{
  "im_end": "<|image_end|>",
  "im_end_token": "<|image_end|>",
  "im_id_end": "<|image_id_end|>",
  "im_id_start": "<|image_id_start|>",
  "im_start": "<|image_start|>",
  "im_start_token": "<|image_start|>",
  "image_feature_size": 64,
  "image_processor_type": "MegrezOImageProcessor",
  "audio_feature_extractor_type": "WhisperFeatureExtractor",
  "feature_size": 128,
  "max_slice_nums": 9,
  "mean": [
    0.5,
    0.5,
    0.5
  ],
  "norm_mean": [
    0.5,
    0.5,
    0.5
  ],
  "norm_std": [
    0.5,
    0.5,
    0.5
  ],
  "patch_size": 14,
  "processor_class": "MegrezOProcessor",
  "sampling_rate": 16000,
  "scale_resolution": 448,
  "slice_end": "<|slice_end|>",
  "slice_end_token": "<|slice_end|>",
  "slice_mode": true,
  "slice_start": "<|slice_start|>",
  "slice_start_token": "<|slice_start|>",
  "std": [
    0.5,
    0.5,
    0.5
  ],
  "unk": "<|unk|>",
  "unk_token": "<|unk|>",
  "pad_token": "<|pad|>",
  "use_image_id": true
}