config.json

{
  "architectures": [
    "VJEPA2ForVideoClassification"
  ],
  "attention_dropout": 0.0,
  "attention_probs_dropout_prob": 0.0,
  "crop_size": 256,
  "drop_path_rate": 0.0,
  "frames_per_clip": 32,
  "hidden_act": "gelu",
  "hidden_dropout_prob": 0.0,
  "hidden_size": 1024,
  "id2label": {
    "0": "['Back', '15som', '05Twis', 'FREE']",
    "1": "['Back', '15som', '15Twis', 'FREE']",
    "2": "['Back', '15som', '25Twis', 'FREE']",
    "3": "['Back', '15som', 'NoTwis', 'PIKE']",
    "4": "['Back', '15som', 'NoTwis', 'TUCK']",
    "5": "['Back', '25som', '15Twis', 'PIKE']",
    "6": "['Back', '25som', '25Twis', 'PIKE']",
    "7": "['Back', '25som', 'NoTwis', 'PIKE']",
    "8": "['Back', '25som', 'NoTwis', 'TUCK']",
    "9": "['Back', '2som', '15Twis', 'FREE']",
    "10": "['Back', '2som', '25Twis', 'FREE']",
    "11": "['Back', '35som', 'NoTwis', 'PIKE']",
    "12": "['Back', '35som', 'NoTwis', 'TUCK']",
    "13": "['Back', '3som', 'NoTwis', 'PIKE']",
    "14": "['Back', '3som', 'NoTwis', 'TUCK']",
    "15": "['Back', 'Dive', 'NoTwis', 'PIKE']",
    "16": "['Back', 'Dive', 'NoTwis', 'TUCK']",
    "17": "['Forward', '15som', '1Twis', 'FREE']",
    "18": "['Forward', '15som', '2Twis', 'FREE']",
    "19": "['Forward', '15som', 'NoTwis', 'PIKE']",
    "20": "['Forward', '1som', 'NoTwis', 'PIKE']",
    "21": "['Forward', '25som', '1Twis', 'PIKE']",
    "22": "['Forward', '25som', '2Twis', 'PIKE']",
    "23": "['Forward', '25som', '3Twis', 'PIKE']",
    "24": "['Forward', '25som', 'NoTwis', 'PIKE']",
    "25": "['Forward', '25som', 'NoTwis', 'TUCK']",
    "26": "['Forward', '35som', 'NoTwis', 'PIKE']",
    "27": "['Forward', '35som', 'NoTwis', 'TUCK']",
    "28": "['Forward', '45som', 'NoTwis', 'TUCK']",
    "29": "['Forward', 'Dive', 'NoTwis', 'PIKE']",
    "30": "['Forward', 'Dive', 'NoTwis', 'STR']",
    "31": "['Inward', '15som', 'NoTwis', 'PIKE']",
    "32": "['Inward', '15som', 'NoTwis', 'TUCK']",
    "33": "['Inward', '25som', 'NoTwis', 'PIKE']",
    "34": "['Inward', '25som', 'NoTwis', 'TUCK']",
    "35": "['Inward', '35som', 'NoTwis', 'TUCK']",
    "36": "['Inward', 'Dive', 'NoTwis', 'PIKE']",
    "37": "['Reverse', '15som', '05Twis', 'FREE']",
    "38": "['Reverse', '15som', '15Twis', 'FREE']",
    "39": "['Reverse', '15som', '25Twis', 'FREE']",
    "40": "['Reverse', '15som', '35Twis', 'FREE']",
    "41": "['Reverse', '15som', 'NoTwis', 'PIKE']",
    "42": "['Reverse', '25som', '15Twis', 'PIKE']",
    "43": "['Reverse', '25som', 'NoTwis', 'PIKE']",
    "44": "['Reverse', '25som', 'NoTwis', 'TUCK']",
    "45": "['Reverse', '35som', 'NoTwis', 'TUCK']",
    "46": "['Reverse', 'Dive', 'NoTwis', 'PIKE']",
    "47": "['Reverse', 'Dive', 'NoTwis', 'TUCK']"
  },
  "image_size": 256,
  "in_chans": 3,
  "initializer_range": 0.02,
  "label2id": {
    "LABEL_0": 0,
    "LABEL_1": 1,
    "LABEL_10": 10,
    "LABEL_11": 11,
    "LABEL_12": 12,
    "LABEL_13": 13,
    "LABEL_14": 14,
    "LABEL_15": 15,
    "LABEL_16": 16,
    "LABEL_17": 17,
    "LABEL_18": 18,
    "LABEL_19": 19,
    "LABEL_2": 2,
    "LABEL_20": 20,
    "LABEL_21": 21,
    "LABEL_22": 22,
    "LABEL_23": 23,
    "LABEL_24": 24,
    "LABEL_25": 25,
    "LABEL_26": 26,
    "LABEL_27": 27,
    "LABEL_28": 28,
    "LABEL_29": 29,
    "LABEL_3": 3,
    "LABEL_30": 30,
    "LABEL_31": 31,
    "LABEL_32": 32,
    "LABEL_33": 33,
    "LABEL_34": 34,
    "LABEL_35": 35,
    "LABEL_36": 36,
    "LABEL_37": 37,
    "LABEL_38": 38,
    "LABEL_39": 39,
    "LABEL_4": 4,
    "LABEL_40": 40,
    "LABEL_41": 41,
    "LABEL_42": 42,
    "LABEL_43": 43,
    "LABEL_44": 44,
    "LABEL_45": 45,
    "LABEL_46": 46,
    "LABEL_47": 47,
    "LABEL_5": 5,
    "LABEL_6": 6,
    "LABEL_7": 7,
    "LABEL_8": 8,
    "LABEL_9": 9
  },
  "layer_norm_eps": 1e-06,
  "mlp_ratio": 4,
  "model_type": "vjepa2",
  "num_attention_heads": 16,
  "num_hidden_layers": 24,
  "num_pooler_layers": 3,
  "patch_size": 16,
  "pred_hidden_size": 384,
  "pred_mlp_ratio": 4.0,
  "pred_num_attention_heads": 12,
  "pred_num_hidden_layers": 12,
  "pred_num_mask_tokens": 10,
  "pred_zero_init_mask_tokens": true,
  "qkv_bias": true,
  "torch_dtype": "float32",
  "transformers_version": "4.53.0.dev0",
  "tubelet_size": 2,
  "use_SiLU": false,
  "wide_SiLU": true
}