| { |
| "_name_or_path": "laion/clap-htsat-fused", |
| "aff_block_r": 4, |
| "architectures": [ |
| "ClapAudioModelWithProjection" |
| ], |
| "attention_probs_dropout_prob": 0.0, |
| "depths": [ |
| 2, |
| 2, |
| 6, |
| 2 |
| ], |
| "drop_path_rate": 0.0, |
| "enable_fusion": true, |
| "enable_patch_fusion": true, |
| "enable_patch_layer_norm": true, |
| "flatten_patch_embeds": true, |
| "fusion_num_hidden_layers": 2, |
| "fusion_type": null, |
| "hidden_act": "gelu", |
| "hidden_dropout_prob": 0.1, |
| "hidden_size": 768, |
| "initializer_factor": 1.0, |
| "layer_norm_eps": 1e-05, |
| "mlp_ratio": 4.0, |
| "model_type": "clap_audio_model", |
| "num_attention_heads": [ |
| 4, |
| 8, |
| 16, |
| 32 |
| ], |
| "num_classes": 527, |
| "num_hidden_layers": 4, |
| "num_mel_bins": 64, |
| "patch_embed_input_channels": 1, |
| "patch_embeds_hidden_size": 96, |
| "patch_size": 4, |
| "patch_stride": [ |
| 4, |
| 4 |
| ], |
| "projection_dim": 512, |
| "projection_hidden_act": "relu", |
| "projection_hidden_size": 768, |
| "qkv_bias": true, |
| "spec_size": 256, |
| "torch_dtype": "float32", |
| "transformers_version": "4.40.1", |
| "window_size": 8 |
| } |
|
|