AutoEMatic introduces runtime-adaptive mixed-precision inference for LLMs
A new paper, AutoEMatic, proposes automatic exponent-mantissa refinement that adapts LLM quantization precision dynamically during inference, unlike existing post-training quantization methods that fix precision beforehand.
Read the full story at arXiv cs.AI ↗
Timeline · 1 report
- 2026-11-10 05:00 · arXiv cs.AI
AutoEMatic: Runtime-Adaptive Mixed-Precision LLM Inference via Automatic Exponent-Mantissa Refinement