AI Paper Atlas

Research map / Architectures and efficiency / Quantization

TWLA: Achieving Ternary Weights and Low-Bit Activations for LLMs via Post-Training Quantization

Zhixiong Zhao, Zukang Xu, Zhixuan Chen, Xing Hu, Zhe jiang, Dawei Yang

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

Read the paperCheck your idea against 137,217 papers

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
add E2M-ATQ (ablation)TWLA (LLaMA2-13B, A16) · C4C4 perplexity: 6e3 -> 18.32Improved
add KOTMS to E2M-ATQ (ablation)TWLA (LLaMA2-13B, A16) · C4C4 perplexity: 18.32 -> 8.64Improved
add E2M-ATQ to KOTMS (ablation)TWLA (LLaMA2-13B, A16) · C4C4 perplexity: 57.98 -> 8.64Improved
add KOTMS to E2M-ATQ (ablation)TWLA (LLaMA2-13B, A16) · MMLUMMLU accuracy: 30.15 -> 44.86Improved
add KOTMS to E2M-ATQ (ablation)TWLA (Qwen3-14B, A16) · MMLUMMLU accuracy: 57.64 -> 69.15Improved
add KOTMS to E2M-ATQ (ablation)TWLA (Qwen3-14B, A16) · C4C4 perplexity: 23.64 -> 17.26Improved
add KOTMS (no ILA-AMP) to E2M-ATQ at 4-bit activations (ablation)TWLA (LLaMA2-13B, A4) · C4C4 perplexity: 3e3 -> 25.03Improved
add ILA-AMP to E2M-ATQ + KOTMS (ablation)TWLA (LLaMA2-13B, A4) · C4C4 perplexity: 25.03 -> 10.07Improved
add ILA-AMP to E2M-ATQ + KOTMS (ablation)TWLA (LLaMA2-13B, A4) · MMLUMMLU accuracy: 27.52 -> 38.17Improved
add ILA-AMP to E2M-ATQ + KOTMS (ablation)TWLA (Qwen3-14B, A4) · MMLUMMLU accuracy: 47.60 -> 60.82Improved
add KOTMS to E2M-ATQ + ILA-AMP (ablation)TWLA (LLaMA2-13B, A4) · C4C4 perplexity: 55.86 -> 10.07Improved
use TWLA W1.58 instead of PT2-LLM W1.58LLaMA3-8B A16 · 7-task avg accuracyaccuracy: 39.04 -> 62.98Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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