AI Paper Atlas

Research map / Architectures and efficiency / Quantization

Float8@2bits: Entropy Coding Enables Data-Free Model Compression

Patrick Putzky, Martin Genzel, Mattes Mollenhauer, Sebastian Schulze, Thomas Wollmann, Stefan Dietzel

ICML 2026 · Architectures and efficiency › Quantization · approach: Post-training weight quantization · 0 citations (OpenAlex)

Read the paper

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use EntQuant instead of HQQ at 2 bitsLLaMA-2 70B · C4perplexity: 32.24 -> 6.47Improved
use EntQuant instead of HQQ at 3 bitsLLaMA-2 70B · C4perplexity: 6.02 -> 5.74Improved
use EntQuant (data-free) instead of calibration-based GPTQ at 2 bitsLLaMA-2 70B · zero-shot averageaccuracy: 34.4 -> 68.6Improved
use EntQuant instead of QuIP# at 2 bitsLLaMA-2 70B · zero-shot averageaccuracy: 70.9 -> 68.6Hurt
compress below 2 bits (1.7 bits) (ablation)EntQuant · LLaMA-3.1 8B C4perplexity: 9.74 (3-bit) -> 135.24 (1.7-bit)Hurt
use block-wise decompression instead of layer-wise decoding (ablation)EntQuant inference · inference speedspeed: approximately 50% speed-upImproved
exclude layers with super weights (ablation)EntQuant Int8 · LLaMA-2 7B C4perplexityImproved
exclude layers with super weights (ablation)EntQuant Float8 · LLaMA-2 7B C4perplexityImproved
exclude layers with super weights (ablation)HQQ at 2 bits · LLaMA-2 7B C4perplexityNo clear effect
use Int8 base format instead of Float8 (ablation)EntQuant · LLaMA-2 7B C4perplexityNo clear effect
add dynamic 8-bit activation quantization (W8A8) (ablation)EntQuant 3-bit · LLaMA-2 70B C4perplexity: 5.74 -> 7.35Hurt
add dynamic 8-bit activation quantization (W8A8) (ablation)EntQuant 3-bit · LLaMA-2 7B C4perplexity: 7.55 -> 7.76Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy