[2606.02823] Qift: Shift-Friendly No-Zero W2 Post-Training Quantization for Rotated W2A4/KV4 LLM Inference
![[2606.02823] Qift: Shift-Friendly No-Zero W2 Post-Training Quantization for Rotated W2A4/KV4 LLM Inference](/static/img/curated/desks/afbytes-us.png)
AI disclosure
Summary
Abstract page for arXiv paper 2606.02823: Qift: Shift-Friendly No-Zero W2 Post-Training Quantization for Rotated W2A4/KV4 LLM Inference