mlsanityData to 14 Oct 2026

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

arXiv 2606.13054 ↗First posted 11 Jun 2026cs.LG cs.AIcited by 8

On the map: Efficient LLM inference and serving → KV cache quantization

Abstract (arXiv)

Large language models (LLMs) exhibit exceptional general language processing capabilities, but their memory and compute costs hinder deployment. Ternarization has emerged as a promising compression technique, offering significant reductions in model size and inference complexity. However, existing methods struggle with heavy-tailed activation distributions and therefore keep activations in high precision, fundamentally limiting end-to-end inference acceleration. To overcome this limitation, we propose TWLA, a post-training quantization (PTQ) framework that achieves 1.58-bit weight compression and 4-bit activation quantization while maintaining high accuracy. TWLA comprises three components: (1) Euclidean-to-Manifold Asymmetric Ternary Quantizer (E2M-ATQ) minimizes layer-output error under weight ternarization via a two-stage optimization from Euclidean initialization to manifold relocation; (2) Kronecker Orthogonal Tri-Modal Shaping (KOTMS) applies a Kronecker-structured orthogonal rotation to reshape weights into ternary-friendly tri-modal distributions, while the shared rotation statistically suppresses activation outliers; and (3) Inter-Layer Aware Activation Mixed Precision (ILA-AMP) explicitly introduces adjacent-layer second-order interaction costs in bit allocation and jointly optimizes for the layer-wise disparity of activation quantization gains induced by the shared orthogonal transform, preventing cascades triggered by a few weak layers. Extensive experiments demonstrate that TWLA maintains high accuracy under W1.58A4, while delivering significant inference acceleration. The code is available at https://github.com/Kishon-zzx/TWLA.

Key facts. Cited by 8 papers in the map (data through 14 Oct 2026). Placed in KV cache quantization, one of 12 directions in Efficient LLM inference and serving. Builds on 32 of its 38 references that are in the map. First posted on arXiv 11 Jun 2026.

Builds on

32 of 38 references.

Closest ancestors · 1

Same direction, newest first

  1. 2024QuaRot: Outlier-Free 4-Bit Inference in Rotated LLMs30 Mar

Other references · 31

  1. 2019Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformercited by 14,599
  2. 2021Training Verifiers to Solve Math Word Problemscited by 10,109
  3. 2021Evaluating Large Language Models Trained on Codecited by 8,320
  4. 2020Measuring Massive Multitask Language Understandingcited by 7,963
  5. 2018Think you have Solved Question Answering? Try ARC, the AI2 Reasoning Challengecited by 4,886
  6. 2019HellaSwag: Can a Machine Really Finish Your Sentence?cited by 4,100
  7. 2016Pointer Sentinel Mixture Modelscited by 3,096
  8. 2019PIQA: Reasoning about Physical Commonsense in Natural Languagecited by 2,883
  9. 2022GPTQ: Accurate Post-Training Quantization for Generative Pre-trained Transformerscited by 1,919
  10. 2016The LAMBADA dataset: Word prediction requiring a broad discourse contextcited by 992
  11. 2023SpQR: A Sparse-Quantized Representation for Near-Lossless LLM Weight Compressioncited by 367
  12. 2016Trained Ternary Quantizationcited by 208
19 more
  1. 2020TernaryBERT: Distillation-aware Ultra-low Bit BERTcited by 148
  2. 2024SliM-LLM: Salience-Driven Mixed-Precision Quantization for Large Language Modelscited by 63
  3. 2024ResQ: Mixed-Precision Quantization of Large Language Models with Low-Rank Residualscited by 36
  4. 2024APTQ: Attention-aware Post-Training Mixed-Precision Quantization for Large Language Modelscited by 34
  5. 2016Ternary Neural Networks for Resource-Efficient AI Applicationscited by 33
  6. 2025Pathology-Aware Prototype Evolution via LLM-Driven Semantic Disambiguation for Multicenter Diabetic Retinopathy Diagnosiscited by 22
  7. 2025BitNet v2: Native 4-bit Activations with Hadamard Transformation for 1-bit LLMscited by 18
  8. 2024Towards Effective Data-Free Knowledge Distillation via Diverse Diffusion Augmentationcited by 13
  9. 2025QUARK: Quantization-Enabled Circuit Sharing for Transformer Acceleration by Exploiting Common Patterns in Nonlinear Operationscited by 11
  10. 2025SpecQuant: Spectral Decomposition and Adaptive Truncation for Ultra-Low-Bit LLMs Quantizationcited by 10
  11. 2024Adaptive Dataset Quantizationcited by 9
  12. 2025RobuQ: Pushing DiTs to W1.58A2 via Robust Activation Quantizationcited by 8
  13. 2026KBVQ-MoE: KLT-guided SVD with Bias-Corrected Vector Quantization for MoE Large Language Modelscited by 8
  14. 2025BiMaCoSR: Binary One-Step Diffusion Model Leveraging Flexible Matrix Compression for Real Super-Resolutioncited by 5
  15. 2026BWLA: Breaking the Barrier of W1AX Post-Training Quantization for LLMscited by 5
  16. 2025DBellQuant: Breaking the Bell with Double-Bell Transformation for LLMs Post Training Binarizationcited by 4
  17. 2025TreeQ: Pushing the Quantization Boundary of Diffusion Transformer via Tree-Structured Mixed-Precision Searchcited by 3
  18. 2026LoPRo: Enhancing Low-Rank Quantization via Permuted Block-Wise Rotationcited by 3
  19. 2026FLRQ: Faster LLM Quantization with Flexible Low-Rank Matrix Sketchingcited by 2

Built on by

Cited by year (2026 to 14 Oct)

Citing papers per year: 2026: 8.

8 papers.

  1. 2026Q-Strata: Hierarchical Bit Allocation for Mixed-Precision Quantization of Mixture-of-Experts LLMscited by 3
  2. 2026All for 1-Bit: Towards Genuine 1-Bit Post-Training Quantization for LLMscited by 2
  3. 2026Capability-Stratified Degradation in Ternary Language Modelscited by 1
  4. 2026Breaking the 1.58-bit Barrier for Ternary LLMscited by 1
  5. 2026Post-Training Ternarization of Qwen3-4B Capability, Effective Bit Budget, Storage Compression, and Deploymentcited by 0
  6. 2026Scaling Post-Training Ternarisation to Qwen3-8B Capability Retention, Reproduction, Lossless Packing, and Packed Executioncited by 0
  7. 2026JustQuant: You Don't Need Smoothing, SVD, or Rotation for 4-Bit Activation Quantizationcited by 0
  8. 2026LoopQuant: Accurate and Efficient Post-Training Quantization for Looped Transformerscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025PT2-LLM: Post-Training Ternarization for Large Language Models
  2. 2023FPTQ: Fine-grained Post-Training Quantization for Large Language Models
  3. 2024LRQ: Optimizing Post-Training Quantization for Large Language Models by Learning Low-Rank Weight-Scaling Matrices
  4. 2025Achieving binary weight and activation for LLMs using Post-Training Quantization
  5. 2023Norm Tweaking: High-performance Low-bit Quantization of Large Language Models
  6. 2025PTQTP: Post-Training Quantization to Trit-Planes for Large Language Models
  7. 2024Post Training Quantization of Large Language Models with Microscaling Formats
  8. 2024BiLLM: Pushing the Limit of Post-Training Quantization for LLMs