Transformers
vision_asym_qwen3_vl_processor / preprocessor_config.json
TYTTYTTYT's picture
Vision-Asym Qwen3-VL processor: 2D R=4 align, visual_token_mask + keep_mask, selectable-text block padding
838c455 verified
Raw
History Blame Contribute Delete
1.08 kB
{
"auto_map": {
"AutoImageProcessor": "image_processing_qwen2_vl_fast.ZFQwen2VLImageProcessorFast",
"AutoProcessor": "processing_qwen3_vl.VisionAsymQwen3VLProcessor",
"AutoVideoProcessor": "video_processing_qwen3_vl.VisionAsymQwen3VLVideoProcessor"
},
"crop_size": null,
"data_format": "channels_first",
"default_to_square": true,
"device": null,
"disable_grouping": null,
"do_center_crop": null,
"do_convert_rgb": true,
"do_normalize": true,
"do_pad": null,
"do_rescale": true,
"do_resize": true,
"focus_size": 4,
"image_mean": [
0.5,
0.5,
0.5
],
"image_processor_type": "ZFQwen2VLImageProcessorFast",
"image_std": [
0.5,
0.5,
0.5
],
"input_data_format": null,
"max_pixels": null,
"merge_size": 2,
"min_pixels": null,
"pad_size": null,
"patch_size": 16,
"processor_class": "VisionAsymQwen3VLProcessor",
"resample": 3,
"rescale_factor": 0.00392156862745098,
"return_tensors": null,
"size": {
"longest_edge": 16777216,
"shortest_edge": 65536
},
"temporal_patch_size": 2
}