {"prompt": "Amen break 174 BPM", "seconds_start": 0, "seconds_total": 10}, {"prompt": "Sound of a dog barking followed by a man speaking.", "seconds_start": 0, "seconds_total": 10}, {"prompt": "Sound of a busy metro station with a cat meowing.", "seconds_start": 0, "seconds_total": 10} ], "demo_cfg_scales": [5,7] } } }