transformers

4a19887f - tests: override prepare_video_inputs in llava_next_video to use tiny frames

tests: override prepare_video_inputs in llava_next_video to use tiny frames The base class creates (8, 3, 30, 400) frames which, after anyres tiling in LlavaNextVideo, produces ~100 MB of tensors in test_processor_text_has_no_visual. Override with (2, 3, 8, 8) frames to keep memory minimal. Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>