Zineng Tang
		
	commited on
		
		
					Commit 
							
							·
						
						c7f8711
	
1
								Parent(s):
							
							5b3818c
								
Upload config.json
Browse files- config.json +43 -0
    	
        config.json
    ADDED
    
    | @@ -0,0 +1,43 @@ | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | |
|  | 
|  | |
| 1 | 
            +
            {
         | 
| 2 | 
            +
              "attention_probs_dropout_prob": 0.0,
         | 
| 3 | 
            +
              "audio_mask_ratio": 0.15,
         | 
| 4 | 
            +
              "audio_patch_size": [
         | 
| 5 | 
            +
                16,
         | 
| 6 | 
            +
                16
         | 
| 7 | 
            +
              ],
         | 
| 8 | 
            +
              "audio_size": [
         | 
| 9 | 
            +
                1024,
         | 
| 10 | 
            +
                128
         | 
| 11 | 
            +
              ],
         | 
| 12 | 
            +
              "decoder_hidden_size": 384,
         | 
| 13 | 
            +
              "decoder_intermediate_size": 1536,
         | 
| 14 | 
            +
              "decoder_num_attention_heads": 6,
         | 
| 15 | 
            +
              "decoder_num_hidden_layers": 4,
         | 
| 16 | 
            +
              "hidden_act": "gelu",
         | 
| 17 | 
            +
              "hidden_dropout_prob": 0.0,
         | 
| 18 | 
            +
              "hidden_size": 768,
         | 
| 19 | 
            +
              "id2label": {
         | 
| 20 | 
            +
                "0": "LABEL_0"
         | 
| 21 | 
            +
              },
         | 
| 22 | 
            +
              "image_size": 224,
         | 
| 23 | 
            +
              "initializer_range": 0.02,
         | 
| 24 | 
            +
              "intermediate_size": 3072,
         | 
| 25 | 
            +
              "label2id": {
         | 
| 26 | 
            +
                "LABEL_0": 0
         | 
| 27 | 
            +
              },
         | 
| 28 | 
            +
              "layer_norm_eps": 1e-12,
         | 
| 29 | 
            +
              "model_type": "tvlt",
         | 
| 30 | 
            +
              "norm_pix_loss": true,
         | 
| 31 | 
            +
              "num_attention_heads": 12,
         | 
| 32 | 
            +
              "num_audio_channels": 1,
         | 
| 33 | 
            +
              "num_channels": 3,
         | 
| 34 | 
            +
              "num_frames": 8,
         | 
| 35 | 
            +
              "num_hidden_layers": 12,
         | 
| 36 | 
            +
              "patch_size": 16,
         | 
| 37 | 
            +
              "qkv_bias": true,
         | 
| 38 | 
            +
              "task_mae": true,
         | 
| 39 | 
            +
              "task_matching": true,
         | 
| 40 | 
            +
              "transformers_version": "4.25.0.dev0",
         | 
| 41 | 
            +
              "use_mean_pooling": true,
         | 
| 42 | 
            +
              "video_mask_ratio": 0.75
         | 
| 43 | 
            +
            }
         |