{
  "_name_or_path": "MCG-NJU/videomae-base-finetuned-kinetics",
  "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": "tap dancing",
    "1": "breakdancing",
    "2": "belly dancing",
    "3": "dancing charleston",
    "4": "dancing ballet",
    "5": "square dancing",
    "6": "jumpstyle dancing",
    "7": "salsa dancing",
    "8": "robot dancing",
    "9": "country line dancing",
    "10": "dancing macarena",
    "11": "mosh pit dancing",
    "12": "dancing gangnam style",
    "13": "swing dancing",
    "14": "tango dancing"
  },
  "image_size": 224,
  "initializer_range": 0.02,
  "intermediate_size": 3072,
  "label2id": {
    "belly dancing": 2,
    "breakdancing": 1,
    "country line dancing": 9,
    "dancing ballet": 4,
    "dancing charleston": 3,
    "dancing gangnam style": 12,
    "dancing macarena": 10,
    "jumpstyle dancing": 6,
    "mosh pit dancing": 11,
    "robot dancing": 8,
    "salsa dancing": 7,
    "square dancing": 5,
    "swing dancing": 13,
    "tango dancing": 14,
    "tap dancing": 0
  },
  "layer_norm_eps": 1e-12,
  "model_type": "videomae",
  "norm_pix_loss": false,
  "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": "4.30.1",
  "tubelet_size": 2,
  "use_mean_pooling": true
}