Native Binary-Weight Training Outperforms Zero-Shot Sign Projection in MNIST Diffusion
Does severe failure after projecting a diffusion model to one-bit weights show that its binary representation is unusable? We test that implication, rather than propose a new quantizer, in a small MNIST residual U-Net. We compare floating-point training, zero-shot per-channel sign/scale projection of a trained model, b...
Ayushman Bhattacharya, Anwesha Chakraborty, Nihal Gazi
· Zenodo (CERN European Organi... · 0 citations