| { | |
| "architectures": [ | |
| "QaDiTModel" | |
| ], | |
| "auto_map": { | |
| "AutoConfig": "configuration_qadit.QaDiTConfig", | |
| "AutoModel": "modelling_qadit.QaDiTModel" | |
| }, | |
| "depth": 12, | |
| "dtype": "float32", | |
| "guidance_scale": 4.0, | |
| "hidden_size": 768, | |
| "latent_channels": 8, | |
| "latent_freq": 16, | |
| "latent_scale": 0.9503493000009796, | |
| "latent_time": 256, | |
| "load_auxiliaries": true, | |
| "logit_normal_mean": 0.0, | |
| "logit_normal_std": 1.0, | |
| "mlp_ratio": 4.0, | |
| "model_type": "qadit", | |
| "num_heads": 12, | |
| "num_inference_steps": 50, | |
| "num_train_timesteps": 1000, | |
| "patch_size": 2, | |
| "repa_layer": 4, | |
| "sample_rate": 16000, | |
| "schedule": "cosine", | |
| "text_dim": 1024, | |
| "text_max_length": 64, | |
| "text_model": "google/flan-t5-large", | |
| "transformers_version": "5.14.1", | |
| "vae_model": "cvssp/audioldm-s-full-v2", | |
| "vae_subfolder": "vae", | |
| "vocoder_model": "cvssp/audioldm-s-full-v2", | |
| "vocoder_subfolder": "vocoder" | |
| } | |