Recurrent-State Quantization

Latest papers

9 papers6 papers in the last four weeks, against 2 the four weeks before. 0.0% of all new papers.

PACE: Periodic and Channel-Aligned Encoding of Four-Bit Recurrent States
Oct 21, 2026Patrick Haller, Filipe Laitenberger, Alan AkbikRecurrent-State Quantization LLM Quantization
Humboldt-Universität zu Berlin
Preserving Updates in Low-Bit Recurrent State Quantization
Oct 11, 2026Haotian Li, Liran Dong, Kaicheng Xiao +1Recurrent-State Quantization Low-Bit Quantization
The Chinese University of Hong Kong
When Dynamic Scaling Fails: A Scale-State Feedback Governs Recurrent-State Quantization in Linear-Attention and SSM LLMs
Oct 10, 2026Yuhang JiangRecurrent-State Quantization LLM Quantization
Independent Researcher
Low-Discrepancy Dither for Quantized Recurrent State Caches
Sep 30, 2026Snigdha Chandan KhilarLLM Quantization Recurrent-State Quantization
Independent Researcher
STEPQuant: When and Where Errors Matter in Delta-Rule Recurrent State Quantization
Sep 29, 2026Bingchen Yao, Haobo Xu, Haokun Lin +6LLM Quantization LLM Inference Acceleration
Zhejiang University · Tsinghua University · NLPR & MAIS, Institute of Automation, CAS +3
LeapQuant: Efficient Linear Attention with Accurate Recurrent State Quantization
Sep 29, 2026Yi Pan, Haocheng Xi, Kan Zhu +10LLM Quantization LLM Inference Acceleration
UC Berkeley · University of Washington · MIT +2
Low-Bit Recurrent States in Hybrid Language Models
Sep 25, 2026Hongren Chen, Jiayang HeHybrid Language Models Recurrent-State Quantization
When Quantization Breaks Memory: Recurrent-State Write-Back in Low-Precision Temporal Inference
Sep 3, 2026Ismail Erbas, Xavier Intes, Vikas PandeyRecurrent Neural Networks Recurrent-State Quantization
Department of Biomedical Engineering Center for Modeling, Simulation, and Imaging in Medicine Rensselaer Polytechnic Institute Troy, NY 12180
DAMP: Decay-Aware Mixed-Precision Recurrent-State Quantization
Aug 27, 2026Tao Zhang, Jianchao Tan, Pingwei Sun +6Mixed-Precision Quantization LLM Quantization
South China University of Technology · Meituan · East China Normal University

Oct 21, 2026J/K move · Enter open · S save

PACE: Periodic and Channel-Aligned Encoding of Four-Bit Recurrent States

Patrick Haller, Filipe Laitenberger, Alan Akbik

Humboldt-Universität zu Berlin

Recurrent-State Quantization LLM Quantization Neural Network Quantization 4-Bit Quantization Post-Training Quantization Mixed-Precision Quantization

Open paper AskPDF arXiv

Linear-attention language models keep their context in a recurrent state of fixed size, where an inference system stores one copy of this state for every sequence it decodes at the same time. Storing the state at lower precision therefore lets more sequences share an accelerator. Existing state quantizers round the state after every generated token and report that four bits lose accuracy unless the bit width is raised or the model is fine-tuned.\par We introduce PACE, a post-hoc four-bit state quantizer. PACE rounds the state only once every few generated tokens, as chunked kernels already do once per chunk during prefill, and keeps the tokens generated in between in a small buffer. Rounding less often reduces the error that accumulates in the state. Additionally, we find that the large entries of the gated-delta state gather in columns, because some channels of the update written at each token remain large across many tokens. PACE therefore shares each quantization scale within one column. It also keeps the heads that retain rounding error longest at eight bits. On four hybrid models from 4B to 35B parameters, PACE with mixed four- and eight-bit heads is competitive with eight-bit DAMP on the three larger models at about two thirds of its memory, PACE with four-bit heads throughout scores higher than Q-Mamba at equal memory, and rounding every few tokens preserves the decoding speedup of the buffer.