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.
arXiv:2512. 00956v3 Announce Type: replace Abstract: Quantizing LLM weights and activations is a standard approach for efficient deployment, but a few extreme outliers can stretch the dynamic range and amplify low-bit quantization errors.
PrismQuant introduces a rotation framework that aligns the leading activation eigenspace with the constant group subspace of asymmetric grouped INT4 quantizers, optimizing the alignment via Ky Fan trace maximization and providing a closed‑form solution. The method employs compact Householder transformations for efficient, gradient‑free construction and applies a predictive range law to relate unaligned activation energy and group size to quantization variation. Experiments on large language models up to 70B parameters show PrismQuant achieving state‑of‑the‑art perplexity and accuracy, while delivering significant speedups and memory reductions in deployment scenarios.
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.
arXiv:2608. 08081v1 Announce Type: cross Abstract: Large mixture-of-experts (MoE) language models with 26--120 billion parameters exceed the memory capacity of consumer devices through three simultaneous pressures: resident weight matrices, key-value (KV) cache state that grows linearly with context, and dozens of expert sublayers that must be paged on demand.
arXiv:2607. 23047v1 Announce Type: cross Abstract: Mixed-precision quantization improves the accuracy of post-training quantization by allocating higher bitwidths to sensitive layers, but existing methods solve the allocation for a single fixed memory budget.
arXiv:2606. 02823v1 Announce Type: new Abstract: Two-bit weight quantization is attractive for memory-efficient LLM inference, but the standard W2 level set {-2,-1,0,+1} often collapses under aggressive W2A4/KV4 settings.
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 (...
arXiv:2606. 26587v1 Announce Type: cross Abstract: Low-bit floating-point formats and semi-structured sparsity are increasingly supported by modern accelerators, yet combining them for LLM activation compression remains challenging: activations contain input-dependent outliers that dominate block scales in FP4 quantization, and directly applying N:M sparsity masks discards moderate values, coupling sparsification loss with quantization error.
arXiv:2610.14745v1 Announce Type: new Abstract: Extreme low-bit weight-only quantization is an effective approach to reducing the deployment costs of large language models. Below 3 bits, scalar quant...
PRQuant introduces a training‑free, low‑overhead method for low‑bit quantization of linear layers by permuting input channels that cause the largest quantization error into contiguous tail blocks and precomputing residual weight sub‑tensors. The approach eliminates the need for online gathering during inference, converting scattered residual compensation into a regular tail‑augmented GEMM and thereby reducing latency. Experiments show that PRQuant lowers down‑projection reconstruction error and outperforms standard MXFP4 and other post‑training quantization baselines on five downstream benchmarks, improving accuracy by up to 1.24 points on Qwen3‑4B‑Instruct‑2507.
ConQuR introduces a lightweight post‑training rotation calibration for large language model activation quantization. By learning orthogonal rotations that align normalized activations with the corners of an inscribed hypercube, the method distributes activation energy evenly and can be updated online without storing activations. Experiments on Llama‑2 and Llama‑3 models (3B–70B) show competitive or improved perplexity and reasoning performance while avoiding costly training or large offline storage.
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.