transformers

d6d6dcc7 - tests: add _setup_image_processor to QianfanOCR with size=64 and max_patches=1 to reduce memory

tests: add _setup_image_processor to QianfanOCR with size=64 and max_patches=1 to reduce memory Default GotOcr2ImageProcessor uses 448×448 with max_patches=12, producing up to 27 MB pixel_values per image. Reducing to 64×64 with max_patches=1 cuts peak test RSS ~90 MB. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>