transformers

c2abdef3 - Fix `base_model_prefix` in the PP-OCR detectors, CLIPSeg, TimesFM, CLIP and VideoPrism heads so base checkpoints load (#48871)

Fix `base_model_prefix` in the PP-OCR detectors, CLIPSeg, TimesFM, CLIP and VideoPrism heads so base checkpoints load (#48871) * Fix `base_model_prefix` in the PP-OCR detectors, CLIPSeg, TimesFM, CLIP and VideoPrism heads so base checkpoints load * Handle tuple entries of MODEL_MAPPING_NAMES in test_base_model_prefix_loads_base_weights `MODEL_MAPPING_NAMES["funnel"]` is a tuple, so the six Funnel heads were silently skipped by the string comparison; normalise the entry to a tuple of names and test membership. --------- Co-authored-by: Raushan Turganbay <raushan@huggingface.co>