# Compression

Compression is the reduction of digital data size by encoding it to eliminate redundancy, including methods such as run-length encoding and Lempel-Ziv compression.

This is one page of public article previews, not the complete archive. Follow Next page to continue. Summaries are not the original full articles.

## StarMPI-SVD: Distributed-Memory Parallel Star-M Tensor SVD

DevFeed: [StarMPI-SVD: Distributed-Memory Parallel Star-M Tensor SVD](<https://devfeed.tech/articles/starmpi-svd-distributed-memory-parallel-star-m-tensor-svd-146136.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.26451>)

Author: Naman Pesricha, Grey Ballard, Srinivas Eswar, Md Taufique Hussain, Vishwas Rao, Phani Motamarri

Published: 2026-11-20T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Distributed Computing](<https://devfeed.tech/sources/arxiv-distributed-computing.md>)

Topics: [Distributed Systems & Parallel Computing](<https://devfeed.tech/topics/distributed-systems-parallel-computing.md>), [High-Performance Computing](<https://devfeed.tech/topics/high-performance-computing.md>), [Algorithms](<https://devfeed.tech/topics/algorithms.md>), [Compression](<https://devfeed.tech/topics/compression.md>)

Tags: [adaptive](<https://devfeed.tech/tags/adaptive.md>), [asynchronous](<https://devfeed.tech/tags/asynchronous.md>), [communication](<https://devfeed.tech/tags/communication.md>), [compression](<https://devfeed.tech/tags/compression.md>), [cs-dc](<https://devfeed.tech/tags/cs-dc.md>), [data](<https://devfeed.tech/tags/data.md>), [datasets](<https://devfeed.tech/tags/datasets.md>), [distributed](<https://devfeed.tech/tags/distributed.md>), [efficiency](<https://devfeed.tech/tags/efficiency.md>), [experiments](<https://devfeed.tech/tags/experiments.md>), [kernels](<https://devfeed.tech/tags/kernels.md>), [matrix](<https://devfeed.tech/tags/matrix.md>), [nodes](<https://devfeed.tech/tags/nodes.md>), [order](<https://devfeed.tech/tags/order.md>), [parallel](<https://devfeed.tech/tags/parallel.md>), [scalability](<https://devfeed.tech/tags/scalability.md>), [scaling](<https://devfeed.tech/tags/scaling.md>), [singular-value-decomposition](<https://devfeed.tech/tags/singular-value-decomposition.md>), [size](<https://devfeed.tech/tags/size.md>), [star](<https://devfeed.tech/tags/star.md>), [svd](<https://devfeed.tech/tags/svd.md>)

### AI overview

The work presents the first distributed-memory implementations of t-SVDM-I and adaptive t-SVDM-II for N-way tensors, extending star-M compression beyond shared-memory settings. t-SVDM-I uses invertible transformations and slice-wise truncated SVDs at a prescribed rank; t-SVDM-II selects slice-wise ranks to meet an error tolerance. Asynchronous, non-blocking communication is used for data redistribution to address computational and communication bottlenecks. Strong-scaling experiments report at least 70% parallel efficiency on up to 128 nodes (6,144 cores); the methods are applied to datasets up to 768 GB.

### Source excerpt

arXiv:2611.26451v1 Announce Type: new Abstract: The star-M tensor algebra generalizes the optimal compression guarantees of the matrix singular value decomposition (SVD) to higher-order tensors. The t-SVDM-I decomposition is computed using a set of invertible transformations and slice-wise truncated SVDs with a prescribed rank, and its adaptive, multirank variant t-SVDM-II can determine slice-wise ranks to satisfy a prescribed error tolerance. Existing implementations of these algorithms are limited to shared-memory environments, restricting their applicability to modern large-scale datasets. In this work, we present the first distributed-memory implementations of t-SVDM-I and t-SVDM-II, which are generalized to N-way tensors. To address computational and communication bottlenecks, we design asynchronous, non-blocking communication strategies for efficient data redistribution to obtain highly efficient key computational kernels. Strong-scaling experiments demonstrate the efficiency and scalability of our approach (at least 70% parallel efficiency on up to 128 nodes, or 6144 cores), and we apply the star-M compression techniques to a set of application datasets of size up to 768 GB.

## SCAR: State-Space Compression for AI-Based Radio Resource Management of Vehicular Services

DevFeed: [SCAR: State-Space Compression for AI-Based Radio Resource Management of Vehicular Services](<https://devfeed.tech/articles/scar-state-space-compression-for-ai-based-radio-resource-management-of-vehicular-services-144855.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2508.06243>)

Author: Ioan-Sorin Comsa, Purav Shah, Karthik Vaidhyanathan, Deepak Gangadharan, Christof Imhof, Per Bergamin, Aryan Kaushik, Gabriel-Miro Muntean, Ramona Trestian

Published: 2026-11-20T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [clustering](<https://devfeed.tech/topics/clustering.md>), [Representation learning](<https://devfeed.tech/topics/representation-learning.md>), [Compression](<https://devfeed.tech/topics/compression.md>), [World models](<https://devfeed.tech/topics/world-models.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [algorithms](<https://devfeed.tech/tags/algorithms.md>), [clustering](<https://devfeed.tech/tags/clustering.md>), [compression](<https://devfeed.tech/tags/compression.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [cs-ne](<https://devfeed.tech/tags/cs-ne.md>), [cs-sy](<https://devfeed.tech/tags/cs-sy.md>), [edge](<https://devfeed.tech/tags/edge.md>), [eess-sy](<https://devfeed.tech/tags/eess-sy.md>), [fairness](<https://devfeed.tech/tags/fairness.md>), [learning](<https://devfeed.tech/tags/learning.md>), [machine-learning](<https://devfeed.tech/tags/machine-learning.md>), [management](<https://devfeed.tech/tags/management.md>), [management-policies](<https://devfeed.tech/tags/management-policies.md>), [metrics](<https://devfeed.tech/tags/metrics.md>), [ml](<https://devfeed.tech/tags/ml.md>), [network](<https://devfeed.tech/tags/network.md>), [radio](<https://devfeed.tech/tags/radio.md>), [results](<https://devfeed.tech/tags/results.md>), [rl](<https://devfeed.tech/tags/rl.md>), [sast](<https://devfeed.tech/tags/sast.md>), [services](<https://devfeed.tech/tags/services.md>), [simulation](<https://devfeed.tech/tags/simulation.md>), [state-space](<https://devfeed.tech/tags/state-space.md>)

### AI overview

SCAR compresses high-dimensional vehicular network Channel Quality Indicator data into compact states for reinforcement-learning-based radio resource management. Its edge-assisted framework uses clustering and radial basis function networks to preserve information relevant to decisions, allowing policies to adjust the throughput-fairness trade-off against NGMN service-level fairness objectives. In simulations, compression raised average occupancy of the NGMN feasible region from 56.32% to 88.07% across ten RL algorithms, with the best configuration reaching 97.76%; the abstract reports limited throughput degradation and competitive computational overhead. SAST-based clustering reduced compression distortion by about 10% versus conventional K-means. These findings are simulation results, not evidence of deployment performance.

### Source excerpt

arXiv:2508.06243v3 Announce Type: replace Abstract: The increasing demand for connected vehicular services poses significant challenges for AI-based network and service management due to the high volume and rapid variability of network state information. Traditional management and control mechanisms struggle to scale when processing fine-grained metrics such as Channel Quality Indicators (CQIs) in dynamic vehicular environments. To address this challenge, we propose SCAR (State-Space Compression for AI-Based Radio Resource Management), an edge-assisted framework that abstracts high-dimensional CQI information into compact representations for AI-based management. SCAR employs machine-learning (ML)-based compression techniques, including clustering and radial basis function (RBF) networks, to reduce the dimensionality of CQI-derived state information while preserving essential features relevant to management decisions. The resulting compressed states are used to train reinforcement learning (RL)-based management policies that dynamically adapt the throughput-fairness trade-off according to service-level fairness objectives defined by the NGMN. Simulation results show that CQI compression improves average NGMN feasible-region occupancy from 56.32% to 88.07% across ten RL algorithms, reaching 97.76% for the best-performing RL configuration, with limited throughput degradation and competitive computational overhead. Furthermore, simulated annealing with stochastic tunneling (SAST)-based clustering reduces state compression distortion by approximately 10% compared with conventional K-means clustering. These results demonstrate SCAR's ability to provide compact state representations for RL-based network and service management while improving service-level fairness in dynamic vehicular systems.

## Complete-SVD: A Functional Norm Approach to Low-Rank LLM Compression

DevFeed: [Complete-SVD: A Functional Norm Approach to Low-Rank LLM Compression](<https://devfeed.tech/articles/complete-svd-a-functional-norm-approach-to-low-rank-llm-compression-145134.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.26059>)

Author: Alper Kalle, Theo Rudkiewicz, Mohamed-Oumar Ouerfelli, Mohamed Tamaazousti

Published: 2026-11-20T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Compression](<https://devfeed.tech/topics/compression.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>), [post-training](<https://devfeed.tech/topics/post-training.md>), [LLM optimization](<https://devfeed.tech/topics/llm-optimization.md>), [Neural Network](<https://devfeed.tech/topics/neural-network.md>)

Tags: [compression](<https://devfeed.tech/tags/compression.md>), [cs-cl](<https://devfeed.tech/tags/cs-cl.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [large-language-models](<https://devfeed.tech/tags/large-language-models.md>), [llm](<https://devfeed.tech/tags/llm.md>), [low-rank](<https://devfeed.tech/tags/low-rank.md>), [post-training](<https://devfeed.tech/tags/post-training.md>), [svd](<https://devfeed.tech/tags/svd.md>)

### AI overview

Complete-SVD compresses LLMs by minimizing the L2 distance between original and compressed model logits, giving equal weight to every logit change. Linearization and a Kronecker-factored approximation turn this objective into a weighted SVD with a closed-form solution. Because its output-side factor is intractable to compute directly, the method estimates it without bias using random probes, requiring K backward passes per sequence regardless of vocabulary size or sequence length. With uniform ranks and WikiText-2 calibration, it reports the lowest C4 and PTB perplexity at every matched setting, and at compression ratio 0.7 with LEMS rank allocation. The related KL-SVD variant lowers calibration perplexity in most settings and raises downstream accuracy, but also raises cross-corpus perplexity. Complete-SVD improves IFEval accuracy over the strongest prior output-aware methods throughout the reported comparisons.

### Source excerpt

arXiv:2611.26059v1 Announce Type: new Abstract: Post-training low-rank compression of neural networks must not only approximate the original weights but also preserve the model's input-output behavior. Recent output-aware methods measure the compression error beyond the compressed layer, at the model output, weighting logit changes by the observed next token or by the model's predicted probabilities. We propose Complete-SVD, a method for large language models that minimizes the $L_2$ distance between the logits of the original and compressed models, weighting all logit changes equally. Linearizing the network around the original weights, together with a Kronecker-factored approximation, reduces this objective to a weighted SVD with a closed-form solution. Its output-side factor is intractable to compute; we estimate it without bias from random probes, at a cost of $K$ backward passes per sequence, independent of vocabulary size and sequence length. At matched settings with uniform ranks and WikiText-2 calibration, Complete-SVD achieves the lowest C4 and PTB perplexity in every setting, and at ratio 0.7 under LEMS rank allocation. Replacing the $L_2$ distance by the Kullback-Leibler divergence (KL-SVD), with the same estimation framework, lowers calibration perplexity in most settings and raises downstream accuracy, but raises cross-corpus perplexity. Complete-SVD also improves IFEval instruction-following accuracy over the strongest prior output-aware methods throughout.

## Aggron: Aggressive Compression of Agentic KV Caches during Head-of-Line Stalls in the Request Queue

DevFeed: [Aggron: Aggressive Compression of Agentic KV Caches during Head-of-Line Stalls in the Request Queue](<https://devfeed.tech/articles/aggron-aggressive-compression-of-agentic-kv-caches-during-head-of-line-stalls-in-the-request-queue-146134.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.26435>)

Author: Rishabh Ravi, Tony Hong, Chen Wang, Yue Zhu, Mengmei Ye, Hubertus Franke, Fan Lai, Josep Torrellas

Published: 2026-11-20T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Distributed Computing](<https://devfeed.tech/sources/arxiv-distributed-computing.md>)

Topics: [LLM optimization](<https://devfeed.tech/topics/llm-optimization.md>), [AI Context Management](<https://devfeed.tech/topics/ai-context-management.md>), [Compression](<https://devfeed.tech/topics/compression.md>), [queueing theory](<https://devfeed.tech/topics/queueing-theory.md>), [performance-engineering](<https://devfeed.tech/topics/performance-engineering.md>), [vllm](<https://devfeed.tech/topics/vllm.md>)

Tags: [agentic](<https://devfeed.tech/tags/agentic.md>), [cache](<https://devfeed.tech/tags/cache.md>), [capacity](<https://devfeed.tech/tags/capacity.md>), [compression](<https://devfeed.tech/tags/compression.md>), [context-compression](<https://devfeed.tech/tags/context-compression.md>), [cs-ar](<https://devfeed.tech/tags/cs-ar.md>), [cs-dc](<https://devfeed.tech/tags/cs-dc.md>), [engine](<https://devfeed.tech/tags/engine.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [hardware](<https://devfeed.tech/tags/hardware.md>), [kv-cache](<https://devfeed.tech/tags/kv-cache.md>), [latency](<https://devfeed.tech/tags/latency.md>), [llm](<https://devfeed.tech/tags/llm.md>), [load](<https://devfeed.tech/tags/load.md>), [memory](<https://devfeed.tech/tags/memory.md>), [model](<https://devfeed.tech/tags/model.md>), [queue](<https://devfeed.tech/tags/queue.md>), [request](<https://devfeed.tech/tags/request.md>), [serving](<https://devfeed.tech/tags/serving.md>), [stalled](<https://devfeed.tech/tags/stalled.md>), [throughput](<https://devfeed.tech/tags/throughput.md>)

### AI overview

Aggron addresses KV-cache limits in long-running LLM agents by summarizing selected cache content at every turn, keeping each summarization request lightweight. It runs these requests in a separate queue while the corresponding agent request is stalled behind a head-of-line block, using capacity left idle in the main queue; it launches summarization only when it predicts completion before that agent request executes. The abstract says this approach retains accuracy and, versus state-of-the-art asynchronous summarization at high load, reduces P99 program latency by 27-31% and increases throughput by 1.34-1.40x. The supplied text is an abstract, so it does not provide evaluation details or broader comparisons.

### Source excerpt

arXiv:2611.26435v1 Announce Type: new Abstract: LLM agents run long, multi-turn programs whose contexts grow at every step, and serving them is limited by KV cache capacity. To alleviate this limitation, systems use context compression techniques, which typically wait until the KV cache crosses a large threshold and then summarize large spans of history at once, either in the background while the agent progresses or on the critical path. These summarizations can be done using the same model running on the same engine, or using the same or a different model running on extra hardware. When running on the same engine, summarization competes with the requests being served for both compute and memory capacity. To maximize the cost-effectiveness of summarization, this paper presents a new serving system called {\em Aggron}. Aggron summarizes agentic requests at every turn and selectively summarizes only certain parts of the KV cache. As a result, Aggron retains accuracy and keeps summarization requests lightweight. Moreover, Aggron launches summarization requests from a separate {\em Summarization Queue} while the corresponding agentic request whose KV cache we want to compress is stalled in the main request queue due to head-of-line stalls. The Summarization queue utilizes compute and memory capacity left over by the head-of-line stalls in the main queue. Further, to maximize efficiency, the server issues summarizations only when it predicts that they can complete before the corresponding agentic request will finally execute. In our evaluation, we compare Aggron to state-of-the-art asynchronous summarization. At high load, Aggron reduces P99 program latency by $27$--$31\%$, and raises throughput by $1.34$--$1.40\times$.

## Valkey 9.2 Delivers More Performance, Zstd Compression & Lower Memory Overhead

DevFeed: [Valkey 9.2 Delivers More Performance, Zstd Compression & Lower Memory Overhead](<https://devfeed.tech/articles/valkey-9-2-delivers-more-performance-zstd-compression-lower-memory-overhead-143043.md>)

Original publisher: [Read original article](<https://www.phoronix.com/news/Valkey-9.2>)

Author: Michael Larabel

Published: 2026-11-19T10:41:47Z

Content type: release

Language: en

Sources: [Phoronix](<https://devfeed.tech/sources/phoronix.md>)

Topics: [valkey](<https://devfeed.tech/topics/valkey.md>), [Compression](<https://devfeed.tech/topics/compression.md>), [database-optimization](<https://devfeed.tech/topics/database-optimization.md>)

Tags: [compression](<https://devfeed.tech/tags/compression.md>), [desktop-linux](<https://devfeed.tech/tags/desktop-linux.md>), [linux-benchmarking](<https://devfeed.tech/tags/linux-benchmarking.md>), [linux-hardware-benchmarks](<https://devfeed.tech/tags/linux-hardware-benchmarks.md>), [linux-hardware-reviews](<https://devfeed.tech/tags/linux-hardware-reviews.md>), [linux-how-to](<https://devfeed.tech/tags/linux-how-to.md>), [linux-performance](<https://devfeed.tech/tags/linux-performance.md>), [linux-server-benchmarks](<https://devfeed.tech/tags/linux-server-benchmarks.md>), [memory](<https://devfeed.tech/tags/memory.md>), [open-source-graphics](<https://devfeed.tech/tags/open-source-graphics.md>), [overhead](<https://devfeed.tech/tags/overhead.md>), [performance](<https://devfeed.tech/tags/performance.md>), [phoronix](<https://devfeed.tech/tags/phoronix.md>), [phoronix-test-suite](<https://devfeed.tech/tags/phoronix-test-suite.md>), [replication](<https://devfeed.tech/tags/replication.md>), [ubuntu-benchmarks](<https://devfeed.tech/tags/ubuntu-benchmarks.md>), [ubuntu-hardware](<https://devfeed.tech/tags/ubuntu-hardware.md>), [valkey](<https://devfeed.tech/tags/valkey.md>), [zstd](<https://devfeed.tech/tags/zstd.md>)

### AI overview

Valkey 9.2 combines performance work with changes aimed at reducing memory overhead. It adds B+ tree sorted sets, Zstd compression for RDB persistence and replication (described as achieving greater compression ratios than existing LZ4), forkless snapshotting and full synchronization to replicas, and server-side hot key detection. The new DELEX command conditionally deletes a string key when its value matches or differs from a given value. The excerpt also notes other optimizations and dozens of bug fixes.

### Source excerpt

Valkey 9.2 is out today as the latest feature update to this leading fork of the Redis in-memory, key-value database...

## CommitKV: Commit-Aligned KV Cache Compression for Multimodal Agents

DevFeed: [CommitKV: Commit-Aligned KV Cache Compression for Multimodal Agents](<https://devfeed.tech/articles/commitkv-commit-aligned-kv-cache-compression-for-multimodal-agents-143297.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2608.07855>)

Author: Weizhong Huang, Jinchao Zhang, Xiawu Zheng

Published: 2026-11-19T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Compression](<https://devfeed.tech/topics/compression.md>), [Cache](<https://devfeed.tech/topics/cache.md>), [sglang](<https://devfeed.tech/topics/sglang.md>), [vllm](<https://devfeed.tech/topics/vllm.md>)

Tags: [agents](<https://devfeed.tech/tags/agents.md>), [attention](<https://devfeed.tech/tags/attention.md>), [benchmarks](<https://devfeed.tech/tags/benchmarks.md>), [budget](<https://devfeed.tech/tags/budget.md>), [cache](<https://devfeed.tech/tags/cache.md>), [compression](<https://devfeed.tech/tags/compression.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [experiments](<https://devfeed.tech/tags/experiments.md>), [gpu-memory](<https://devfeed.tech/tags/gpu-memory.md>), [inference](<https://devfeed.tech/tags/inference.md>), [inference-costs](<https://devfeed.tech/tags/inference-costs.md>), [kv-cache](<https://devfeed.tech/tags/kv-cache.md>), [measurement](<https://devfeed.tech/tags/measurement.md>), [memory](<https://devfeed.tech/tags/memory.md>), [models](<https://devfeed.tech/tags/models.md>), [multimodal](<https://devfeed.tech/tags/multimodal.md>), [multimodal-models](<https://devfeed.tech/tags/multimodal-models.md>), [qwen3](<https://devfeed.tech/tags/qwen3.md>), [redundancy](<https://devfeed.tech/tags/redundancy.md>), [sglang](<https://devfeed.tech/tags/sglang.md>), [times](<https://devfeed.tech/tags/times.md>)

### AI overview

CommitKV reduces multimodal agents' KV cache costs by comparing the deletion effect of the same page before a tool call completes and after its observation returns. It retires pages with low influence in both windows, temporarily protects pages influential before the call, then fills the cache budget using attention importance, key redundancy, and token-level deletion effects. Across four open-weight models and eight benchmarks, it achieved the highest average accuracy among the compared compression methods on each model. For Qwen3-VL-8B, accuracy was 69.85% at 512 tokens, 0.91% below FullKV; at 256 tokens, it beat the strongest compression baseline by 8.44%. In matched deployments with SGLang and vLLM at 256 tokens, peak allocated GPU memory fell 54.37%-55.23% and end-to-end inference was 2.12-2.17x faster than FullKV.

### Source excerpt

arXiv:2608.07855v2 Announce Type: replace Abstract: Multimodal agents gather visual evidence through reasoning and tool use, extending their ability to solve tasks beyond the initial input. However, accumulating observations increases KV cache memory and inference costs. Token eviction reduces these costs by retaining useful cached tokens. Existing methods use attention, redundancy, or stage-specific queries to rank tokens, without explicitly comparing the same history across tool interactions. When observations shift attention, they may mistake temporarily reduced influence for lost usefulness and evict information needed later. We propose CommitKV, a training-free KV cache compression method that accounts for this changing evidence use. Commit-aligned measurement compares the deletion effects of the same page before a tool call completes and after its observation returns. Lifecycle-aware compression retires pages with low influence in both windows and temporarily protects pages influential before the call. Joint scoring combines attention importance, key redundancy, and token-level deletion effects to fill the remaining budget. Extensive experiments cover four open-weight multimodal models and eight benchmarks. CommitKV achieves the highest average accuracy among the compared compression methods on all four models. On Qwen3-VL-8B, it achieves 69.85% at a 512-token budget, only 0.91% below FullKV, and exceeds the strongest compression baseline by 8.44% at 256 tokens. Deploying the Qwen3-VL-8B agent with SGLang and vLLM at a 256-token budget reduces peak allocated GPU memory by 54.37%-55.23% and accelerates end-to-end inference by 2.12-2.17$\times$ over FullKV on the matched workload.

## Data-Driven Variable-Length Compression of 3D Representations

DevFeed: [Data-Driven Variable-Length Compression of 3D Representations](<https://devfeed.tech/articles/data-driven-variable-length-compression-of-3d-representations-141678.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.21802>)

Author: Yvette Y. Lin, Sai Karthikey Pentapati, Sarvesh Sundaram, Guandao Yang, Sara Fridovich-Keil

Published: 2026-11-18T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Compression](<https://devfeed.tech/topics/compression.md>), [3D](<https://devfeed.tech/topics/3d.md>), [Representation learning](<https://devfeed.tech/topics/representation-learning.md>), [signed distance functions](<https://devfeed.tech/topics/signed-distance-functions.md>), [dataset](<https://devfeed.tech/topics/dataset.md>)

Tags: [3d](<https://devfeed.tech/tags/3d.md>), [abstract](<https://devfeed.tech/tags/abstract.md>), [adapting](<https://devfeed.tech/tags/adapting.md>), [applications](<https://devfeed.tech/tags/applications.md>), [complexity](<https://devfeed.tech/tags/complexity.md>), [compression](<https://devfeed.tech/tags/compression.md>), [cs-cv](<https://devfeed.tech/tags/cs-cv.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [data-driven](<https://devfeed.tech/tags/data-driven.md>), [dataset](<https://devfeed.tech/tags/dataset.md>), [efficiency](<https://devfeed.tech/tags/efficiency.md>), [exploits](<https://devfeed.tech/tags/exploits.md>), [fast](<https://devfeed.tech/tags/fast.md>), [fields](<https://devfeed.tech/tags/fields.md>), [fixed](<https://devfeed.tech/tags/fixed.md>), [general](<https://devfeed.tech/tags/general.md>), [inference](<https://devfeed.tech/tags/inference.md>), [leverage](<https://devfeed.tech/tags/leverage.md>), [ratio](<https://devfeed.tech/tags/ratio.md>), [shapes](<https://devfeed.tech/tags/shapes.md>), [size](<https://devfeed.tech/tags/size.md>), [variable](<https://devfeed.tech/tags/variable.md>)

### AI overview

The method learns dataset-wide structure to compress 3D representations at variable lengths, adapting each shape's representation size to its complexity through fast feed-forward inference. It works on general latent representations and is reported to improve rate-distortion for both meshes and signed distance fields. The supplied text does not include quantitative results or details of the evaluation.

### Source excerpt

arXiv:2611.21802v1 Announce Type: cross Abstract: Compressing 3D shapes is important for the efficiency of many applications, yet existing methods either cannot leverage dataset-level statistics or use a fixed compression ratio for all shapes. We propose a learned variable-length compression method that exploits structure across a training dataset while adapting compressed representation size to individual shape complexity using only fast feed-forward inference. Our method operates on general latent representations and improves rate-distortion for both meshes and signed distance fields. Code is available at https://github.com/voilalab/varlen-3D.

## Similarity-Aware Client-Side Reordering for Server-Side Storage Compression in Distributed File Systems

DevFeed: [Similarity-Aware Client-Side Reordering for Server-Side Storage Compression in Distributed File Systems](<https://devfeed.tech/articles/similarity-aware-client-side-reordering-for-server-side-storage-compression-in-distributed-file-systems-140390.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.22847>)

Author: Yuanhong Huang, Zicong Chen, Yili Ma, Jinwu Yang, Yulong Chen, Guangming Tan, Dingwen Tao

Published: 2026-11-18T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Distributed Computing](<https://devfeed.tech/sources/arxiv-distributed-computing.md>)

Topics: [systems](<https://devfeed.tech/topics/systems.md>), [Compression](<https://devfeed.tech/topics/compression.md>), [ZFS](<https://devfeed.tech/topics/zfs-file-system-features.md>)

Tags: [adaptive](<https://devfeed.tech/tags/adaptive.md>), [backend](<https://devfeed.tech/tags/backend.md>), [block](<https://devfeed.tech/tags/block.md>), [blocks](<https://devfeed.tech/tags/blocks.md>), [bounded](<https://devfeed.tech/tags/bounded.md>), [candidate](<https://devfeed.tech/tags/candidate.md>), [compression](<https://devfeed.tech/tags/compression.md>), [cs-dc](<https://devfeed.tech/tags/cs-dc.md>), [data](<https://devfeed.tech/tags/data.md>), [datasets](<https://devfeed.tech/tags/datasets.md>), [distributed](<https://devfeed.tech/tags/distributed.md>), [efficiency](<https://devfeed.tech/tags/efficiency.md>), [file](<https://devfeed.tech/tags/file.md>), [file-system](<https://devfeed.tech/tags/file-system.md>), [framework](<https://devfeed.tech/tags/framework.md>), [implement](<https://devfeed.tech/tags/implement.md>), [layout](<https://devfeed.tech/tags/layout.md>), [server](<https://devfeed.tech/tags/server.md>), [similarity](<https://devfeed.tech/tags/similarity.md>), [storage](<https://devfeed.tech/tags/storage.md>), [systems](<https://devfeed.tech/tags/systems.md>)

### AI overview

The framework improves server-side compression by reordering blocks on the client before storage, placing related blocks within the same compression scope while retaining the existing server compressor. It works in bounded windows, periodically chooses among candidate layouts using lightweight compression of sampled data, and restores logical order on reads with compact window-local metadata. In SkyFS with Zstandard on an OpenZFS backend, it reduced physical storage by up to 42.9% versus logical-order writes; sequential-read throughput was 94-108% of baseline and random-read IOPS 91-99%. Layout construction was the main added write cost.

### Source excerpt

arXiv:2611.22847v1 Announce Type: new Abstract: As growing data volumes strain the capacity of distributed file systems, server-side storage compression transparently reduces disk usage, but its effectiveness is limited when similar blocks land in separate compression records and cannot share matches. We present a client-side reordering framework that transforms the physical block layout before persistence, so that related blocks are co-located within the same compression scope without replacing the existing storage-server compressor. It operates on bounded reordering windows, periodically selects among candidate layouts via lightweight compression of sampled data, and recovers the original logical order during reads with compact window-local metadata. We implement the framework in the SkyFS distributed file system, using the Zstandard (Zstd) compressor on the OpenZFS storage backend. Results on diverse datasets show that the framework reduces physical storage by up to 42.9% relative to logical-order writes, while achieving 94-108% of baseline sequential-read throughput and 91-99% of random-read IOPS. Client-side layout construction is the primary additional write cost, showing that adaptive similarity-aware reordering substantially improves storage efficiency with acceptable overhead.

## Lasso: Towards Sustainable AI-Generated Media Detection with Intel QAT On-Chip Compression

DevFeed: [Lasso: Towards Sustainable AI-Generated Media Detection with Intel QAT On-Chip Compression](<https://devfeed.tech/articles/lasso-towards-sustainable-ai-generated-media-detection-with-intel-qat-on-chip-compression-137360.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.18895>)

Author: Darsh Asher, Joshua Kalyanapu, Kaushal Mhapsekar, Bita Aslrousta, Amna Hassan, Debayan Sengupta, Samira Mirbagher Ajorpaz

Published: 2026-11-17T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>), [arXiv: Cryptography and Security](<https://devfeed.tech/sources/arxiv-cryptography-and-security.md>)

Topics: [Reverse Engineering](<https://devfeed.tech/topics/reverse-engineering.md>), [Sustainable Computing](<https://devfeed.tech/topics/sustainable-computing.md>), [Compression](<https://devfeed.tech/topics/compression.md>), [Generative AI](<https://devfeed.tech/topics/generative-ai.md>)

Tags: [ai](<https://devfeed.tech/tags/ai.md>), [ai-generated-content](<https://devfeed.tech/tags/ai-generated-content.md>), [ai-generated-media](<https://devfeed.tech/tags/ai-generated-media.md>), [artifacts](<https://devfeed.tech/tags/artifacts.md>), [arxiv](<https://devfeed.tech/tags/arxiv.md>), [binoculars](<https://devfeed.tech/tags/binoculars.md>), [chip](<https://devfeed.tech/tags/chip.md>), [code](<https://devfeed.tech/tags/code.md>), [compression](<https://devfeed.tech/tags/compression.md>), [content](<https://devfeed.tech/tags/content.md>), [cost](<https://devfeed.tech/tags/cost.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-ar](<https://devfeed.tech/tags/cs-ar.md>), [cs-cr](<https://devfeed.tech/tags/cs-cr.md>), [cs-cv](<https://devfeed.tech/tags/cs-cv.md>), [data](<https://devfeed.tech/tags/data.md>), [data-structure](<https://devfeed.tech/tags/data-structure.md>), [detection](<https://devfeed.tech/tags/detection.md>), [device](<https://devfeed.tech/tags/device.md>), [domains](<https://devfeed.tech/tags/domains.md>), [edge](<https://devfeed.tech/tags/edge.md>), [energy](<https://devfeed.tech/tags/energy.md>), [intel](<https://devfeed.tech/tags/intel.md>), [media](<https://devfeed.tech/tags/media.md>)

### AI overview

Lasso uses timing differences in Intel QuickAssist Technology's compression accelerator as a signal for detecting AI-generated media. For text, the abstract reports 100% accuracy on original DetectRL text, 92% average after paraphrasing, and 82% across original, paraphrased, and perturbed text; accuracy reaches 95% for matched instruction-tuned generators, including an unseen generator. On RAID, it averages 82% when tested on a different generator and 74% across eight held-out domains. Detection takes 0.74 microseconds, at least 83,784 times faster than five baselines, and uses over 99% less energy per document than Binoculars and DetectGPT. It also reports 67.5% accuracy for video, 70% for code, and 78% for images. These are abstract-level reported results; evaluation details are not included in the supplied excerpt.

### Source excerpt

arXiv:2611.18895v1 Announce Type: cross Abstract: Synthetic text, images, and video are rising. Software-only detectors are power-hungry, too slow, and costly, and do not scale with demand. When software cannot scale, historically hardware often comes to the rescue. We perform the first reverse engineering of a compression accelerator--Intel QuickAssist Technology (QAT) microarchitecture and--find different amount of time on each, a novel data structure-dependent property. Using this information, we then build Lasso, a low-cost, high-performance detector for AI-generated content that turns the accelerator into a measuring instrument. For text detection, Lasso achieves 100% accuracy on original DetectRL text and retains 92% average accuracy after paraphrasing, compared with 78% for Binoculars and 55% for DetectGPT. Across original, paraphrased, and perturbed text, it achieves 82% average accuracy, outperforming all five baselines. With topic, length, and formatting matched, accuracy reaches 95% for instruction-tuned generators, including when the tested generator is unseen. On RAID, accuracy averages 82% when trained on one generator and tested on another, and 74% across eight held-out domains. Detection takes 0.74 microseconds, a speedup of at least 83,784x across five baselines. At the same time, Lasso consumes over 99% less energy per document than both Binoculars and DetectGPT. At the same time, Lasso detects AI-generated videos, software code, and images with 67.5%, 70% and 78% accuracy, respectively, with low latency and power opening avenues for practical real time and edge device detection of multimodality AI artifacts for the first time. No other compared technique supports detection across all four modalities: text, video, images, and code.

## MiSTiQ: MicroScaling Trellis Quantization for Extreme KV Cache Compression

DevFeed: [MiSTiQ: MicroScaling Trellis Quantization for Extreme KV Cache Compression](<https://devfeed.tech/articles/mistiq-microscaling-trellis-quantization-for-extreme-kv-cache-compression-139142.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.21216>)

Author: Ofir Gordon, Lior Dikstein, Paul Balanca, Idan Achituve, Douglas Orr, Hai Victor Habi

Published: 2026-11-17T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [quantization](<https://devfeed.tech/topics/quantization.md>), [Compression](<https://devfeed.tech/topics/compression.md>), [NVFP4](<https://devfeed.tech/topics/nvfp4.md>)

Tags: [attention](<https://devfeed.tech/tags/attention.md>), [beam](<https://devfeed.tech/tags/beam.md>), [bits](<https://devfeed.tech/tags/bits.md>), [block](<https://devfeed.tech/tags/block.md>), [cache](<https://devfeed.tech/tags/cache.md>), [compression](<https://devfeed.tech/tags/compression.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [hardware](<https://devfeed.tech/tags/hardware.md>), [inference](<https://devfeed.tech/tags/inference.md>), [kv-cache](<https://devfeed.tech/tags/kv-cache.md>), [language-models](<https://devfeed.tech/tags/language-models.md>), [large-language-models](<https://devfeed.tech/tags/large-language-models.md>), [large-language-models-llms](<https://devfeed.tech/tags/large-language-models-llms.md>), [llama3](<https://devfeed.tech/tags/llama3.md>), [llms](<https://devfeed.tech/tags/llms.md>), [memory](<https://devfeed.tech/tags/memory.md>), [nvfp4](<https://devfeed.tech/tags/nvfp4.md>), [perplexity](<https://devfeed.tech/tags/perplexity.md>), [precision](<https://devfeed.tech/tags/precision.md>), [quantization](<https://devfeed.tech/tags/quantization.md>), [trellis](<https://devfeed.tech/tags/trellis.md>)

### AI overview

MiSTiQ combines trellis-coded KV-cache storage and microscaling (MX) formats to target both extreme compression and low-precision attention computation. Trellis quantization with beam search explores large reconstruction spaces without large codebooks, while MX-Hash functions produce reconstruction values directly in the target MX format using scale-conditioned statistics of block-normalized Gaussian inputs. The abstract reports results on Llama3 and Qwen3 at about 1.25 effective bits per element: lower WikiText perplexity and higher GSM8K accuracy than evaluated baselines across several block-scale formats, and the highest LongBench average for both models with NVFP4. The excerpt does not provide numerical scores or implementation details.

### Source excerpt

arXiv:2611.21216v1 Announce Type: new Abstract: Modern large language models (LLMs) increasingly support longer contexts and serve more concurrent requests, increasing both sequence lengths and batch sizes. Both trends increase KV-cache size and memory traffic during inference. KV-cache quantization reduces these costs, with recent approaches using codebook-based vector quantization to achieve extremely low bit widths. However, hardware constraints limit codebook size, and most methods reconstruct the cache in higher-precision floating-point formats for attention, limiting the computational benefits of quantization. To address these limitations, we introduce MiSTiQ, a drop-in replacement for codebook-based vector quantizers that combines trellis-coded storage with microscaling (MX) formats for low-precision computation. We combine trellis quantization with beam search to encode the KV cache over large reconstruction spaces without large codebooks. To support MX computation, we derive MX-Hash functions that generate reconstruction values directly in the target MX element format. These functions use the scale-conditioned marginal distribution of block-normalized Gaussian inputs. On Llama3 and Qwen3, MiSTiQ achieves lower WikiText perplexity and higher GSM8K accuracy than the evaluated baselines across several block-scale formats at approximately 1.25 effective bits per element. With NVFP4, it also achieves the highest LongBench average on both models.

## MeltKV: Memory-Aligned KV Cache Compression via Variable-Bit Packed Vector Quantization for Long-Context LLMs

DevFeed: [MeltKV: Memory-Aligned KV Cache Compression via Variable-Bit Packed Vector Quantization for Long-Context LLMs](<https://devfeed.tech/articles/meltkv-memory-aligned-kv-cache-compression-via-variable-bit-packed-vector-quantization-for-long-context-llms-138770.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.18652>)

Author: Jiyun Han, Seongjoon Cho, Seungkyu Choi

Published: 2026-11-17T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [quantization](<https://devfeed.tech/topics/quantization.md>), [LLM optimization](<https://devfeed.tech/topics/llm-optimization.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>), [Cache](<https://devfeed.tech/topics/cache.md>), [Compression](<https://devfeed.tech/topics/compression.md>), [long-context](<https://devfeed.tech/topics/long-context.md>)

Tags: [alignment](<https://devfeed.tech/tags/alignment.md>), [bits](<https://devfeed.tech/tags/bits.md>), [cache](<https://devfeed.tech/tags/cache.md>), [compress](<https://devfeed.tech/tags/compress.md>), [compression](<https://devfeed.tech/tags/compression.md>), [context](<https://devfeed.tech/tags/context.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [data-types](<https://devfeed.tech/tags/data-types.md>), [decoding](<https://devfeed.tech/tags/decoding.md>), [experiments](<https://devfeed.tech/tags/experiments.md>), [framework](<https://devfeed.tech/tags/framework.md>), [index](<https://devfeed.tech/tags/index.md>), [inference](<https://devfeed.tech/tags/inference.md>), [kv-cache](<https://devfeed.tech/tags/kv-cache.md>), [llms](<https://devfeed.tech/tags/llms.md>), [long-context](<https://devfeed.tech/tags/long-context.md>), [long-context-llms](<https://devfeed.tech/tags/long-context-llms.md>), [memory](<https://devfeed.tech/tags/memory.md>), [quantization](<https://devfeed.tech/tags/quantization.md>), [variable](<https://devfeed.tech/tags/variable.md>), [vector](<https://devfeed.tech/tags/vector.md>)

### AI overview

MeltKV targets memory use and decoding latency in long-context LLM serving by combining variable-bit packing of grouped key-channel indices with head-wise codebook sharing. It packs indices into standard data types to eliminate redundant bits and reduces codebook size while maintaining accuracy. The abstract reports compression down to 1 bit and, at comparable perplexity, 48.1% lower memory use and 52.9% lower latency than state-of-the-art vector-quantization methods. These are reported experimental results; the excerpt provides no further evaluation details.

### Source excerpt

arXiv:2611.18652v1 Announce Type: new Abstract: Large Language Models (LLMs) increasingly rely on long contexts, making Key-Value (KV) caches dominate memory usage and inference latency. Existing vector quantization (VQ) methods effectively compress KV caches to sub-4-bit levels but often overlook inefficiencies in memory alignment and decoding speed. This work introduces MeltKV, a memory-latency co-optimized KV cache quantization framework for long-context LLM serving. MeltKV employs a memory-aligned variable-bit index packing scheme that adaptively assigns bit-widths to grouped key channels and compactly stores indices within standard data-types, eliminating redundant bits. Additionally, head-wise codebook sharing reduces the total codebook size while maintaining accuracy. Experiments demonstrate that MeltKV effectively compresses KV representations to sub-4-bit levels down to 1 bit and, at comparable perplexity, achieves 48.1% lower memory usage and 52.9% lower latency than state-of-the-art VQ methods.

## SplitZip: Ultra Fast Lossless KV Compression for Disaggregated LLM Serving

DevFeed: [SplitZip: Ultra Fast Lossless KV Compression for Disaggregated LLM Serving](<https://devfeed.tech/articles/splitzip-ultra-fast-lossless-kv-compression-for-disaggregated-llm-serving-136961.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2605.01708>)

Author: Yipin Guo, Siddharth Joshi

Published: 2026-11-17T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Distributed Computing](<https://devfeed.tech/sources/arxiv-distributed-computing.md>)

Topics: [Compression](<https://devfeed.tech/topics/compression.md>), [LLM optimization](<https://devfeed.tech/topics/llm-optimization.md>), [LLM Techniques](<https://devfeed.tech/topics/llm-techniques.md>), [GPU optimization](<https://devfeed.tech/topics/gpu-optimization.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>)

Tags: [bf16](<https://devfeed.tech/tags/bf16.md>), [compression](<https://devfeed.tech/tags/compression.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-dc](<https://devfeed.tech/tags/cs-dc.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [decoding](<https://devfeed.tech/tags/decoding.md>), [disaggregation](<https://devfeed.tech/tags/disaggregation.md>), [encoding](<https://devfeed.tech/tags/encoding.md>), [fp8](<https://devfeed.tech/tags/fp8.md>), [gpu](<https://devfeed.tech/tags/gpu.md>), [kv-cache](<https://devfeed.tech/tags/kv-cache.md>), [llm-serving](<https://devfeed.tech/tags/llm-serving.md>), [throughput](<https://devfeed.tech/tags/throughput.md>)

### AI overview

SplitZip targets the KV-cache transfer bottleneck in disaggregated LLM serving with a GPU-friendly lossless codec that preserves tensors bitwise and does not modify model execution. It encodes frequent floating-point exponents using a calibrated top-16 fixed-length codebook, while rare exponents go through a sparse position/value escape stream; this avoids online histogramming and enables efficient GPU encoding and decoding. On real BF16 activations, the reported compression and decompression throughputs are 613.3 and 2181.8 GB/s. End-to-end experiments report up to 1.32x faster BF16 KV transfer, 1.30x faster time to first token, and 1.23x higher request throughput. The approach also extends to FP8 KV caches, with up to 1.14x compression over native E5M2.

### Source excerpt

arXiv:2605.01708v4 Announce Type: replace Abstract: Contemporary systems serving large language models (LLMs) have adopted prefill-decode disaggregation to load-balance between the compute-bound prefill phase and the memory-bound decode phase. Under this design, prefill workers generate a KV cache that must be transferred to decode workers before generation can begin. With these workers residing on different physical systems, this transfer becomes a significant bottleneck to serving LLMs at scale, especially for long-input and agentic workloads. Existing lossless codecs are unsuitable here as they primarily target offline weight compression, run on CPUs, or use variable-length coding whose compression cannot keep up with KV production during prefill. We introduce SplitZip, a GPU-friendly lossless compressor for KV cache transfer that preserves KV tensors bitwise and integrates into existing serving frameworks without modifying model execution. SplitZip exploits redundancy in floating-point exponents of KV activations, encoding frequent exponent values with fixed-length codes and routing rare exponents through a sparse escape stream of (position, value). A calibrated top-16 exponent codebook eliminates online histogramming, while the regular dense path and sparse escape correction make both encoding and decoding efficient on GPUs. On real BF16 activation tensors, SplitZip achieves $613.3$ GB/s compression throughput and $2181.8$ GB/s decompression throughput, outperforming prior lossless compressors on the critical codec path. End-to-end transfer experiments show up to $1.32\times$ speedup for BF16 KV cache transfer, $1.30\times$ speedup for TTFT, and $1.23\times$ increase in Request Throughput. The same approach extends to FP8 KV caches, providing up to $1.14\times$ compression over native E5M2. Code is available at https://github.com/Intelligent-Microsystems-Lab/SplitZip

## Practical Semantic Communication for General-Purpose Networks: Compression of High-Dimensional Embeddings Under Quantization

DevFeed: [Practical Semantic Communication for General-Purpose Networks: Compression of High-Dimensional Embeddings Under Quantization](<https://devfeed.tech/articles/practical-semantic-communication-for-general-purpose-networks-compression-of-high-dimensional-embeddings-under-quantization-139103.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.20863>)

Author: Leonardo Roque, Louis-Adrien Dufr\\\`ene, Quentin Lampin, Gr\\'egoire Lefebvre, Mohamad Assaad

Published: 2026-11-17T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Compression](<https://devfeed.tech/topics/compression.md>), [Embeddings](<https://devfeed.tech/topics/embeddings.md>), [quantization](<https://devfeed.tech/topics/quantization.md>), [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>)

Tags: [channel](<https://devfeed.tech/tags/channel.md>), [classification](<https://devfeed.tech/tags/classification.md>), [communication](<https://devfeed.tech/tags/communication.md>), [compression](<https://devfeed.tech/tags/compression.md>), [continuous](<https://devfeed.tech/tags/continuous.md>), [cs-it](<https://devfeed.tech/tags/cs-it.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [embeddings](<https://devfeed.tech/tags/embeddings.md>), [errors](<https://devfeed.tech/tags/errors.md>), [experiments](<https://devfeed.tech/tags/experiments.md>), [framework](<https://devfeed.tech/tags/framework.md>), [gaussian](<https://devfeed.tech/tags/gaussian.md>), [information](<https://devfeed.tech/tags/information.md>), [math-it](<https://devfeed.tech/tags/math-it.md>), [matrix](<https://devfeed.tech/tags/matrix.md>), [model](<https://devfeed.tech/tags/model.md>), [networks](<https://devfeed.tech/tags/networks.md>), [quantization](<https://devfeed.tech/tags/quantization.md>), [semantic](<https://devfeed.tech/tags/semantic.md>), [text](<https://devfeed.tech/tags/text.md>), [text-classification](<https://devfeed.tech/tags/text-classification.md>), [text-embeddings](<https://devfeed.tech/tags/text-embeddings.md>), [wireless](<https://devfeed.tech/tags/wireless.md>)

### AI overview

The paper proposes a framework for sending task-relevant information in high-dimensional embeddings over wireless channels despite quantization and channel errors. Based on a relaxed Gaussian Information Bottleneck model, it supports non-Gaussian targets, derives closed-form gradients for optimizing the compression matrix, and gives bounds relating continuous representations to discrete noisy performance. In text-classification experiments, it reportedly outperforms task-agnostic quantization-based compression, preserving semantic fidelity at low bit budgets. The supplied abstract does not include implementation details or quantitative results.

### Source excerpt

arXiv:2611.20863v1 Announce Type: cross Abstract: This paper introduces a practical framework for semantic communication over wireless channels, focusing on the compression of high-dimensional embeddings under quantization and channel errors. Building upon a relaxed Gaussian Information Bottleneck (GIB) model, the approach provides analytical solutions for task-relevant information preservation, even with non-Gaussian targets. The framework relaxes traditional Gaussian assumptions, enabling effective compression for discrete and complex data such as text embeddings. It derives closed-form gradients for optimizing the compression matrix and establishes bounds linking continuous representations to discrete, noisy system performance. Comprehensive experiments on text classification tasks demonstrate that the proposed method outperforms task-agnostic quantization-based compression methods, maintaining high semantic fidelity at low bit budgets per embedding. The approach offers a scalable, interpretable, and resource-efficient solution for deploying semantic communication in general-purpose networks.

## Hybrid Compression for Memory-Efficient Matrix-Vector Multiplication

DevFeed: [Hybrid Compression for Memory-Efficient Matrix-Vector Multiplication](<https://devfeed.tech/articles/hybrid-compression-for-memory-efficient-matrix-vector-multiplication-138788.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.18792>)

Author: Felipe A. Louza, Giovanni Manzini, Guilherme P. Telles

Published: 2026-11-17T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Compression](<https://devfeed.tech/topics/compression.md>), [Embeddings](<https://devfeed.tech/topics/embeddings.md>)

Tags: [abstract](<https://devfeed.tech/tags/abstract.md>), [algorithms](<https://devfeed.tech/tags/algorithms.md>), [alphabet](<https://devfeed.tech/tags/alphabet.md>), [also](<https://devfeed.tech/tags/also.md>), [ans](<https://devfeed.tech/tags/ans.md>), [applying](<https://devfeed.tech/tags/applying.md>), [approach](<https://devfeed.tech/tags/approach.md>), [arithmetic](<https://devfeed.tech/tags/arithmetic.md>), [arxiv](<https://devfeed.tech/tags/arxiv.md>), [bandwidth](<https://devfeed.tech/tags/bandwidth.md>), [can](<https://devfeed.tech/tags/can.md>), [coder](<https://devfeed.tech/tags/coder.md>), [column](<https://devfeed.tech/tags/column.md>), [combines](<https://devfeed.tech/tags/combines.md>), [component](<https://devfeed.tech/tags/component.md>), [compression](<https://devfeed.tech/tags/compression.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [hybrid](<https://devfeed.tech/tags/hybrid.md>), [matrix](<https://devfeed.tech/tags/matrix.md>), [memory](<https://devfeed.tech/tags/memory.md>), [vector](<https://devfeed.tech/tags/vector.md>)

### AI overview

The hybrid strategy improves mm-RePair's compression of real-valued machine-learning matrices by splitting the CSRV sequence into entries suited to RePair grammar compression and remaining entries encoded with ans-fold. It also sorts and delta encodes each row and remaps the alphabet to a contiguous range, while enabling matrix-vector multiplication directly on compressed data. In experiments on word embeddings and transformer hidden states, it was more compact than original mm-RePair on every input and reduced peak multiplication memory, with the largest gains on dense matrices with many distinct values and competitive multiplication times. It beat the column-reordering variant on most matrices; combining both approaches further compressed sparse matrices but greatly increased compression time. Its thread scaling matched original mm-RePair.

### Source excerpt

arXiv:2611.18792v1 Announce Type: new Abstract: Iterative machine-learning algorithms repeatedly access large real-valued matrices through matrix-vector products, and their performance is limited by memory bandwidth rather than arithmetic throughput. Lossless compression schemes that support computation directly over the compressed data reduce both the data moved and the memory footprint of these workloads. The mm-RePair framework [11] follows this approach by applying the RePair grammar compressor to the Compressed Sparse Row/Value (CSRV) representation of a matrix and performing matrix-vector multiplications directly on the compressed data. Its effectiveness, however, degrades on matrices that are dense or contain many distinct values. In this paper we present a hybrid strategy for the mm-RePair framework that combines grammar and entropy compression. The CSRV sequence is partitioned into a grammar-compressible component, formed by the entries that mm-RePair can compress effectively, and an entropy-compressible component holding the remaining entries, which is encoded with the ans-fold coder [24]. Compression improves further by sorting and delta encoding the entries of each row in the CSRV representation, and by remapping the alphabet to a contiguous range. Experiments on real-valued machine-learning matrices, including word embeddings and transformer hidden states, show that our hybrid strategy is more compact than the original \mmRepair on every input and reduces the peak memory of multiplication, with the largest gains on dense matrices with many distinct values, at competitive multiplication times. It also outperforms the column-reordering variant of \mmRepair on most of the matrices, and combining the two is even more compact on the sparse ones, at the cost of a much longer compression time. Finally, our strategy scales with the number of threads as well as the original \mmRepair does.

## RACE: Ranking-Uncertainty Aware Capacity-Adaptive Customization for Efficient Sequential Recommendation

DevFeed: [RACE: Ranking-Uncertainty Aware Capacity-Adaptive Customization for Efficient Sequential Recommendation](<https://devfeed.tech/articles/race-ranking-uncertainty-aware-capacity-adaptive-customization-for-efficient-sequential-recommendation-135178.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.18244>)

Author: Tianqi Liu, Kaipu Sun, Kairui Fu, Wenkai Wang, Sihao Liu, Hui Zang, Ran Chen, Tao Wang, Shengyu Zhang

Published: 2026-11-16T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [personalization](<https://devfeed.tech/topics/personalization.md>), [Compression](<https://devfeed.tech/topics/compression.md>)

Tags: [adaptive](<https://devfeed.tech/tags/adaptive.md>), [capacity](<https://devfeed.tech/tags/capacity.md>), [compression](<https://devfeed.tech/tags/compression.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-ir](<https://devfeed.tech/tags/cs-ir.md>), [customization](<https://devfeed.tech/tags/customization.md>), [data](<https://devfeed.tech/tags/data.md>), [datasets](<https://devfeed.tech/tags/datasets.md>), [efficiency](<https://devfeed.tech/tags/efficiency.md>), [inference](<https://devfeed.tech/tags/inference.md>), [input](<https://devfeed.tech/tags/input.md>), [intrinsic](<https://devfeed.tech/tags/intrinsic.md>), [layer](<https://devfeed.tech/tags/layer.md>), [lightweight](<https://devfeed.tech/tags/lightweight.md>), [list](<https://devfeed.tech/tags/list.md>), [metric](<https://devfeed.tech/tags/metric.md>), [model](<https://devfeed.tech/tags/model.md>), [model-compression](<https://devfeed.tech/tags/model-compression.md>), [race](<https://devfeed.tech/tags/race.md>), [ranking](<https://devfeed.tech/tags/ranking.md>), [recommendation](<https://devfeed.tech/tags/recommendation.md>), [uncertainty](<https://devfeed.tech/tags/uncertainty.md>)

### AI overview

RACE adapts sequential recommendation models to each instance's apparent complexity, aiming to avoid wasting computation on simple cases or underserving complex ones. It measures ranking uncertainty from the distributional stability of top-K predictions, then uses a lightweight self-supervised predictor on early-layer features to estimate that uncertainty. The estimate guides two forms of compression: a hypernetwork-style extractor identifies redundant model channels, and importance-aware progressive filtering reduces input-sequence data. The abstract reports better accuracy, efficiency, and adaptability than state-of-the-art baselines across four datasets; it provides no dataset names or quantitative results.

### Source excerpt

arXiv:2611.18244v1 Announce Type: cross Abstract: Efficient sequential recommendation has emerged as a promising paradigm for providing real-time personalized service. To deploy models under resource-constrained environment, model compression becomes a necessity, yet existing methods typically apply uniform compression ratios, overlooking the heterogeneity in user intent complexity. This "one-size-fits-all" approach leads to a fundamental Capacity-Complexity Mismatch, where computational resources are either wasted on simple instances or insufficient for complex ones. To address this, we propose RACE (Ranking-uncertainty Aware Capacity-Adaptive Customization for Efficient Sequential Recommendation), a framework that dynamically adapts both model structures and input sequences with uncertainty-aware compression ratios. To ensure capacity-adaptive inference, RACE captures the model's intrinsic uncertainty of its predictions and transforms it into a proactive resource allocator. Specifically, we derive a list-wise ranking-uncertainty metric from the distributional stability of top-$K$ predictions, which serves as a measurable proxy for instance complexity. A lightweight predictor is then trained via self-supervision to anticipate this uncertainty from early-layer features. Guided by this signal, RACE performs instance-specific slimming in two dimensions: a hypernetwork-style channel sensitivity extractor identifies parameter redundancy, while importance-aware progressive sequence filtering mitigates data redundancy. Extensive experiments on four popular datasets demonstrate that RACE outperforms state-of-the-art baselines on accuracy, efficiency, and adaptability.

## RecMem: Recurrent Memory Compression for Long-Sequence Generative Recommendation

DevFeed: [RecMem: Recurrent Memory Compression for Long-Sequence Generative Recommendation](<https://devfeed.tech/articles/recmem-recurrent-memory-compression-for-long-sequence-generative-recommendation-135145.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.17921>)

Author: Haoyi Hu, Gaoyang Guo, Yunjia Xi, Wen Chen, Weiwen Liu, Jian Wu, Yuning Jiang, Weinan Zhang

Published: 2026-11-16T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [Compression](<https://devfeed.tech/topics/compression.md>), [personalization](<https://devfeed.tech/topics/personalization.md>)

Tags: [compression](<https://devfeed.tech/tags/compression.md>), [consistency](<https://devfeed.tech/tags/consistency.md>), [context](<https://devfeed.tech/tags/context.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-ir](<https://devfeed.tech/tags/cs-ir.md>), [decoder](<https://devfeed.tech/tags/decoder.md>), [fixed](<https://devfeed.tech/tags/fixed.md>), [framework](<https://devfeed.tech/tags/framework.md>), [generative](<https://devfeed.tech/tags/generative.md>), [history](<https://devfeed.tech/tags/history.md>), [independent](<https://devfeed.tech/tags/independent.md>), [information](<https://devfeed.tech/tags/information.md>), [interaction](<https://devfeed.tech/tags/interaction.md>), [invariance](<https://devfeed.tech/tags/invariance.md>), [layers](<https://devfeed.tech/tags/layers.md>), [memory](<https://devfeed.tech/tags/memory.md>), [memory-compression](<https://devfeed.tech/tags/memory-compression.md>), [orders](<https://devfeed.tech/tags/orders.md>), [predictions](<https://devfeed.tech/tags/predictions.md>), [recommendation](<https://devfeed.tech/tags/recommendation.md>), [recurrent](<https://devfeed.tech/tags/recurrent.md>), [sequence](<https://devfeed.tech/tags/sequence.md>)

### AI overview

RecMem compresses long recommendation histories into a fixed set of memory slots, keeping memory state and decoder context independent of history length. A bifurcated encoder-decoder with slot-wise forget-gated updates learns predictive information through recommendation supervision; stream-consistency training aligns sequential and packed processing. On MerRec, 16 slots and a 216-token decoder context retained 87.9% of HSTU-1200's Recall@10 and 97.0% of its Recall@200, with the paper reporting better results than PHSTU and evaluated recent-only and retrieval baselines. The abstract also reports a horizon-independent state-discrepancy bound under contraction and approximate semigroup consistency.

### Source excerpt

arXiv:2611.17921v1 Announce Type: new Abstract: Generative recommendation benefits from long interaction histories, but the cost of processing them directly grows with history length. We propose RecMem, a recurrent memory compression framework guided by three desiderata: space invariance, semantic fidelity, and lifelong evolvability. RecMem compresses historical segments into a fixed set of memory slots using a bifurcated encoder-decoder with shared lower layers and slot-wise forget-gated updates. Recommendation supervision trains the memory to preserve predictive information, while stream-consistency training aligns predictions from sequential and packed processing. By construction, the memory state and decoder context are independent of history length; under contraction and approximate semigroup consistency, we further derive a horizon-independent bound on the state discrepancy between these two processing orders. On MerRec, RecMem uses 16 memory slots and a 216-token decoder context, compared with 1,200 tokens for HSTU-1200, while retaining 87.9% of HSTU-1200's Recall@10 and 97.0% of its Recall@200. It outperforms PHSTU and the evaluated recent-only and retrieval baselines, and analyses of distant history show that archived interactions contribute predictive signal.

## JSONBench is back, and ClickHouse still leads

DevFeed: [JSONBench is back, and ClickHouse still leads](<https://devfeed.tech/articles/jsonbench-is-back-and-clickhouse-still-leads-136654.md>)

Original publisher: [Read original article](<https://clickhouse.com/blog/jsonbench-is-back-clickhouse-still-leads>)

Author: Tom Schreiber; Mark Needham

Published: 2026-11-16T00:30:00Z

Content type: news

Language: en

Sources: [ClickHouse Blog](<https://devfeed.tech/sources/clickhouse-blog.md>)

Topics: [clickhouse](<https://devfeed.tech/topics/clickhouse.md>), [JSON](<https://devfeed.tech/topics/json.md>), [DuckDB](<https://devfeed.tech/topics/duckdb.md>), [reproducibility](<https://devfeed.tech/topics/reproducibility.md>), [elasticsearch](<https://devfeed.tech/topics/elasticsearch.md>), [Compression](<https://devfeed.tech/topics/compression.md>)

Tags: [benchmark](<https://devfeed.tech/tags/benchmark.md>), [clickhouse](<https://devfeed.tech/tags/clickhouse.md>), [database](<https://devfeed.tech/tags/database.md>), [engines](<https://devfeed.tech/tags/engines.md>), [json](<https://devfeed.tech/tags/json.md>)

### AI overview

In the described billion-document Bluesky event aggregation, ClickHouse returned results in 661 ms on a cold run, while MongoDB took 32 minutes 26 seconds on the same machine type and published configurations. The article attributes ClickHouse's JSON analytics performance to columnar storage, compression, and parallel processing that reads the fields a query needs. JSONBench has expanded from five to 27 database engines; ClickHouse also leads the broader comparison, which reports combined scores rather than query-time multipliers and penalizes unsuccessful queries and incomplete loads. The benchmark is open source, with public data, queries, configurations, and results, so others can reproduce measurements or contribute systems and improvements. The article also cites ClickHouse JSON changes including faster reads, lower memory use, indexes, Parquet support, and type hints without rewriting historical data.

### Source excerpt

JSONBench now compares 27 database engines. ClickHouse still leads on a billion JSON documents. Explore the results and run the benchmark yourself.

## Beyond Compressing Everything: Contrastive Memory Learning for Selective Context Compression

DevFeed: [Beyond Compressing Everything: Contrastive Memory Learning for Selective Context Compression](<https://devfeed.tech/articles/beyond-compressing-everything-contrastive-memory-learning-for-selective-context-compression-132229.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.16384>)

Author: Sikui Zhang, Haowei Liu, Ziyun Gan, Chunfeng Yuan, Guangze Gao, Nian Liu, Boyu Cai, Bo Li, Xiaolei Lv, Jun Gao, Bing Li, Weiming Hu

Published: 2026-11-13T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [AI Context Management](<https://devfeed.tech/topics/ai-context-management.md>), [Representation learning](<https://devfeed.tech/topics/representation-learning.md>), [Compression](<https://devfeed.tech/topics/compression.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>)

Tags: [attention](<https://devfeed.tech/tags/attention.md>), [benchmarks](<https://devfeed.tech/tags/benchmarks.md>), [compression](<https://devfeed.tech/tags/compression.md>), [context-compression](<https://devfeed.tech/tags/context-compression.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cl](<https://devfeed.tech/tags/cs-cl.md>), [learning](<https://devfeed.tech/tags/learning.md>), [memory](<https://devfeed.tech/tags/memory.md>)

### AI overview

CoMem compresses context selectively for each query, aiming to retain relevant evidence within a limited number of memory slots. It trains document-and-query-conditioned memories with Contrastive Compressor Learning and mines hard negatives by pairing one document with different queries, encouraging distinct evidence selection for each. It also enables bidirectional attention among memory slots to address information-flow limits from causal masking. The abstract reports that CoMem outperforms existing context compression methods across compression ratios on question answering and RULER-NIAH benchmarks; it provides no numerical results or further evaluation details.

### Source excerpt

arXiv:2611.16384v1 Announce Type: cross Abstract: As large language models (LLMs) advance toward increasingly long inputs, soft context compression offers a promising approach to efficient inference by converting long inputs into compact memory slots. Under limited memory budgets, effective context compression remains challenging because task-relevant evidence varies substantially across queries, ranging from localized evidence to multiple distant spans or information dispersed throughout the full document. Instead of compressing everything indiscriminately, we propose CoMem, a query-guided selective context compression framework that learns memory representations conditioned on both the document and the query. CoMem develops Contrastive Compressor Learning to achieve query-sensitive yet context-preserving memory representations. To inject a strong prior over query-dependent evidence distributions, we devise a same-context hard negative mining strategy by pairing a single document with multiple queries, requiring the compressor to selectively retain information according to each query's distinct evidence needs. To address the restricted information flow imposed by causal masking, we further enable bidirectional attention among memory slots. Experiments demonstrate that CoMem consistently outperforms existing context compression methods across compression ratios on question answering and RULER-NIAH benchmarks. Our code is available at https://github.com/scar-on/CoMem.

## The risk of KV cache compression

DevFeed: [The risk of KV cache compression](<https://devfeed.tech/articles/the-risk-of-kv-cache-compression-132812.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2607.01520>)

Author: Lukas Haverbeck, Carmen Amo Alonso, Andres Felipe Posada-Moreno, Sebastian Trimpe, Marco Pavone

Published: 2026-11-13T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Compression](<https://devfeed.tech/topics/compression.md>), [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [model architecture](<https://devfeed.tech/topics/model-architecture.md>), [Algorithms](<https://devfeed.tech/topics/algorithms.md>)

Tags: [algorithm](<https://devfeed.tech/tags/algorithm.md>), [attention](<https://devfeed.tech/tags/attention.md>), [autoregressive](<https://devfeed.tech/tags/autoregressive.md>), [cache](<https://devfeed.tech/tags/cache.md>), [compression](<https://devfeed.tech/tags/compression.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [design](<https://devfeed.tech/tags/design.md>), [design-principles](<https://devfeed.tech/tags/design-principles.md>), [future](<https://devfeed.tech/tags/future.md>), [guarantees](<https://devfeed.tech/tags/guarantees.md>), [inference](<https://devfeed.tech/tags/inference.md>), [information](<https://devfeed.tech/tags/information.md>), [intrinsic](<https://devfeed.tech/tags/intrinsic.md>), [kv-cache](<https://devfeed.tech/tags/kv-cache.md>), [masking](<https://devfeed.tech/tags/masking.md>), [minimax](<https://devfeed.tech/tags/minimax.md>), [novel](<https://devfeed.tech/tags/novel.md>), [practical](<https://devfeed.tech/tags/practical.md>), [prefill](<https://devfeed.tech/tags/prefill.md>), [principles](<https://devfeed.tech/tags/principles.md>), [risk](<https://devfeed.tech/tags/risk.md>)

### AI overview

The paper characterizes the minimax risk of KV cache compression in terms of a cache's intrinsic compressibility, aiming to identify when accurate compression is possible. It finds that when attention can sharply retrieve individual tokens, compression can uniformly outperform random sampling only by using information about future queries, regardless of how redundant the cache is. For causal masking, the authors derive design principles intended to map efficiently to prefill and autoregressive decoding with minimax-optimal risk, and instantiate them in a practical algorithm. The supplied abstract does not provide implementation details or empirical results.

### Source excerpt

arXiv:2607.01520v2 Announce Type: replace Abstract: Transformer inference on long sequences is expensive because softmax attention repeatedly reads from a large KV cache. The prevalent approach to this bottleneck is KV cache compression, which replaces the full cache with a compact summary. Despite its practical importance, the design of such summaries is largely driven by empirical experimentation. On the theoretical side, existing results show that KV cache compression can be impossible in the worst case, but offer little systematic guidance for designing algorithms in regimes where accurate compression is possible. We bridge this gap by characterizing the minimax risk of KV cache compression in terms of the intrinsic compressibility of a cache, revealing when and how accurate compression is possible. In particular, we show that, once attention can sharply retrieve individual tokens, compression can only uniformly outperform random sampling by exploiting information about future queries, regardless how redundant the cache is inherently. Additionally, our results yield novel design principles for KV cache compression under causal masking that map efficiently to prefill and autoregressive decoding while achieving minimax-optimal risk. We instantiate these principles in a practical algorithm and illustrate how our theoretical findings can be put to practical use. Overall, our results opens a principled avenue for KV cache compression with theoretical guarantees.

## STeP-C: Real-Time Neural Compression for Distributed Fiber-Optic Sensing at Scale

DevFeed: [STeP-C: Real-Time Neural Compression for Distributed Fiber-Optic Sensing at Scale](<https://devfeed.tech/articles/step-c-real-time-neural-compression-for-distributed-fiber-optic-sensing-at-scale-133178.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.17126>)

Author: Tingfeng Li, Ming-Fang Huang, Shaobo Han, Yue Tian

Published: 2026-11-13T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Compression](<https://devfeed.tech/topics/compression.md>), [real-time](<https://devfeed.tech/topics/real-time.md>)

Tags: [acoustic-sensing](<https://devfeed.tech/tags/acoustic-sensing.md>), [acquisition](<https://devfeed.tech/tags/acquisition.md>), [arithmetic](<https://devfeed.tech/tags/arithmetic.md>), [arithmetic-coding](<https://devfeed.tech/tags/arithmetic-coding.md>), [buffer](<https://devfeed.tech/tags/buffer.md>), [compression](<https://devfeed.tech/tags/compression.md>), [cs-cv](<https://devfeed.tech/tags/cs-cv.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [distributed](<https://devfeed.tech/tags/distributed.md>), [entropy](<https://devfeed.tech/tags/entropy.md>), [entropy-coding](<https://devfeed.tech/tags/entropy-coding.md>), [fiber](<https://devfeed.tech/tags/fiber.md>), [fp16](<https://devfeed.tech/tags/fp16.md>), [gpu](<https://devfeed.tech/tags/gpu.md>), [inference](<https://devfeed.tech/tags/inference.md>), [live](<https://devfeed.tech/tags/live.md>), [neural](<https://devfeed.tech/tags/neural.md>), [offline](<https://devfeed.tech/tags/offline.md>), [optimized](<https://devfeed.tech/tags/optimized.md>), [real-time](<https://devfeed.tech/tags/real-time.md>), [scale](<https://devfeed.tech/tags/scale.md>), [sensing](<https://devfeed.tech/tags/sensing.md>)

### AI overview

STeP-C demonstrates real-time neural compression on a live distributed acoustic sensing system, reporting 98.85% file-size reduction at 31.04 dB full-band PSNR and 535 MB/s throughput, above the stated 324 MB/s requirement for 1 kHz, 81 km streams. Its design separates spatial encoding, temporal prediction, and residual coding. To address entropy coding as the streaming bottleneck, it uses GPU-side symbol computation, cached probability tables, parallel GIL-free arithmetic coding, FP16 inference, batched coding, and a GPU ring buffer. The abstract reports better performance than conventional codecs at practical streaming bitrates, transfer to unseen channel ranges within the same acquisition system without retraining, and detector consistency of 96.62% high-confidence class agreement and 100% Jaccard similarity on detected event types. These claims are limited to the reported testbed and excerpt.

### Source excerpt

arXiv:2611.17126v1 Announce Type: cross Abstract: Distributed fiber-optic sensing (DFOS) technologies such as Distributed Acoustic Sensing (DAS) generate continuous, high-resolution spatio-temporal data at rates exceeding hundreds of megabytes per second, creating severe challenges for real-time storage and transmission. While neural compression methods can achieve strong rate-distortion performance on sensing data, prior approaches either require per-segment retraining incompatible with streaming, or have only been validated offline. None have been demonstrated end-to-end on a continuously streaming DAS acquisition system at scale. We present STeP-C (Spatio-Temporal Predictive Compression), a neural compression framework that decouples spatial encoding, temporal prediction, and entropy-optimized residual coding, trained end-to-end with a rate-distortion objective motivated by conditional entropy reduction. Critically, we identify entropy coding as the dominant bottleneck for streaming deployment. We address it through GPU-side symbol computation, cached probability tables, and parallel GIL-free arithmetic coding, together with FP16 inference and batched entropy coding, enabling a full real-time pipeline built around a GPU ring buffer. We validate this pipeline on a live DAS testbed, achieving a 98.85% file size reduction at 31.04 dB full-band PSNR and an average throughput of 535 MB/s, exceeding the 324 MB/s requirement for 1 kHz, 81 km DAS streams. We further show that STeP-C substantially outperforms conventional codecs at practical streaming bitrates, can be applied to unseen channel ranges within the same acquisition system without retraining, and largely maintains downstream detector consistency, with 96.62% high-confidence class agreement and 100% Jaccard similarity on detected event types.

## FedAdaVR: Adaptive Variance Reduction for Robust Federated Learning under Limited Client Participation

DevFeed: [FedAdaVR: Adaptive Variance Reduction for Robust Federated Learning under Limited Client Participation](<https://devfeed.tech/articles/fedadavr-adaptive-variance-reduction-for-robust-federated-learning-under-limited-client-participation-131507.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2601.22204>)

Author: S M Ruhul Kabir Howlader, Xiao Chen, Yifei Xie, Lu Liu

Published: 2026-11-13T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Distributed Computing](<https://devfeed.tech/sources/arxiv-distributed-computing.md>), [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Federated Learning](<https://devfeed.tech/topics/federated-learning.md>), [quantization](<https://devfeed.tech/topics/quantization.md>), [Compression](<https://devfeed.tech/topics/compression.md>)

Tags: [adaptive](<https://devfeed.tech/tags/adaptive.md>), [algorithm](<https://devfeed.tech/tags/algorithm.md>), [bias](<https://devfeed.tech/tags/bias.md>), [challenges](<https://devfeed.tech/tags/challenges.md>), [client](<https://devfeed.tech/tags/client.md>), [clients](<https://devfeed.tech/tags/clients.md>), [communication](<https://devfeed.tech/tags/communication.md>), [compression](<https://devfeed.tech/tags/compression.md>), [convergence](<https://devfeed.tech/tags/convergence.md>), [convex](<https://devfeed.tech/tags/convex.md>), [cs-dc](<https://devfeed.tech/tags/cs-dc.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [datasets](<https://devfeed.tech/tags/datasets.md>), [errors](<https://devfeed.tech/tags/errors.md>), [experiments](<https://devfeed.tech/tags/experiments.md>), [federated](<https://devfeed.tech/tags/federated.md>), [federated-learning](<https://devfeed.tech/tags/federated-learning.md>), [fp16](<https://devfeed.tech/tags/fp16.md>), [memory](<https://devfeed.tech/tags/memory.md>), [reduction](<https://devfeed.tech/tags/reduction.md>), [robust](<https://devfeed.tech/tags/robust.md>), [variance](<https://devfeed.tech/tags/variance.md>)

### AI overview

FedAdaVR addresses errors from partial client participation by combining an adaptive optimizer with variance reduction: it reuses each absent client's most recently stored update to emulate that client's participation. The paper reports that, with Adagrad, the method removes first-order participation bias and improves monotonically with cohort size. FedAdaVR-Quant compresses historical updates in server memory while keeping client-server communication in full precision; reported memory reductions are 50% with FP16, 75% with Int8, and 87.5% with Int4, with competitive and sometimes superior model performance. Experiments on vision and NLP datasets are reported to outperform state-of-the-art baselines. These are claims from the supplied abstract excerpt.

### Source excerpt

arXiv:2601.22204v3 Announce Type: replace-cross Abstract: Federated learning (FL) encounters substantial challenges due to heterogeneity, leading to gradient noise, client drift, and partial client participation errors, the last of which is the most pervasive but remains insufficiently addressed in current literature. In this paper, we propose FedAdaVR, a novel FL algorithm aimed at solving heterogeneity issues caused by partial client participation by incorporating an adaptive optimiser with a variance reduction technique. This method takes advantage of the most recent stored updates from clients, even when they are absent from the current training round, thereby emulating their presence. Furthermore, we propose FedAdaVR-Quant, a server-side memory compression variant that stores historical client updates in quantised form (while keeping client-server communication in full precision), reducing the server memory footprint of FedAdaVR by 50% (FP16), 75% (Int8), and 87.5% (Int4) while maintaining highly competitive, and occasionally superior model performance. We analyse the non-convex convergence behaviour of FedAdaVR with Adagrad and prove that our proposed framework eliminates the first-order partial client participation bias while improving monotonically with cohort size. Extensive experiments conducted on multiple vision and NLP datasets demonstrate that FedAdaVR consistently outperforms state-of-the-art baseline methods.

## GeoPair: Geometry-Preserving Cross-Layer Factorization for Training-Free Transformer Compression

DevFeed: [GeoPair: Geometry-Preserving Cross-Layer Factorization for Training-Free Transformer Compression](<https://devfeed.tech/articles/geopair-geometry-preserving-cross-layer-factorization-for-training-free-transformer-compression-132827.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2609.25963>)

Author: Baher Mohammad, Ammar Ali, Stamatios Lefkimmiatis

Published: 2026-11-13T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Machine Learning](<https://devfeed.tech/sources/arxiv-machine-learning.md>)

Topics: [Compression](<https://devfeed.tech/topics/compression.md>), [Neural Network](<https://devfeed.tech/topics/neural-network.md>)

Tags: [abstract](<https://devfeed.tech/tags/abstract.md>), [activation](<https://devfeed.tech/tags/activation.md>), [approach](<https://devfeed.tech/tags/approach.md>), [architectures](<https://devfeed.tech/tags/architectures.md>), [art](<https://devfeed.tech/tags/art.md>), [arxiv](<https://devfeed.tech/tags/arxiv.md>), [better](<https://devfeed.tech/tags/better.md>), [compression](<https://devfeed.tech/tags/compression.md>), [cross](<https://devfeed.tech/tags/cross.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [engineering](<https://devfeed.tech/tags/engineering.md>), [foundation](<https://devfeed.tech/tags/foundation.md>), [framework](<https://devfeed.tech/tags/framework.md>), [free](<https://devfeed.tech/tags/free.md>), [functional](<https://devfeed.tech/tags/functional.md>), [independent](<https://devfeed.tech/tags/independent.md>), [isolation](<https://devfeed.tech/tags/isolation.md>), [layer](<https://devfeed.tech/tags/layer.md>), [layers](<https://devfeed.tech/tags/layers.md>), [training](<https://devfeed.tech/tags/training.md>), [transformer](<https://devfeed.tech/tags/transformer.md>)

### AI overview

GeoPair is a training-free transformer compression framework that sequentially optimizes cross-layer weight pairings and shared-dictionary factorizations. It selects structurally compatible projections to preserve each layer's distinct calibration geometry, instead of forcing adjacent layers to share a basis or heuristically merging activation statistics. Combined with structured sparsity, the approach produces weight decompositions that the abstract says preserve functional fidelity. It reports outperforming independent structured decompositions and alternative pairwise factorizations across architectures, scales, and modalities; the excerpt gives no quantitative results.

### Source excerpt

arXiv:2609.25963v2 Announce Type: replace Abstract: Transformer architectures exhibit cross-layer redundancies, yet post-training compression pipelines typically optimize layers in isolation or rely on heuristic grouping strategies that disregard layer-specific activation geometries. We introduce a principled, training-free framework that sequentially optimizes cross-layer weight pairings and shared-dictionary factorizations. Rather than forcing weights of adjacent layers to share a basis or heuristically merging activation statistics, our approach identifies structurally compatible projections and learns a shared representation that better preserves each layer's distinct calibration geometry. Coupled with structured sparsity, this yields highly efficient weight decompositions without sacrificing functional fidelity. Across diverse architectures, scales, and modalities, our method achieves state-of-the-art results, consistently outperforming independent structured weight decompositions and alternative pairwise weight factorizations, which operate under heuristic grouping strategies. By replacing heuristic engineering strategies with a convergent, optimization-driven pipeline, we establish a theoretically grounded foundation for scalable, transformer compression across different modalities.

## A Multi-Word Tokenizer for 204 Languages That Needs Fewer Tokens Than GPT-4o's, Including on English

DevFeed: [A Multi-Word Tokenizer for 204 Languages That Needs Fewer Tokens Than GPT-4o's, Including on English](<https://devfeed.tech/articles/a-multi-word-tokenizer-for-204-languages-that-needs-fewer-tokens-than-gpt-4o-s-including-on-english-128249.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.13059>)

Author: Johin Johny, Ankur Ghosh, Shervin Sam

Published: 2026-11-11T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Artificial Intelligence](<https://devfeed.tech/sources/arxiv-artificial-intelligence.md>)

Topics: [tokenization](<https://devfeed.tech/topics/tokenization.md>), [Compression](<https://devfeed.tech/topics/compression.md>)

Tags: [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cl](<https://devfeed.tech/tags/cs-cl.md>), [english](<https://devfeed.tech/tags/english.md>), [fewer](<https://devfeed.tech/tags/fewer.md>), [gpt](<https://devfeed.tech/tags/gpt.md>), [languages](<https://devfeed.tech/tags/languages.md>), [tokens](<https://devfeed.tech/tags/tokens.md>)

### AI overview

On FLORES-200, Havon v4 uses fewer tokens than GPT-4o's o200k_base on 161 of 204 languages at the same 200,019-token vocabulary size, including all 25 Indian languages (median 0.74x GPT-4o's count); it also uses 6.1% fewer on English. The tokenizer adds a multi-word SuperBPE training stage to v3. Results vary by language and vocabulary size: gains are largest for English, Hindi and Romance languages, while Khmer, Burmese and Tibetan gain little and lose subword merges. Havon v5 improves code compression over v4, using 1.03-1.22x GPT-4o's tokens versus v4's 1.11-1.29x, while retaining reported gains on Indian languages and English. These are compression measurements; the authors have not evaluated models trained with v4 or v5.

### Source excerpt

arXiv:2611.13059v1 Announce Type: cross Abstract: Languages of India need 1.6 to 14 times as many tokens as English on GPT-4o's tokenizer for the same content, which makes them more expensive and slower to serve. We train tokenizers from scratch on human-written text in about 200 languages. Havon v3 is a byte-level BPE tokenizer with 262,144 tokens and a pre-tokenizer that keeps combining marks with their letters and chunks scripts written without spaces. On FLORES-200 devtest it needs fewer tokens than GPT-4o's o200k_base on all 25 languages of India (median 0.74 of GPT-4o's count) and on 172 of 204 languages, but 7% more on English. Havon v4 adds a second, multi-word training stage (SuperBPE) to v3. At exactly GPT-4o's vocabulary size of 200,019 tokens, v4 needs 6.1% fewer tokens than GPT-4o on English and fewer on 161 of 204 languages and all 25 languages of India (median 0.74); at the same size, v3 needs 9.4% more on English and wins on 133. With 262,144 tokens, v4 needs 9.6% fewer tokens on English and fewer on 182 of 204 languages, and with 524,288 tokens it needs fewer on 203 of 204. The gains are largest for English, Hindi and the Romance languages; scripts without spaces, such as Khmer, Burmese and Tibetan, gain little and lose subword merges. Havon v5 replaces v4's Python-only code sample with permissively licensed code in 15 languages: on code it needs 1.03 to 1.22 times GPT-4o's tokens, against 1.11 to 1.29 for v4, while keeping all 25 languages of India (median 0.74, all significant under a sentence-level bootstrap) and English (0.94). These are compression results only: we have not yet evaluated a model trained with v4 or v5.

## SplitCom: Communication-efficient Split Federated Fine-tuning of LLMs via Temporal Compression

DevFeed: [SplitCom: Communication-efficient Split Federated Fine-tuning of LLMs via Temporal Compression](<https://devfeed.tech/articles/splitcom-communication-efficient-split-federated-fine-tuning-of-llms-via-temporal-compression-127831.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2602.10564>)

Author: Tao Li, Yulin Tang, Yiyang Song, Cong Wu, Xihui Liu, Pan Li, Xianhao Chen

Published: 2026-11-11T05:00:00Z

Content type: article

Language: en

Sources: [arXiv: Networking](<https://devfeed.tech/sources/arxiv-networking.md>)

Topics: [Federated Learning](<https://devfeed.tech/topics/federated-learning.md>), [Fine-tuning](<https://devfeed.tech/topics/fine-tuning.md>), [LLM Techniques](<https://devfeed.tech/topics/llm-techniques.md>), [Compression](<https://devfeed.tech/topics/compression.md>), [Distributed Systems & Parallel Computing](<https://devfeed.tech/topics/distributed-systems-parallel-computing.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>)

Tags: [activation](<https://devfeed.tech/tags/activation.md>), [adaptive](<https://devfeed.tech/tags/adaptive.md>), [architecture](<https://devfeed.tech/tags/architecture.md>), [communication](<https://devfeed.tech/tags/communication.md>), [compression](<https://devfeed.tech/tags/compression.md>), [cs-ni](<https://devfeed.tech/tags/cs-ni.md>), [experiments](<https://devfeed.tech/tags/experiments.md>), [exploits](<https://devfeed.tech/tags/exploits.md>), [federated](<https://devfeed.tech/tags/federated.md>), [federated-learning](<https://devfeed.tech/tags/federated-learning.md>), [fine-tuning](<https://devfeed.tech/tags/fine-tuning.md>), [framework](<https://devfeed.tech/tags/framework.md>), [llms](<https://devfeed.tech/tags/llms.md>), [performance](<https://devfeed.tech/tags/performance.md>), [reduction](<https://devfeed.tech/tags/reduction.md>), [reinforcement-learning](<https://devfeed.tech/tags/reinforcement-learning.md>), [resource](<https://devfeed.tech/tags/resource.md>), [server](<https://devfeed.tech/tags/server.md>), [simulations](<https://devfeed.tech/tags/simulations.md>), [split](<https://devfeed.tech/tags/split.md>), [temporal](<https://devfeed.tech/tags/temporal.md>)

### AI overview

SplitCom reduces communication in split federated fine-tuning by uploading activations only when they deviate noticeably from prior epochs, exploiting temporal redundancy. It uses adaptive upload thresholds controlled either by bang-bang control or DDPG-based reinforcement learning, and dimensionality reduction to ease client memory demands. Its U-shaped variant is designed so the server never accesses client labels. Simulations and laboratory experiments report up to 98.8% lower uplink communication in the standard configuration and up to 95.8% lower total communication in the U-shaped variant, without noticeably compromising model performance.

### Source excerpt

arXiv:2602.10564v3 Announce Type: replace Abstract: Federated fine-tuning of on-device large language models (LLMs) mitigates privacy concerns by preventing raw data sharing. However, the intensive computational and memory demands pose significant challenges for resource-constrained edge devices. To overcome these limitations, split federated learning (SFL) emerges as a promising solution that partitions the model into lightweight client-side and compute-intensive server-side sub-models, thus offloading the primary training workload to a powerful server. Nevertheless, the high-dimensional activation exchanges in SFL lead to excessive communication overhead. To overcome this, we propose SplitCom, a communication-efficient SFL framework for LLMs that exploits temporal redundancy in activations across consecutive training epochs. Inspired by video compression, the core innovation of our framework lies in selective activation uploading only when a noticeable deviation from previous epochs occurs. To balance communication efficiency and learning performance, we introduce two adaptive threshold control schemes based on 1) bang-bang control or 2) deep deterministic policy gradient (DDPG)-based reinforcement learning. Moreover, we implement dimensionality reduction techniques to alleviate client-side memory requirements. Furthermore, we extend SplitCom to the U-shape architecture, ensuring the server never accesses clients' labels. Extensive simulations and laboratory experiments demonstrate that SplitCom reduces uplink communication costs by up to 98.8\,\% in its standard configuration and total communication costs by up to 95.8\,\% in its U-shape variant without noticeably compromising model performance.

[Next page](<https://devfeed.tech/topics/compression.md?cursor=WyIyMDI2LTExLTExVDA1OjAwOjAwKzAwOjAwIiwgIjdjZDAyNzJkLTExOGItNDMyZS04ZmNiLWM5Mzc3MGM3ZTBmNiJd>)