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": "Birds",
    "1": "I dont know",
    "2": "address",
    "3": "angry",
    "4": "apple",
    "5": "bad",
    "6": "bathroom",
    "7": "beautiful",
    "8": "body",
    "9": "bowl",
    "10": "bread",
    "11": "call",
    "12": "christmas",
    "13": "clock",
    "14": "cold",
    "15": "date",
    "16": "die",
    "17": "dog",
    "18": "door",
    "19": "eat",
    "20": "evening",
    "21": "expensive",
    "22": "family",
    "23": "grow",
    "24": "handkerchief",
    "25": "please"
  },
  "image_size": 224,
  "initializer_range": 0.02,
  "intermediate_size": 3072,
  "label2id": {
    "Birds": 0,
    "I dont know": 1,
    "address": 2,
    "angry": 3,
    "apple": 4,
    "bad": 5,
    "bathroom": 6,
    "beautiful": 7,
    "body": 8,
    "bowl": 9,
    "bread": 10,
    "call": 11,
    "christmas": 12,
    "clock": 13,
    "cold": 14,
    "date": 15,
    "die": 16,
    "dog": 17,
    "door": 18,
    "eat": 19,
    "evening": 20,
    "expensive": 21,
    "family": 22,
    "grow": 23,
    "handkerchief": 24,
    "please": 25
  },
  "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": "4.38.1",
  "tubelet_size": 2,
  "use_mean_pooling": false
}