Pith. sign in

REVIEW 35 cited by

QuaRot: Outlier-Free 4-Bit Inference in Rotated LLMs

Not yet reviewed by Pith; the record is open.

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2404.00456 v2 pith:WABLRAOR submitted 2024-03-30 cs.LG

QuaRot: Outlier-Free 4-Bit Inference in Rotated LLMs

classification cs.LG
keywords quarotllmsquantizationwithoutactivationsbitscachehidden
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved
0 comments
Share X Bluesky LinkedIn Reddit HN
read the original abstract

We introduce QuaRot, a new Quantization scheme based on Rotations, which is able to quantize LLMs end-to-end, including all weights, activations, and KV cache in 4 bits. QuaRot rotates LLMs in a way that removes outliers from the hidden state without changing the output, making quantization easier. This computational invariance is applied to the hidden state (residual) of the LLM, as well as to the activations of the feed-forward components, aspects of the attention mechanism, and to the KV cache. The result is a quantized model where all matrix multiplications are performed in 4 bits, without any channels identified for retention in higher precision. Our 4-bit quantized LLaMa2-70B model has losses of at most 0.47 WikiText-2 perplexity and retains 99% of the zero-shot performance. We also show that QuaRot can provide lossless 6 and 8 bit LLaMa2 models without any calibration data using round-to-nearest quantization. Code is available at: https://github.com/spcl/QuaRot.

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Forward citations

Cited by 35 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. HeadQ: Model-Visible Distortion and Score-Space Correction for KV-Cache Quantization

    cs.LG 2026-05 conditional novelty 8.0

    HeadQ removes 84-94% of excess perplexity from 2-bit key quantization by storing low-rank residuals in a calibration-learned query basis for score-space correction and using A²-weighted distortion for values.

  2. MOSAIC: Masked Outsourcing of Secure AI Computations

    cs.CR 2026-07 conditional novelty 7.0

    A secure approximate matrix-multiplication outsourcing protocol with O((m+n)l) client overhead that keeps 70B transformer perplexity within ~1% of BF16 at sigma=0.5 under claimed 140-bit LWE/LPN parameters.

  3. The Speedup Paradox: Rethinking Inference Speed-Quality Trade-off in Embodied Tasks

    cs.RO 2026-06 unverdicted novelty 7.0

    TISED framework reveals paradoxical effects where inference optimizations can lengthen task completion time on static tasks or raise success rates on dynamic tasks in embodied AI.

  4. STAR-KV: Low-Rank KV Cache Compression via Soft Thresholding for Adaptive Rank Control

    cs.LG 2026-06 unverdicted novelty 7.0

    STAR-KV applies differentiable soft thresholding for per-head and per-block adaptive low-rank KV cache compression, combined with hybrid decomposition and low-rank-aware quantization, achieving up to 75% compression a...

  5. {\Omega}-QVLA: Robust Quantization for Vision-Language-Action Models via Composite Rotation and Per-step Scaling

    cs.CV 2026-05 unverdicted novelty 7.0

    Omega-QVLA is a post-training quantization framework achieving uniform W4A4 for VLA models' LLM backbone and DiT action head via composite SVD-Hadamard rotation and per-step scaling, matching FP16 success rates on LIBERO.

  6. When Quantization Is Free: An int4 KV Cache That Outruns fp16 on Apple Silicon

    cs.PF 2026-05 unverdicted novelty 7.0

    A single fused int4 KV cache kernel on Apple Silicon outperforms fp16 in latency with 3x memory compression and near-zero quality loss on tested models.

  7. QuantVLA: Scale-Calibrated Post-Training Quantization for Vision-Language-Action Models

    cs.LG 2026-02 unverdicted novelty 7.0

    QuantVLA is the first post-training quantization framework for VLA models that quantizes the diffusion transformer action head and reports higher task success rates than full-precision baselines with roughly 70% memor...

  8. Four Over Six: More Accurate NVFP4 Quantization with Adaptive Block Scaling

    cs.CL 2025-12 conditional novelty 7.0

    Four Over Six adaptively scales blocks in NVFP4 quantization to smaller FP4 values, making representable value distributions more uniform and reducing quantization error especially for near-maximal values.

  9. LightRot: A Light-Weighted Rotation Scheme and Architecture for Accurate Low-Bit Large Language Model Inference

    cs.AR 2026-07 conditional novelty 6.0

    LightRot uses grouped local rotation plus outlier alignment to make 4-bit LLaMA inference accurate and cheap, claiming 27.4 TOPS/W on a 28nm accelerator.

  10. MXSens: Sensitivity-Aware Mixed-Precision Quantization for Efficient LLM Inference

    cs.LG 2026-07 conditional novelty 6.0

    MXSens allocates 8-bit precision to the 32 most sensitive columns per layer, 6-bit to moderately sensitive columns, and 4-bit elsewhere in MXINT, improving WikiText-2 perplexity over prior 4-bit LLM quantization methods.

  11. The Speedup Paradox: Rethinking Inference Speed-Quality Trade-off in Embodied Tasks

    cs.RO 2026-06 unverdicted novelty 6.0

    TISED decomposes inference optimization effects on embodied tasks and identifies paradoxical outcomes where faster per-step inference can increase task completion time on static tasks or raise success rates on dynamic tasks.

  12. GRINQH: Graded Input-based Quantization Hierarchy for Efficient LLM Generation

    cs.LG 2026-06 unverdicted novelty 6.0

    GRINQH introduces a graded input-based quantization hierarchy that dynamically assigns multi-precision weights using activation magnitudes as importance proxy, unifying quantization with sparsification to improve LLM ...

  13. MonaVec: A Training-Free Embedded Vector Search Kernel for Edge and Offline AI Systems

    cs.IR 2026-06 unverdicted novelty 6.0

    MonaVec provides a training-free 4-bit vector quantization and deterministic search kernel using Randomized Hadamard Transform and ChaCha20 seeding for embedded and offline use.

  14. Qift: Shift-Friendly No-Zero W2 Post-Training Quantization for Rotated W2A4/KV4 LLM Inference

    cs.LG 2026-06 unverdicted novelty 6.0

    Qift defines a fixed no-zero W2 level set for rotated weights that improves W2A4 perplexity and accuracy on LLaMA-2-7B and LLaMA-3.1-8B over the standard {-2,-1,0,1} set.

  15. HeadQ: Model-Visible Distortion and Score-Space Correction for KV-Cache Quantization

    cs.LG 2026-05 unverdicted novelty 6.0

    HeadQ reduces 84-94% of excess perplexity in 2-bit key quantization by adding low-rank logit corrections in a calibration-learned query basis, with further gains from an A^2-weighted value policy.

  16. Technical Report: Activation Residual Hessian Quantization (ARHQ) for Low-Bit LLM Quantization

    cs.LG 2026-04 unverdicted novelty 6.0

    ARHQ isolates error-sensitive weight directions in LLMs via truncated SVD on the scaled matrix W G_x^{1/2} from activation residuals, improving SNR and preserving performance under aggressive low-bit quantization.

  17. MemExplorer: Navigating the Heterogeneous Memory Design Space for Agentic Inference NPUs

    cs.AR 2026-04 unverdicted novelty 6.0

    MemExplorer optimizes heterogeneous memory systems for agentic LLM inference on NPUs and reports up to 2.3x higher energy efficiency than baselines under fixed power budgets.

  18. Rethinking Residual Errors in Compensation-based LLM Quantization

    cs.LG 2026-04 conditional novelty 6.0

    Redefining residual errors to include compensation-aware discrepancies and realigning calibration to full-precision outputs improves GPTQ and GPTAQ performance on LLMs.

  19. Price of metric universality in vector quantization is at most 0.11 bit

    cs.IT 2026-02 conditional novelty 6.0

    A universal vector-quantization codebook exists that is within 0.11 bit/coordinate of covariance-adaptive waterfilling simultaneously for all input covariances, for Gaussian weights.

  20. Double: Breaking the Acceleration Limit via Double Retrieval Speculative Parallelism

    cs.CL 2026-01 unverdicted novelty 6.0

    Double achieves up to 5.3x inference speedup on 70B LLMs via synchronous double retrieval speculative parallelism that is lossless and outperforms trained baselines like EAGLE-3.

  21. You Had One Job: Per-Task Quantization Using LLMs' Hidden Representations

    cs.CL 2025-11 conditional novelty 6.0

    TAQ estimates per-layer importance from hidden representations and output sensitivity on task calibration data to allocate mixed precision in a training-free PTQ setting, outperforming task-agnostic baselines on accur...

  22. TurboQuant: Online Vector Quantization with Near-optimal Distortion Rate

    cs.LG 2025-04 unverdicted novelty 6.0

    TurboQuant achieves near-optimal vector quantization distortion for both MSE and inner products via random rotation and per-coordinate scalar quantization, with a formal proof that it matches lower bounds within a fac...

  23. Efficient Reasoning on the Edge

    cs.LG 2026-03 accept novelty 5.5

    LoRA adapters, budget-forced GRPO, dynamic switching, parallel verification and FPTQuant enable practical chain-of-thought reasoning on quantized Qwen2.5-7B for edge devices.

  24. SigmaScale: LLM Compression with SVD-based Low-Rank Decomposition and Learned Scaling Matrices

    cs.CL 2026-06 unverdicted novelty 5.0

    Learned diagonal scaling matrices optimized with activation-aware loss reduce effective rank in LLM weight matrices and yield competitive perplexity and zero-shot results versus prior SVD methods on Llama 3.1 8B and Qwen3-8B.

  25. QVGGT: Post-Training Quantized Visual Geometry Grounded Transformer

    cs.CV 2026-05 unverdicted novelty 5.0

    QVGGT uses per-block mixed-precision analysis, outlier token filtering with PCA compensation, and task-aware scale search to achieve near-lossless W4A16 quantization of VGGT with 3-4.9x memory savings and 2.8x speedup.

  26. 31.1 A 14.08-to-135.69Token/s ReRAM-on-Logic Stacked Outlier-Free Large-Language-Model Accelerator with Block-Clustered Weight-Compression and Adaptive Parallel-Speculative-Decoding

    cs.AR 2026-05 unverdicted novelty 5.0

    A ReRAM-on-logic stacked chip delivers 14.08-135.69 tokens/s LLM inference with block-clustered compression and adaptive parallel speculative decoding, yielding 4.46-7.17x speedup over standard methods.

  27. HeadQ: Model-Visible Distortion and Score-Space Correction for KV-Cache Quantization

    cs.LG 2026-05 unverdicted novelty 5.0

    HeadQ applies score-space logit corrections for keys and attention-weighted surrogates for values to KV-cache quantization, removing 84-94% of excess perplexity in 2-bit key experiments across six models.

  28. StreamIndex: Memory-Bounded Compressed Sparse Attention via Streaming Top-k

    cs.LG 2026-05 accept novelty 5.0

    Chunked streaming top-k enables CSA indexer execution at 1M sequence length with 6.21 GB peak memory and >=0.998 recall on synthetic V4-shaped inputs.

  29. High-Rate Quantized Matrix Multiplication I

    cs.IT 2026-01 unverdicted novelty 5.0

    High-rate quantization theory yields accurate approximations for the distortion of absmax INT and FP schemes in generic weight-plus-activation matrix multiplication.

  30. You Had One Job: Per-Task Quantization Using LLMs' Hidden Representations

    cs.CL 2025-11 reject novelty 5.0

    A per-layer, per-task bit-allocation rule based on hidden-activation entropy and variance reportedly matches full-precision TriviaQA accuracy within about one point — but the paper's tables and abstract are internally...

  31. Combating the Memory Walls: Optimization Pathways for Long-Context Agentic LLM Inference

    cs.AR 2025-09 unverdicted novelty 5.0

    PLENA introduces a co-designed system with three optimization pathways for long-context agentic LLM inference, claiming up to 2.23x throughput over A100 and 4.04x energy efficiency.

  32. MixLLM: LLM Quantization with Global Mixed-precision between Output-features and Highly-efficient System Design

    cs.LG 2024-12 unverdicted novelty 5.0

    MixLLM uses global output-feature importance to set mixed bit-widths for LLM quantization and adds two-step dequantization plus software pipelining for system efficiency.

  33. Influence-Inspired Spectral Rotations for Extreme Low-Bit LLM Quantization

    cs.LG 2026-05 unverdicted novelty 4.0

    A WHT rotation plus per-coordinate activation-energy rescaling before auto-round quantization lowers WikiText-2 perplexity 15-58% versus vanilla auto-round at W2A16 on models from 135M to 1.5B parameters.

  34. DuQuant++: Fine-grained Rotation Enhances Microscaling FP4 Quantization

    cs.CV 2026-04 unverdicted novelty 4.0

    DuQuant++ adapts outlier-aware fine-grained rotation to MXFP4 by matching block size to the 32-element microscaling group, enabling a single rotation that smooths distributions and achieves SOTA performance on LLaMA-3...

  35. A Survey on Efficient Inference for Large Language Models

    cs.CL 2024-04 accept novelty 3.0

    The paper surveys techniques to speed up and reduce the resource needs of LLM inference, organized by data-level, model-level, and system-level changes, with comparative experiments on representative methods.