config.json

{
  "architectures": [
    "SiTMAEForPreTraining"
  ],
  "attention_probs_dropout_prob": 0.0,
  "attention_type": "eager",
  "decoder_hidden_size": 512,
  "decoder_intermediate_size": 2048,
  "decoder_num_attention_heads": 16,
  "decoder_num_hidden_layers": 8,
  "group_images": false,
  "hidden_act": "gelu",
  "hidden_dropout_prob": 0.0,
  "hidden_size": 768,
  "id2label": {},
  "initializer_range": 0.02,
  "intermediate_size": 3072,
  "label2id": {},
  "layer_norm_eps": 1e-12,
  "mask_ratio": 0.75,
  "max_seq_len": [
    6000,
    1
  ],
  "model_type": "sit_mae",
  "norm_pix_loss": true,
  "num_attention_heads": 12,
  "num_channels": 2,
  "num_hidden_layers": 12,
  "patch_size": [
    8,
    1
  ],
  "qkv_bias": true,
  "torch_dtype": "float32",
  "transformers_version": "4.44.2",
  "use_fs": true
}