config.json

{
  "_name_or_path": "MCG-NJU/videomae-base",
  "architectures": [
    "VideoMAEForVideoClassification"
  ],
  "attention_probs_dropout_prob": 0.0,
  "decoder_hidden_size": 384,
  "decoder_intermediate_size": 1536,
  "decoder_num_attention_heads": 6,
  "decoder_num_hidden_layers": 4,
  "hidden_act": "gelu",
  "hidden_dropout_prob": 0.0,
  "hidden_size": 768,
  "id2label": {
    "0": "drinking",
    "1": "eating",
    "2": "fastening_seat_belt",
    "3": "fetching_an_object",
    "4": "interacting_with_phone",
    "5": "opening_bottle",
    "6": "placing_an_object",
    "7": "putting_on_jacket",
    "8": "reading_magazine",
    "9": "reading_newspaper",
    "10": "sitting_still",
    "11": "taking_off_jacket",
    "12": "talking_on_phone",
    "13": "using_multimedia_display",
    "14": "working_on_laptop",
    "15": "writing"
  },
  "image_size": 224,
  "initializer_range": 0.02,
  "intermediate_size": 3072,
  "label2id": {
    "drinking": 0,
    "eating": 1,
    "fastening_seat_belt": 2,
    "fetching_an_object": 3,
    "interacting_with_phone": 4,
    "opening_bottle": 5,
    "placing_an_object": 6,
    "putting_on_jacket": 7,
    "reading_magazine": 8,
    "reading_newspaper": 9,
    "sitting_still": 10,
    "taking_off_jacket": 11,
    "talking_on_phone": 12,
    "using_multimedia_display": 13,
    "working_on_laptop": 14,
    "writing": 15
  },
  "layer_norm_eps": 1e-12,
  "model_type": "videomae",
  "norm_pix_loss": true,
  "num_attention_heads": 12,
  "num_channels": 3,
  "num_frames": 16,
  "num_hidden_layers": 12,
  "patch_size": 16,
  "problem_type": "single_label_classification",
  "qkv_bias": true,
  "torch_dtype": "float32",
  "transformers_version": "4.46.3",
  "tubelet_size": 2,
  "use_mean_pooling": false
}