Conference
Open access
Sep 2026
HaRP-ViT: Hadamard Rotation for Low-Bit Vision Transformer Post-Training Quantization
HaRP-ViT (Hadamard Rotation for Post-Training Quantization of Vision Transformers), a training-free randomized Hadamard transform applied exclusively after the GELU activation is proposed, making the post-GELU tensor better suited to standard per-tensor uniform quantization.
Mustafa Ibrahim Aksu, F. Ş. Torun
· 2026 11th International Conf... · 0 citations