{
  "_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": "Reject",
    "1": "Pass",
    "2": "Invisible",
    "3": "Absent",
    "4": "Torture",
    "5": "Permission",
    "6": "Dishonest",
    "7": "Insult",
    "8": "Complain",
    "9": "Hit",
    "10": "Save",
    "11": "Bring",
    "12": "Request",
    "13": "Start",
    "14": "Seperate",
    "15": "Talk",
    "16": "Turn off the light",
    "17": "Come",
    "18": "Lift",
    "19": "Advise",
    "20": "Encourage",
    "21": "Welding",
    "22": "Trouble",
    "23": "Crying",
    "24": "Whisper",
    "25": "Shave",
    "26": "Qurbani",
    "27": "Embrace",
    "28": "Dig",
    "29": "Making Noise",
    "30": "Bath",
    "31": "Hate",
    "32": "Sleep",
    "33": "Rotate",
    "34": "want",
    "35": "think",
    "36": "kiss",
    "37": "steal",
    "38": "try",
    "39": "tear"
  },
  "image_size": 224,
  "initializer_range": 0.02,
  "intermediate_size": 3072,
  "label2id": {
    "Absent": 3,
    "Advise": 19,
    "Bath": 30,
    "Bring": 11,
    "Come": 17,
    "Complain": 8,
    "Crying": 23,
    "Dig": 28,
    "Dishonest": 6,
    "Embrace": 27,
    "Encourage": 20,
    "Hate": 31,
    "Hit": 9,
    "Insult": 7,
    "Invisible": 2,
    "Lift": 18,
    "Making Noise": 29,
    "Pass": 1,
    "Permission": 5,
    "Qurbani": 26,
    "Reject": 0,
    "Request": 12,
    "Rotate": 33,
    "Save": 10,
    "Seperate": 14,
    "Shave": 25,
    "Sleep": 32,
    "Start": 13,
    "Talk": 15,
    "Torture": 4,
    "Trouble": 22,
    "Turn off the light": 16,
    "Welding": 21,
    "Whisper": 24,
    "kiss": 36,
    "steal": 37,
    "tear": 39,
    "think": 35,
    "try": 38,
    "want": 34
  },
  "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": "4.35.2",
  "tubelet_size": 2,
  "use_mean_pooling": false
}