preprocessor_config.json

{
  "auto_map": {
    "AutoImageProcessor": "image_preprocessing_molmo.MolmoImageProcessor",
    "AutoProcessor": "preprocessing_molmo.MolmoProcessor"
  },
  "base_image_input_size": [
    336,
    336
  ],
  "crop_patches": 24,
  "crop_size": 336,
  "crop_window_patches": 16,
  "crop_window_size": 224,
  "data_format": "channels_first",
  "default_to_square": true,
  "device": "cuda:2",
  "do_center_crop": null,
  "do_convert_rgb": true,
  "do_normalize": true,
  "do_pad": true,
  "do_rescale": true,
  "do_resize": true,
  "do_split_into_crops": true,
  "image_column_token": "<im_col>",
  "image_end_token": "<im_end>",
  "image_mean": [
    0.48145466,
    0.4578275,
    0.40821073
  ],
  "image_padding_mask": true,
  "image_patch_size": 14,
  "image_patch_token": "<im_patch>",
  "image_processor_type": "MolmoImageProcessorFast",
  "image_start_token": "<im_start>",
  "image_std": [
    0.26862954,
    0.26130258,
    0.27577711
  ],
  "image_token_length_h": 12,
  "image_token_length_w": 12,
  "input_data_format": null,
  "max_crops": 12,
  "max_num_crops": 12,
  "overlap_margins": [
    4,
    4
  ],
  "padding_mode": "constant",
  "padding_value": 1.0,
  "patches_per_image_height": 24,
  "patches_per_image_width": 24,
  "processor_class": "MolmoProcessor",
  "resample": 2,
  "rescale_factor": 0.00392156862745098,
  "return_tensors": null,
  "size": {
    "height": 336,
    "width": 336
  },
  "tokens_per_image": 144,
  "tokens_per_image_height": 12,
  "tokens_per_image_width": 12,
  "total_margin_pixels": 112
}