config.json
| 1 | { |
| 2 | "architectures": [ |
| 3 | "VideoMAEForVideoClassification" |
| 4 | ], |
| 5 | "attention_probs_dropout_prob": 0.0, |
| 6 | "decoder_hidden_size": 384, |
| 7 | "decoder_intermediate_size": 1536, |
| 8 | "decoder_num_attention_heads": 6, |
| 9 | "decoder_num_hidden_layers": 4, |
| 10 | "hidden_act": "gelu", |
| 11 | "hidden_dropout_prob": 0.0, |
| 12 | "hidden_size": 768, |
| 13 | "image_size": 224, |
| 14 | "initializer_range": 0.02, |
| 15 | "intermediate_size": 3072, |
| 16 | "layer_norm_eps": 1e-12, |
| 17 | "model_type": "videomae", |
| 18 | "norm_pix_loss": true, |
| 19 | "num_attention_heads": 12, |
| 20 | "num_channels": 3, |
| 21 | "num_frames": 16, |
| 22 | "num_hidden_layers": 12, |
| 23 | "patch_size": 16, |
| 24 | "problem_type": "single_label_classification", |
| 25 | "qkv_bias": true, |
| 26 | "torch_dtype": "float32", |
| 27 | "transformers_version": "4.52.4", |
| 28 | "tubelet_size": 2, |
| 29 | "use_mean_pooling": false |
| 30 | } |
| 31 | |