arXiv:2605.11222v2 Announce Type: replace Abstract: Quantization is an effective strategy to reduce the storage and computation footprint of large language models (LLMs). Post-training quantization (...
The paper revisits Kashin‑decomposition‑based weight quantization for large language models, introducing an improved algorithm that uses a sign‑randomized Discrete Cosine Transform (DCT) instead of a dense random orthogonal matrix. This change reduces per‑iteration cost from ≠(N^2) to ≠(N log N) and, combined with a greedy alternating‑update scheme, guarantees the four‑peak distribution needed for stable 2‑bit clustering while eliminating the need for multi‑restart k‑means. The resulting JAX pipeline, when paired with OPTQ‑style error compensation and QuIP‑style incoherence preprocessing, competes with state‑of‑the‑art quantization methods on OPT, Llama‑2, and Pythia at 4‑bit per channel, and remains numerically stable under stress configurations that cause other methods to diverge.
arXiv:2609.00224v1 Announce Type: cross Abstract: Weight-only post-training quantization (PTQ) can alleviate the computational burden of serving large language models (LLMs) at scale. However, existi...
arXiv:2607. 07964v1 Announce Type: new Abstract: Post-training quantization (PTQ) is a widely adopted technique for compressing large language models (LLMs) without retraining.
arXiv:2601. 21626v2 Announce Type: replace-cross Abstract: Post Training Quantization (PTQ), a mainstream model compression technique, often leads to the paradoxical 'low error, high loss' phenomenon because it focuses solely on minimizing quantization error.
JARQ is a plug‑in refinement method for group‑wise post‑training quantizers that alternates a joint least‑squares fit of all group scales with bounded Babai proposals to move many codes together on the current grid. It solves a bilinear box‑constrained mixed‑integer least‑squares problem without backpropagation, preserves the host model’s bit width, groups, zero points, and inference cost, and improves accuracy across Llama‑2, Llama‑3, and Qwen models when paired with RTN, GPTQ, OmniQuant, and AWQ hosts. In practice, JARQ reduces perplexity in 90 of 96 comparisons, cuts three‑bit RTN perplexity by up to 36%, raises mean multiple‑choice accuracy in 23 of 24 configurations, and enhances QEP, QuaRot, and OJBKQ outputs, all within under a minute per 7B block.
The paper introduces CanonQ, a unified quantization‑aware training framework that separates source canonicalization from task‑aware adaptation to enable extreme low‑bit LLM compression. By applying fixed rotations and energy normalization, CanonQ maps heterogeneous tensors to canonical coordinates, allowing frozen Gaussian‑reference codebooks to be reused across layers and models. Joint training then adapts the network to the coupled errors of weight, activation, and KV cache quantization, achieving significant performance gains—up to 14.28× lower perplexity and 57.9% higher zero‑shot accuracy on LLaMA models, and substantial improvements on instruction‑tuned MobileLLM‑Pro‑1B. "whyItMatters":"The framework demonstrates that carefully designed canonicalization and joint training can dramatically improve the efficiency and accuracy of low‑bit quantized large language models, offering a practical path to deploy powerful LLMs with reduced computational and memory footprints."
arXiv:2607. 02893v1 Announce Type: new Abstract: Low-bit quantization shrinks language models but treats precision as a single global hyper-parameter: every weight uses the same bit-width.
arXiv:2605. 13768v2 Announce Type: replace-cross Abstract: This is the second part of the work investigating quantized matrix multiplication (MatMul).
ThinQuant introduces efficient rotation learning for low‑bit weight and activation quantization of large language models by reducing calibration data through a geometric selection of activations and solving a lower‑dimensional optimization problem via an ADMM algorithm. The method achieves comparable or better quantization performance with dramatically fewer calibration points, completing rotation calibration for Llama‑3‑70B in under 12 minutes and for Llama‑3.1‑405B in just over 2 hours on a single GPU. ThinQuant outperforms existing gradient‑free approaches such as DartQuant and gradient‑based SpinQuant in both speed and perplexity metrics on WikiText‑2.
As large language model inference shifts toward lower precision, post-training quantization (PTQ) becomes increasingly brittle, making quantization-aware training (QAT) essential for preserving model...
HARP (Hadamard‑Preconditioned Adaptive Rotation Processor) is a learnable, structured two‑sided orthogonal processor that replaces fixed randomized Hadamard transforms in post‑training quantization of large language models. By representing rotations as sparse butterfly‑like block‑orthogonal stages and supporting mixed‑radix schedules, HARP adapts the quantization basis to each layer and calibration distribution while maintaining full‑precision equivalence. Across 2–4‑bit settings on Llama models from 1B to 70B, HARP consistently improves perplexity, delivers the strongest zero‑shot gains at 2 bits, and preserves deployment efficiency—achieving 128 tokens per second on Llama 2 7B at 2 bits, roughly 90% of RHT throughput and over twice the speed of FP16.