config.json

{
  "architectures": [
    "M3DCLIP_stage1"
  ],
  "dropout_rate": 0.0,
  "gather_loss": true,
  "hidden_size": 768,
  "img_size": [
    32,
    256,
    256
  ],
  "in_channels": 1,
  "language_model_name_or_path": "/disk1/Data/Yanzhao/M3D_Model/bert-base-uncased",
  "local_loss": false,
  "mask_rate": 0.08,
  "max_text_len": 128,
  "mlp_dim": 3072,
  "model_type": "m3d_clip_stage1",
  "num_heads": 12,
  "num_layers": 12,
  "patch_size": [
    4,
    16,
    16
  ],
  "pos_embed": "perceptron",
  "pretrained_model": "/disk1/Data/Yanzhao/M3D_Model/M3D-CLIP",
  "spatial_dims": 3,
  "torch_dtype": "float32",
  "transformers_version": "4.49.0",
  "use_2D_Encoder": false,
  "use_mask": false,
  "version": "v0",
  "vocab_size": 30522
}