Upload folder using huggingface_hub
Browse files- config.json +64 -0
config.json
ADDED
|
@@ -0,0 +1,64 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"model_type": "dasheng_audiogen",
|
| 3 |
+
"architectures": [
|
| 4 |
+
"DashengAudioGenModel"
|
| 5 |
+
],
|
| 6 |
+
"auto_map": {
|
| 7 |
+
"AutoConfig": "configuration_dasheng_audiogen.DashengAudioGenConfig",
|
| 8 |
+
"AutoModel": "modeling_dasheng_audiogen.DashengAudioGenModel"
|
| 9 |
+
},
|
| 10 |
+
"text_encoder_name": "google/mt5-large",
|
| 11 |
+
"tokenizer_name": "mispeech/dashengtokenizer",
|
| 12 |
+
"use_zero_instruction": true,
|
| 13 |
+
"instruction_seq_len": 1,
|
| 14 |
+
"task_instruction_dim": 1024,
|
| 15 |
+
"sample_rate": 16000,
|
| 16 |
+
"downsampling_ratio": 640,
|
| 17 |
+
"latent_dim": 1280,
|
| 18 |
+
"content_dim": 1024,
|
| 19 |
+
"frame_resolution": 0.005,
|
| 20 |
+
"duration_offset": 1.0,
|
| 21 |
+
"tokenizer_max_length": 512,
|
| 22 |
+
"dit_img_size": 1000,
|
| 23 |
+
"dit_patch_size": 1,
|
| 24 |
+
"dit_in_chans": 1280,
|
| 25 |
+
"dit_out_chans": 1280,
|
| 26 |
+
"dit_input_type": "1d",
|
| 27 |
+
"dit_embed_dim": 1536,
|
| 28 |
+
"dit_depth": 32,
|
| 29 |
+
"dit_num_heads": 24,
|
| 30 |
+
"dit_mlp_ratio": 4.0,
|
| 31 |
+
"dit_qk_norm": "layernorm",
|
| 32 |
+
"dit_norm_layer": "layernorm",
|
| 33 |
+
"dit_act_layer": "geglu",
|
| 34 |
+
"dit_context_norm": true,
|
| 35 |
+
"dit_time_fusion": "ada",
|
| 36 |
+
"dit_ada_sola_rank": 32,
|
| 37 |
+
"dit_ada_sola_alpha": 32,
|
| 38 |
+
"dit_ta_context_dim": 1024,
|
| 39 |
+
"dit_ta_context_fusion": "add",
|
| 40 |
+
"dit_ta_context_norm": true,
|
| 41 |
+
"dit_context_dim": 1024,
|
| 42 |
+
"dit_context_fusion": "cross",
|
| 43 |
+
"dit_context_pe_method": "none",
|
| 44 |
+
"dit_pe_method": "none",
|
| 45 |
+
"dit_rope_mode": "shared",
|
| 46 |
+
"adapter_num_heads": 16,
|
| 47 |
+
"adapter_dropout": 0.2,
|
| 48 |
+
"adapter_duration_grad_scale": 0.1,
|
| 49 |
+
"duration_predictor_filter_channels": 512,
|
| 50 |
+
"duration_predictor_n_layers": 5,
|
| 51 |
+
"duration_predictor_kernel_size": 3,
|
| 52 |
+
"duration_predictor_p_dropout": 0.5,
|
| 53 |
+
"special_tokens": [
|
| 54 |
+
"<|caption|>",
|
| 55 |
+
"<|speech|>",
|
| 56 |
+
"<|sfx|>",
|
| 57 |
+
"<|music|>",
|
| 58 |
+
"<|env|>",
|
| 59 |
+
"<|asr|>",
|
| 60 |
+
"<|speech_start|>",
|
| 61 |
+
"<|speech_end|>"
|
| 62 |
+
],
|
| 63 |
+
"train_special_tokens": true
|
| 64 |
+
}
|