transformers

4341b57c - tests: add _setup_video_processor to SmolVLM with max_image_size=64 to reduce memory

tests: add _setup_video_processor to SmolVLM with max_image_size=64 to reduce memory Video frame tensors default to 364x364 (=9.7 MB per frame batch). Reducing to 64x64 cuts video tensor allocations ~32x while leaving the image processor at 512 (required by test_process_interleaved_images_prompts_*). Video assertions only check frame counts and sign of mean, not spatial dims. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>