config.json

{
  "_name_or_path": "google/vivit-b-16x2-kinetics400",
  "architectures": [
    "VivitForVideoClassification"
  ],
  "attention_probs_dropout_prob": 0.0,
  "hidden_act": "gelu_fast",
  "hidden_dropout_prob": 0.0,
  "hidden_size": 768,
  "id2label": {
    "0": "Neutral",
    "1": "Anger",
    "2": "Disgust",
    "3": "Fear",
    "4": "Happiness",
    "5": "Sadness",
    "6": "Surprise",
    "7": "Other"
  },
  "image_size": 224,
  "initializer_range": 0.02,
  "intermediate_size": 3072,
  "label2id": {
    "Anger": "1",
    "Disgust": "2",
    "Fear": "3",
    "Happiness": "4",
    "Neutral": "0",
    "Other": "7",
    "Sadness": "5",
    "Surprise": "6"
  },
  "layer_norm_eps": 1e-06,
  "model_type": "vivit",
  "num_attention_heads": 12,
  "num_channels": 3,
  "num_frames": 3,
  "num_hidden_layers": 12,
  "qkv_bias": true,
  "torch_dtype": "float32",
  "transformers_version": "4.38.2",
  "tubelet_size": [
    2,
    16,
    16
  ],
  "video_size": [
    32,
    224,
    224
  ]
}