# packing

Published articles for packing.

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

## SHORE: Stability-Hybrid On-Policy Reinforcement Learning for Online Bin Packing

DevFeed: [SHORE: Stability-Hybrid On-Policy Reinforcement Learning for Online Bin Packing](<https://devfeed.tech/articles/shore-stability-hybrid-on-policy-reinforcement-learning-for-online-bin-packing-120737.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.08364>)

Author: Chang Shi, Ann Majewicz Fey

Published: 2026-11-06T05:00:00Z

Content type: article

Language: en

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

Topics: [Reinforcement learning](<https://devfeed.tech/topics/reinforcement-learning.md>), [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [Neural Network](<https://devfeed.tech/topics/neural-network.md>), [Robot Manipulation](<https://devfeed.tech/topics/robot-manipulation.md>), [Isaac](<https://devfeed.tech/topics/isaac.md>)

Tags: [3d](<https://devfeed.tech/tags/3d.md>), [actor](<https://devfeed.tech/tags/actor.md>), [bin-packing](<https://devfeed.tech/tags/bin-packing.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [cs-ro](<https://devfeed.tech/tags/cs-ro.md>), [hybrid](<https://devfeed.tech/tags/hybrid.md>), [learning](<https://devfeed.tech/tags/learning.md>), [online](<https://devfeed.tech/tags/online.md>), [packing](<https://devfeed.tech/tags/packing.md>), [policy](<https://devfeed.tech/tags/policy.md>), [reinforcement-learning](<https://devfeed.tech/tags/reinforcement-learning.md>), [simulation](<https://devfeed.tech/tags/simulation.md>)

### AI overview

SHORE combines a learned probability that a box placement will remain stable with an on-policy estimate of its return if stable. A 3D U-Net learns from executed outcomes and simulated alternatives; a discrete actor-critic uses stability predictions in action values and its policy prior, with gates to limit the effect of an untrained or changing model. Deployment adds no search or simulation overhead. In Isaac Gym, SHORE reached 73% space utilization by 0.4M environment steps, versus about 1M for the same learner without stability modeling. On the expanded 60-box set, it achieved 0.734 utilization versus GOPT's 0.588, with higher placement-failure frequency (1.03% versus 0.63%), indicating a tradeoff of more risk for more packed volume. Ablations suggest simulated alternatives help early supervision, while executed outcomes sustain later performance. The authors also report a real-robot demonstration.

### Source excerpt

arXiv:2611.08364v1 Announce Type: cross Abstract: Online 3D bin packing with a processing buffer requires choosing which box to pack, its orientation, and its placement while keeping the stack stable under unknown future arrivals. A collapse ends the episode and forfeits the remaining volume. Existing methods use geometric rules or hard stability masks, limiting how risk can influence the joint choice of box, orientation, and placement. We propose Stability-Hybrid On-Policy Reinforcement Learning (SHORE), which combines a learned stability probability of a placement with an on-policy estimate of its return conditional on stability. A 3D U-Net learns stability from executed outcomes and simulated alternatives, revealing outcomes of actions underrepresented in early rollouts. A discrete on-policy actor--critic uses stability predictions in both its action values and policy prior, with gates that limit the effect of an untrained or changing model. In deployment, SHORE does not require additional search or simulation overhead. In Isaac Gym, SHORE reaches 73\% space utilization by 0.4M environment steps, while the same learner without stability modeling reaches that level at about 1M. On the expanded 60-box set, SHORE achieves 0.734 space utilization versus 0.588 for the GOPT state-of-the-art baseline, with a higher placement-failure frequency (1.03\% vs. 0.63\%), trading more risk for greater packed volume. The results show faster learning and higher utilization on the more demanding realistic box arrival set. Ablations suggest that simulated alternatives provide useful early supervision, while executed outcomes alone sustain later performance. We also demonstrate the method on a real robot to verify the simplicity of integrating the method to real bin packing pipelines.

## FP4Quant: Efficient 4-Bit Communication Compression for Model-Parallel LLM Training

DevFeed: [FP4Quant: Efficient 4-Bit Communication Compression for Model-Parallel LLM Training](<https://devfeed.tech/articles/fp4quant-efficient-4-bit-communication-compression-for-model-parallel-llm-training-117467.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.04323>)

Author: Man Liu, Xingchen Liu, Hairui Zhao, Guangming Tan, Dingwen Tao

Published: 2026-11-04T05:00:00Z

Content type: article

Language: en

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

Topics: [LLM optimization](<https://devfeed.tech/topics/llm-optimization.md>), [quantization](<https://devfeed.tech/topics/quantization.md>), [distributed-training](<https://devfeed.tech/topics/distributed-training.md>), [Compression](<https://devfeed.tech/topics/compression.md>), [GPU optimization](<https://devfeed.tech/topics/gpu-optimization.md>), [Large Language Model](<https://devfeed.tech/topics/llm.md>), [LLM Techniques](<https://devfeed.tech/topics/llm-techniques.md>), [Distributed Systems & Parallel Computing](<https://devfeed.tech/topics/distributed-systems-parallel-computing.md>), [qwen](<https://devfeed.tech/topics/qwen.md>)

Tags: [communication](<https://devfeed.tech/tags/communication.md>), [compression](<https://devfeed.tech/tags/compression.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [distributed](<https://devfeed.tech/tags/distributed.md>), [fp4](<https://devfeed.tech/tags/fp4.md>), [gpu](<https://devfeed.tech/tags/gpu.md>), [h100](<https://devfeed.tech/tags/h100.md>), [hybrid](<https://devfeed.tech/tags/hybrid.md>), [llm](<https://devfeed.tech/tags/llm.md>), [llm-training](<https://devfeed.tech/tags/llm-training.md>), [packing](<https://devfeed.tech/tags/packing.md>), [parallel](<https://devfeed.tech/tags/parallel.md>), [quality](<https://devfeed.tech/tags/quality.md>), [quantization](<https://devfeed.tech/tags/quantization.md>), [qwen](<https://devfeed.tech/tags/qwen.md>), [scale](<https://devfeed.tech/tags/scale.md>), [throughput](<https://devfeed.tech/tags/throughput.md>), [training](<https://devfeed.tech/tags/training.md>)

### AI overview

FP4Quant targets communication overhead in model-parallel LLM training with a micro-grouped FP4 compression framework. It combines dual-scale quantization to preserve local tensor distributions, warp-cooperative packing for sub-byte data, and a GPU-aware two-level FWHT pipeline to reduce quantization error. On NVIDIA H100 GPUs with Qwen-7B, the authors report up to 1.56x end-to-end throughput versus BF16, with a trade-off between throughput and model quality. The excerpt does not quantify that quality trade-off.

### Source excerpt

arXiv:2611.04323v1 Announce Type: new Abstract: The rapid scaling of large language models (LLMs) has made communication overhead a critical bottleneck in distributed model-parallel training. Although existing low-precision compression methods can effectively reduce communication costs, further reducing the communication footprint to ultra-low-bit formats such as FP4 introduces new challenges, including limited numerical range, sub-byte storage overhead, and increased computational cost from fine-grained quantization. In this paper, we present \textbf{FP4Quant}, a micro-grouped FP4 compression framework for efficient model-parallel LLM training. FP4Quant integrates a micro-grouped dual-scale FP4 quantization mechanism to preserve local tensor distribution characteristics, a warp-cooperative FP4 packing scheme to enable efficient sub-byte data representation, and a hybrid two-level FWHT pipeline to mitigate quantization errors with GPU-aware execution optimization. Experiments on NVIDIA H100 GPUs with Qwen-7B demonstrate up to 1.56$\times$ end-to-end throughput improvement over BF16, revealing a trade-off between throughput and model quality.

## Perseus: A Full-Stack System for Homomorphic Model Inference

DevFeed: [Perseus: A Full-Stack System for Homomorphic Model Inference](<https://devfeed.tech/articles/perseus-a-full-stack-system-for-homomorphic-model-inference-116153.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.04488>)

Author: Alessandro Zirilli, Davide Marincione, Evgenios M. Kornaropoulos, Giuseppe Ateniese, Emanuele Rodol\\\`a

Published: 2026-11-04T05:00:00Z

Content type: release

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: [FHE](<https://devfeed.tech/topics/fhe.md>), [Cryptography](<https://devfeed.tech/topics/cryptography.md>), [LLM optimization](<https://devfeed.tech/topics/llm-optimization.md>), [Advanced AI workflows](<https://devfeed.tech/topics/advanced-ai-workflows.md>)

Tags: [backend](<https://devfeed.tech/tags/backend.md>), [bootstrap](<https://devfeed.tech/tags/bootstrap.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cr](<https://devfeed.tech/tags/cs-cr.md>), [decoding](<https://devfeed.tech/tags/decoding.md>), [encrypted](<https://devfeed.tech/tags/encrypted.md>), [faster](<https://devfeed.tech/tags/faster.md>), [fhe](<https://devfeed.tech/tags/fhe.md>), [full-stack](<https://devfeed.tech/tags/full-stack.md>), [gpt](<https://devfeed.tech/tags/gpt.md>), [homomorphic-encryption](<https://devfeed.tech/tags/homomorphic-encryption.md>), [large-language-models-llms](<https://devfeed.tech/tags/large-language-models-llms.md>), [latency](<https://devfeed.tech/tags/latency.md>), [library](<https://devfeed.tech/tags/library.md>), [model-inference](<https://devfeed.tech/tags/model-inference.md>), [models](<https://devfeed.tech/tags/models.md>), [open](<https://devfeed.tech/tags/open.md>), [optimizations](<https://devfeed.tech/tags/optimizations.md>), [packing](<https://devfeed.tech/tags/packing.md>), [private](<https://devfeed.tech/tags/private.md>), [python](<https://devfeed.tech/tags/python.md>), [token](<https://devfeed.tech/tags/token.md>)

### AI overview

Perseus is an open Python library for end-to-end model inference on fully homomorphically encrypted data. It combines in-kernel regeneration of key-switching keys, a complex Cachemir packing variant, and a bootstrap placer that accounts for packing and magnitude. The abstract reports state-of-the-art latency across multiple models and modalities, with accuracy and perplexity within 3% of plaintext models; for single-token Llama-3-8B decoding, it reports 3.96x higher speed than other implementations. The work also launches a public encrypted GPT-2 speedrun. These performance claims are those reported in the supplied abstract.

### Source excerpt

arXiv:2611.04488v1 Announce Type: new Abstract: Large Language Models (LLMs) have made it exceptionally convenient for users to send their private data to model providers, which process them in plaintext. Fully Homomorphic Encryption (FHE) promises to secure remote inference by computing directly on encrypted data, yet it has remained prohibitively slow. The effort to make it practical remains fragmented, with most existing frameworks being closed-source, which keeps progress from compounding. We present Perseus, a free and open Python software library that supports the full stack of encrypted model inference, from the cryptographic backend and data packing, to encrypted nonlinearities and end-to-end model execution. Perseus ships with built-in novel optimizations, one at each level of the stack: in-kernel regeneration of key-switching keys at the (i) backend level, a complex variant of Cachemir packing at the (ii) data level, and a state-of-the-art bootstrap placer that is both packing- and magnitude-aware at the (iii) model level. Perseus achieves state-of-the-art latency across multiple models and modalities, within $3\%$ of the plaintext models' accuracy and perplexity. Remarkably, it is $3.96\times$ faster at single-token decoding than any other implementation on Llama-3-8B. We initiate, for the first time, a public speedrun on encrypted GPT-2, so that progress can compound.

## Robustness Starts with the Objective: Designing and Certifying Robust LLM-Generated Heuristics

DevFeed: [Robustness Starts with the Objective: Designing and Certifying Robust LLM-Generated Heuristics](<https://devfeed.tech/articles/robustness-starts-with-the-objective-designing-and-certifying-robust-llm-generated-heuristics-113777.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2611.00504>)

Author: Pantea Karimi, Roozbeh Bostandoost, Dany Rouhana, Siva Kesava Reddy Kakarla, Pooria Namyar, Mohammad Alizadeh, Venkat Arun, Behnaz Arzani

Published: 2026-11-03T05:00:00Z

Content type: article

Language: en

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

Topics: [AI Engineering](<https://devfeed.tech/topics/ai-engineering.md>), [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [AI Infrastructure](<https://devfeed.tech/topics/ai-infrastructure.md>)

Tags: [adversarial](<https://devfeed.tech/tags/adversarial.md>), [azure](<https://devfeed.tech/tags/azure.md>), [benchmarks](<https://devfeed.tech/tags/benchmarks.md>), [better](<https://devfeed.tech/tags/better.md>), [case-studies](<https://devfeed.tech/tags/case-studies.md>), [computer](<https://devfeed.tech/tags/computer.md>), [conditional](<https://devfeed.tech/tags/conditional.md>), [confidence](<https://devfeed.tech/tags/confidence.md>), [cost](<https://devfeed.tech/tags/cost.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-cl](<https://devfeed.tech/tags/cs-cl.md>), [cs-dc](<https://devfeed.tech/tags/cs-dc.md>), [cs-pf](<https://devfeed.tech/tags/cs-pf.md>), [framework](<https://devfeed.tech/tags/framework.md>), [gap](<https://devfeed.tech/tags/gap.md>), [gpu](<https://devfeed.tech/tags/gpu.md>), [llm](<https://devfeed.tech/tags/llm.md>), [optimization](<https://devfeed.tech/tags/optimization.md>), [optimum](<https://devfeed.tech/tags/optimum.md>), [packing](<https://devfeed.tech/tags/packing.md>), [performance](<https://devfeed.tech/tags/performance.md>), [regressions](<https://devfeed.tech/tags/regressions.md>), [robust](<https://devfeed.tech/tags/robust.md>), [robustness](<https://devfeed.tech/tags/robustness.md>)

### AI overview

TAILOR guides LLM-generated heuristics by minimizing paired CVaR: the average cost difference from an existing baseline over the worst fraction of workloads. Unlike average-performance objectives, this targets severe regressions; unlike percentile objectives, it accounts for their severity beyond a threshold. The authors report that negative paired CVaR guarantees improvement in average performance and tail optimality gap relative to the unknown optimum, without computing that optimum. Across three case studies, TAILOR had the best mean-tail trade-off among evaluated synthesis objectives. On six of seven optimization benchmarks, its heuristics were certified with 99% confidence to improve both measures over the baseline, and they outperformed known robust, human-designed alternatives in the tail while delivering better or comparable average improvement.

### Source excerpt

arXiv:2611.00504v1 Announce Type: cross Abstract: LLMs are increasingly used to design heuristics for computer systems. However, most synthesis frameworks evaluate candidates by their average performance, which can mask severe regressions on specific workloads. Even adding adversarial training inputs doesn't eliminate these regressions, and percentile objectives (e.g., optimizing for the 95th percentile) overlook how severe regressions can be beyond the selected threshold. We present TAILOR, a framework that guides LLM synthesis using the Conditional Value at Risk (CVaR) of the cost difference between each candidate and an existing baseline on the same workload. This paired CVaR measures the average difference over the worst fraction of workloads. We show that a negative paired CVaR guarantees improvements in both average performance and tail optimality gap relative to the unknown optimum, without ever computing that optimum. We find that TAILOR achieves the best mean--tail trade-off among the synthesis objectives we evaluate across three case studies: Azure GPU batch admission, Azure VM packing, and server placement. Tested across seven diverse optimization benchmarks, TAILOR produces heuristics that are certified with 99% confidence to improve both average performance and tail optimality gaps over the baseline in six of them. On these six benchmarks, TAILOR also outperforms known robust, human-designed alternatives in the tail while achieving better or comparable average-performance improvement over the baseline.

## Handwriting to pen plotting

DevFeed: [Handwriting to pen plotting](<https://devfeed.tech/articles/handwriting-to-pen-plotting-108433.md>)

Original publisher: [Read original article](<https://blog.adafruit.com/2026/10/30/handwriting-to-pen-plotting/>)

Author: Anne Barela

Published: 2026-10-30T15:25:33Z

Content type: news

Language: en

Sources: [Adafruit Industries](<https://devfeed.tech/sources/adafruit-industries-makers-hackers-artists-designers-and-engineers.md>)

Topics: [plotting](<https://devfeed.tech/topics/plotting.md>), [Markdown](<https://devfeed.tech/topics/markdown.md>)

Tags: [ai-assistant](<https://devfeed.tech/tags/ai-assistant.md>), [art](<https://devfeed.tech/tags/art.md>), [automation](<https://devfeed.tech/tags/automation.md>), [code](<https://devfeed.tech/tags/code.md>), [handwriting](<https://devfeed.tech/tags/handwriting.md>), [kitty](<https://devfeed.tech/tags/kitty.md>), [markdown](<https://devfeed.tech/tags/markdown.md>), [newsletter](<https://devfeed.tech/tags/newsletter.md>), [notes](<https://devfeed.tech/tags/notes.md>), [packing](<https://devfeed.tech/tags/packing.md>), [pen](<https://devfeed.tech/tags/pen.md>), [pen-plotter](<https://devfeed.tech/tags/pen-plotter.md>), [plotting](<https://devfeed.tech/tags/plotting.md>), [software](<https://devfeed.tech/tags/software.md>), [svg](<https://devfeed.tech/tags/svg.md>), [tool](<https://devfeed.tech/tags/tool.md>), [tools](<https://devfeed.tech/tags/tools.md>), [writing](<https://devfeed.tech/tags/writing.md>)

### AI overview

Daniel Catt's markdown-to-handwriting tool turns formatted text into dynamically generated, single-stroke cursive SVG paths for a pen plotter, rather than using fonts. It supports headings, underlining, strikethrough and bold; bold is drawn by tracing a word more than once, while italic is omitted. Catt uses a Bantam Tools NextDraw 8511 driven by his AI assistant, Kitty, for todo lists, envelopes and packing lists. The plotter is slower than a nearby laser printer, but pages can be collected after a job finishes. A separate node-based workflow chains a random seed through a "70s Pop" generator, kaleidoscope, and scaling and placement steps; Catt also plotted the graph itself as SVG. He reports spending more time tuning plotting tools than plotting this month.

### Source excerpt

Daniel Catt built a markdown to handwriting tool, written up in his newsletter at newsletter.revdancatt.com. He runs a Bantam Tools NextDraw 8511 pen plotter driven by Kitty, his AI assistant, which writes out his morning todo lists, addresses envelopes and now handles packing lists on A6 index cards and A5 notes. It is slower than [...]

## Do Bio-Inspired Algorithms Save Energy in Data Centres? Separating Server Packing from Migration Search in Virtual Machine Consolidation

DevFeed: [Do Bio-Inspired Algorithms Save Energy in Data Centres? Separating Server Packing from Migration Search in Virtual Machine Consolidation](<https://devfeed.tech/articles/do-bio-inspired-algorithms-save-energy-in-data-centres-separating-server-packing-from-migration-search-in-virtual-machine-consolidation-101612.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2610.34102>)

Author: Yichiet Aun, Yu Hang Woo

Published: 2026-10-27T04:00:00Z

Content type: article

Language: en

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

Topics: [Sustainable Computing](<https://devfeed.tech/topics/sustainable-computing.md>), [virtual machines](<https://devfeed.tech/topics/virtual-machines.md>), [Algorithms](<https://devfeed.tech/topics/algorithms.md>), [reproducibility](<https://devfeed.tech/topics/reproducibility.md>)

Tags: [algorithm](<https://devfeed.tech/tags/algorithm.md>), [algorithms](<https://devfeed.tech/tags/algorithms.md>), [arxiv](<https://devfeed.tech/tags/arxiv.md>), [bio](<https://devfeed.tech/tags/bio.md>), [comparisons](<https://devfeed.tech/tags/comparisons.md>), [completion](<https://devfeed.tech/tags/completion.md>), [control](<https://devfeed.tech/tags/control.md>), [cs-dc](<https://devfeed.tech/tags/cs-dc.md>), [cs-et](<https://devfeed.tech/tags/cs-et.md>), [cs-sy](<https://devfeed.tech/tags/cs-sy.md>), [data](<https://devfeed.tech/tags/data.md>), [data-centre](<https://devfeed.tech/tags/data-centre.md>), [data-centres](<https://devfeed.tech/tags/data-centres.md>), [delivered](<https://devfeed.tech/tags/delivered.md>), [demand](<https://devfeed.tech/tags/demand.md>), [eess-sy](<https://devfeed.tech/tags/eess-sy.md>), [electricity](<https://devfeed.tech/tags/electricity.md>), [energy](<https://devfeed.tech/tags/energy.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [migration](<https://devfeed.tech/tags/migration.md>), [packing](<https://devfeed.tech/tags/packing.md>), [search](<https://devfeed.tech/tags/search.md>), [server](<https://devfeed.tech/tags/server.md>), [virtual-machine](<https://devfeed.tech/tags/virtual-machine.md>)

### AI overview

In the tested CloudSim Plus scenarios, initial VM packing and deactivating empty hosts account for most energy savings; migration search usually adds no benefit and sometimes raises energy use. A matched control kept the same packing and deactivation mechanisms while omitting migration search. Across 24 comparisons of ACO, PSO and a modified genetic algorithm, search increased consumption in 11, matched the control in seven and reduced it in six. Five apparent reductions disappeared after accounting for unfinished tasks and delivered computational work. Only ACO under light heterogeneous load retained a statistically significant 3.22 percentage-point improvement (95% CI 2.96-3.47). The control saved 0.8-81.4% versus a spreading baseline, with savings declining sharply as load rose, and completed workloads without migrations. Results are limited to the two fleet configurations and four workload regimes examined.

### Source excerpt

arXiv:2610.34102v1 Announce Type: new Abstract: Energy-efficient virtual machine (VM) consolidation matters as data centre electricity demand rises. Bio-inspired metaheuristics, including ant colony optimisation (ACO), particle swarm optimisation (PSO), and genetic algorithms, have reported substantial energy savings. However, these improvements are commonly evaluated against conventional allocation policies that differ in both initial VM placement and subsequent migration optimisation, making it difficult to determine the contribution of metaheuristic search. This study reproduces an existing consolidation implementation in CloudSim Plus and introduces a matched control that retains the same initial packing and empty-host deactivation mechanisms but excludes migration search. ACO, PSO, and a modified genetic algorithm (MGA) are evaluated across two fleet configurations and four workload regimes, with ten stochastic replications per condition. The matched control achieves energy savings of 0.8-81.4% relative to the spreading baseline, the amount available falling sharply as load rises, without migrations or incomplete workload execution. Across 24 comparisons, metaheuristic search increases energy consumption in 11 cases, produces identical outcomes in seven, and reduces consumption in six. Of these, ten increases and five reductions remain statistically significant after Holm correction. Five of the six apparent reductions coincide with unfinished tasks and disappear after normalisation by delivered computational work. Only ACO under light heterogeneous load retains a 3.22-percentage-point improvement (95% CI 2.96-3.47). These findings demonstrate that initial packing and host deactivation account for most of the measured energy savings in the examined implementations, highlighting the importance of matched control conditions and workload-completion measures for reliable evaluation of VM consolidation algorithms.

## Homomorphic Encryption with Coset Code Construction

DevFeed: [Homomorphic Encryption with Coset Code Construction](<https://devfeed.tech/articles/homomorphic-encryption-with-coset-code-construction-96611.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2610.31437>)

Author: Haodong Yang, Venkata Gandikota

Published: 2026-10-23T04:00:00Z

Content type: article

Language: en

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

Topics: [homomorphic encryption](<https://devfeed.tech/topics/homomorphic-encryption.md>), [Encryption](<https://devfeed.tech/topics/encryption.md>)

Tags: [addition](<https://devfeed.tech/tags/addition.md>), [code](<https://devfeed.tech/tags/code.md>), [cs-cr](<https://devfeed.tech/tags/cs-cr.md>), [cs-it](<https://devfeed.tech/tags/cs-it.md>), [encryption](<https://devfeed.tech/tags/encryption.md>), [homomorphic-encryption](<https://devfeed.tech/tags/homomorphic-encryption.md>), [math-it](<https://devfeed.tech/tags/math-it.md>), [packing](<https://devfeed.tech/tags/packing.md>)

### AI overview

The construction packs k message symbols into a ciphertext vector of length n using coset codes, reducing the number of ciphertext vectors needed to transmit a message record for a given field and vector length. A compatible family of evaluation codes supports public addition and multiplication up to a predetermined degree. The authors prove bounded chosen-plaintext security under the coordinate-label distinguishability assumption.

### Source excerpt

arXiv:2610.31437v1 Announce Type: cross Abstract: Homomorphic encryption enables computation on input data without decrypting it. This paper extends a construction based on scalar message codes, utilizing coset codes to pack $k$ message symbols into a ciphertext vector of length $n$. For a given field and vector length, this packing method reduces the number of ciphertext vectors required to transmit a message record. A compatible family of evaluation codes supports public addition and multiplication operations up to a predetermined degree. We prove that the scheme achieves bounded chosen-plaintext security under the coordinate-label distinguishability assumption.

## SAR: Learning Global Item-Space Matching and Geometry-Aware Placement Ranking for Buffered Online 3D Bin Packing

DevFeed: [SAR: Learning Global Item-Space Matching and Geometry-Aware Placement Ranking for Buffered Online 3D Bin Packing](<https://devfeed.tech/articles/sar-learning-global-item-space-matching-and-geometry-aware-placement-ranking-for-buffered-online-3d-bin-packing-93549.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2610.25795>)

Author: Xinpeng Lei, Jin Zhang, Yizhe Yang, Yihong Duan, Fucheng Wan

Published: 2026-10-21T04:00:00Z

Content type: article

Language: en

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

Topics: [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [Reinforcement learning](<https://devfeed.tech/topics/reinforcement-learning.md>)

Tags: [3d](<https://devfeed.tech/tags/3d.md>), [bin-packing](<https://devfeed.tech/tags/bin-packing.md>), [buffer](<https://devfeed.tech/tags/buffer.md>), [container](<https://devfeed.tech/tags/container.md>), [counterfactual-analysis](<https://devfeed.tech/tags/counterfactual-analysis.md>), [cs-ai](<https://devfeed.tech/tags/cs-ai.md>), [cs-lg](<https://devfeed.tech/tags/cs-lg.md>), [datasets](<https://devfeed.tech/tags/datasets.md>), [evaluation](<https://devfeed.tech/tags/evaluation.md>), [experiments](<https://devfeed.tech/tags/experiments.md>), [features](<https://devfeed.tech/tags/features.md>), [framework](<https://devfeed.tech/tags/framework.md>), [future](<https://devfeed.tech/tags/future.md>), [generators](<https://devfeed.tech/tags/generators.md>), [global](<https://devfeed.tech/tags/global.md>), [horizon](<https://devfeed.tech/tags/horizon.md>), [learning](<https://devfeed.tech/tags/learning.md>), [loader](<https://devfeed.tech/tags/loader.md>), [matching](<https://devfeed.tech/tags/matching.md>), [online](<https://devfeed.tech/tags/online.md>), [packing](<https://devfeed.tech/tags/packing.md>), [ranking](<https://devfeed.tech/tags/ranking.md>)

### AI overview

Spatial Affordance Retrieval (SAR) separates buffered online 3D bin packing into coordinated item selection and placement ranking. A voxel-token encoder lets the picker match buffered items to residual space without item-specific placement proposals. After selection, the loader ranks feasible rotation-position candidates using their geometry, local geometric features, and comparisons among candidates, without direct access to global spatial tokens. The stages are jointly optimized for a shared long-horizon packing return. On Random, JD, and SF datasets, the best variants exceeded the strongest evaluated online baselines by 0.96-4.20 percentage points in utilization with no buffer (K=1); with K=5, best rates were 93.55%, 91.04%, and 90.41%, respectively. Ablations support complementary roles for the two stages, and experiments report compatibility with multiple candidate generators.

### Source excerpt

arXiv:2610.25795v1 Announce Type: cross Abstract: Online 3D bin packing with a bounded buffer requires irreversible item-selection and placement decisions under limited knowledge of future arrivals. Feasibility alone is insufficient, as feasible actions can produce different residual-space structures and subsequent packing opportunities. A central architectural question is how to allocate spatial reasoning between item selection and placement evaluation. We propose Spatial Affordance Retrieval (SAR), a novel learning-based framework that coordinates global item-space matching with geometry-aware placement ranking. A voxel-token spatial encoder represents 3D container occupancy, enabling a picker to match buffered items to the residual space without requiring item-specific placement proposals as inputs. After item selection, a loader ranks feasible rotation-position candidates using candidate geometry, local geometric features (LGF), and inter-candidate comparison, without directly accessing global spatial tokens. The feasible candidate set connects the two stages, which are jointly optimized under a shared long-horizon packing return. On the Random, JD, and SF datasets, the best SAR variants improve utilization over the strongest evaluated online baselines by 0.96-4.20 percentage points under the no-buffer setting (K=1). With a buffer (K=5), the best utilization rates across SAR variants reach 93.55%, 91.04%, and 90.41%, respectively. Ablations support the complementary roles of global item-space matching and geometry-aware placement evaluation, counterfactual analysis suggests that learned item selection favors greater future packing potential, and experiments demonstrate compatibility with multiple candidate generators.

## KPage: Cross-Block Page Compaction for Hardware-Efficient Sparse Attention

DevFeed: [KPage: Cross-Block Page Compaction for Hardware-Efficient Sparse Attention](<https://devfeed.tech/articles/kpage-cross-block-page-compaction-for-hardware-efficient-sparse-attention-93752.md>)

Original publisher: [Read original article](<https://arxiv.org/abs/2610.28096>)

Author: Junwei Feng, Qiaorui Chen, Chenyu Liang, Zhen Guo, Jingyi Yang, Xiaoling Zhou, Jiahao Zhang, Dongliang Huang, Yiqiao Liao, Chengru Song

Published: 2026-10-21T04:00:00Z

Content type: article

Language: en

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

Topics: [High-Performance Computing](<https://devfeed.tech/topics/high-performance-computing.md>)

Tags: [3d](<https://devfeed.tech/tags/3d.md>), [attention](<https://devfeed.tech/tags/attention.md>), [bandwidth](<https://devfeed.tech/tags/bandwidth.md>), [block](<https://devfeed.tech/tags/block.md>), [blocks](<https://devfeed.tech/tags/blocks.md>), [capacity](<https://devfeed.tech/tags/capacity.md>), [compaction](<https://devfeed.tech/tags/compaction.md>), [cs-dc](<https://devfeed.tech/tags/cs-dc.md>), [cudnn](<https://devfeed.tech/tags/cudnn.md>), [hardware](<https://devfeed.tech/tags/hardware.md>), [kernel](<https://devfeed.tech/tags/kernel.md>), [masking](<https://devfeed.tech/tags/masking.md>), [nvidia](<https://devfeed.tech/tags/nvidia.md>), [open-source](<https://devfeed.tech/tags/open-source.md>), [packing](<https://devfeed.tech/tags/packing.md>), [padding](<https://devfeed.tech/tags/padding.md>), [pages](<https://devfeed.tech/tags/pages.md>), [partitions](<https://devfeed.tech/tags/partitions.md>), [pull-request](<https://devfeed.tech/tags/pull-request.md>), [transformers](<https://devfeed.tech/tags/transformers.md>), [video](<https://devfeed.tech/tags/video.md>)

### AI overview

KPage improves sparse attention efficiency for video Transformer layouts whose selected blocks are only partly occupied. It packs valid 16-row K/V pages across logical-block boundaries into tiles during loading, avoiding compacted tensors in high-bandwidth memory, and routes Q blocks between 64x256 and 128x128 tiles to reduce padding. Exact masking and K/V-owned packing preserve the original sparse graph in forward and backward execution. Against NVIDIA Block-Sparse Attention, reported forward speedups are 1.35x-2.06x across five workloads (1.77x geometric mean), rising to 2.29x on partially occupied 3D layouts; fully occupied blocks are near parity at 1.03x. Physical query-key capacity falls from 334.1G to 177.5G while retaining 160.6G useful interactions. Across eight layouts, KPage reports 1.13x-4.25x forward and 1.02x-2.71x forward-plus-backward speedups over the strongest exact sparse baselines. These are abstract-reported results; open-source release and a cuDNN Frontend pull request are planned after compliance review.

### Source excerpt

arXiv:2610.28096v1 Announce Type: new Abstract: Sparse attention reduces the computational cost of long-sequence video Transformers, but three-dimensional token layouts often leave selected blocks partially occupied, wasting physical tile capacity. We present KPage, a block-sparse attention scheme that packs valid 16-row K/V pages across logical-block boundaries into large tiles directly during loading, without materializing compacted tensors in high-bandwidth memory. KPage also routes Q blocks between 64x256 and 128x128 tiles to reduce query-side padding. Exact masking and K/V-owned page packing preserve the original sparse graph in forward and backward execution. Across five representative workloads, KPage achieves 1.35x-2.06x bare forward-kernel speedup over NVIDIA Block-Sparse Attention (BSA), with a geometric mean of 1.77x. It is near parity for fully occupied blocks (1.03x) but achieves 2.29x geometric-mean speedup on 3D layouts with partial blocks. KPage reduces physical query-key pair capacity from 334.1G to 177.5G while preserving 160.6G useful interactions. Across eight layouts from open-source SSTA, STA, and VSA systems, KPage achieves 1.13x-4.25x forward and 1.02x-2.71x forward-plus-backward speedup over the strongest exact sparse baselines. With independent Q/K partitions, KPage also achieves 1.49x speedup over the strongest exact sparse baseline and 1.57x over BSA on the SVG2 Wan 14B/720p configuration. After compliance review, we will open-source KPage and submit a pull request to NVIDIA cuDNN Frontend.

## Introducing advanced Kubernetes control plane configuration in Amazon EKS

DevFeed: [Introducing advanced Kubernetes control plane configuration in Amazon EKS](<https://devfeed.tech/articles/introducing-advanced-kubernetes-control-plane-configuration-in-amazon-eks-4632.md>)

Original publisher: [Read original article](<https://aws.amazon.com/blogs/containers/introducing-advanced-kubernetes-control-plane-configuration-in-amazon-eks/>)

Author: Ashok Srirama

Published: 2026-08-12T17:23:45Z

Content type: article

Language: en

Sources: [Containers](<https://devfeed.tech/sources/containers.md>)

Topics: [Amazon Elastic Kubernetes Service](<https://devfeed.tech/topics/amazon-elastic-kubernetes-service.md>), [Kubernetes](<https://devfeed.tech/topics/kubernetes.md>), [configuration](<https://devfeed.tech/topics/configuration.md>), [API](<https://devfeed.tech/topics/api.md>), [Availability](<https://devfeed.tech/topics/availability.md>), [Provisioning](<https://devfeed.tech/topics/provisioning.md>)

Tags: [advanced-300](<https://devfeed.tech/tags/advanced-300.md>), [ai-ml](<https://devfeed.tech/tags/ai-ml.md>), [amazon-eks](<https://devfeed.tech/tags/amazon-eks.md>), [amazon-elastic-kubernetes-service](<https://devfeed.tech/tags/amazon-elastic-kubernetes-service.md>), [announcements](<https://devfeed.tech/tags/announcements.md>), [api](<https://devfeed.tech/tags/api.md>), [availability](<https://devfeed.tech/tags/availability.md>), [ci-cd](<https://devfeed.tech/tags/ci-cd.md>), [kubernetes](<https://devfeed.tech/tags/kubernetes.md>), [packing](<https://devfeed.tech/tags/packing.md>), [retention](<https://devfeed.tech/tags/retention.md>), [storage](<https://devfeed.tech/tags/storage.md>)

### AI overview

Amazon EKS introduces advanced Kubernetes control plane configuration, allowing users to set API server, scheduler, and controller manager parameters directly. The article covers pod placement scoring, event retention, and HPA synchronization settings through feature explanations and hands-on walkthroughs.

### Source excerpt

With Amazon EKS, you can now configure Kubernetes control plane components (the API server, scheduler, and controller manager) directly through EKS APIs. This post explains what's configurable and includes two hands-on walkthroughs: enabling MostAllocated bin-packing to optimize pod placement, and tuning event retention duration.

## Solving virtual machine puzzles: How AI is optimizing cloud computing

DevFeed: [Solving virtual machine puzzles: How AI is optimizing cloud computing](<https://devfeed.tech/articles/solving-virtual-machine-puzzles-how-ai-is-optimizing-cloud-computing-6876.md>)

Original publisher: [Read original article](<https://research.google/blog/solving-virtual-machine-puzzles-how-ai-is-optimizing-cloud-computing/>)

Published: 2025-10-17T17:56:35Z

Content type: article

Language: en

Sources: [The latest research from Google](<https://devfeed.tech/sources/the-latest-research-from-google.md>)

Topics: [Artificial Intelligence](<https://devfeed.tech/topics/ai.md>), [Cloud](<https://devfeed.tech/topics/cloud.md>), [Algorithms](<https://devfeed.tech/topics/algorithms.md>), [data centers](<https://devfeed.tech/topics/data-centers.md>), [Server](<https://devfeed.tech/topics/server.md>), [Provisioning](<https://devfeed.tech/topics/provisioning.md>)

Tags: [2025](<https://devfeed.tech/tags/2025.md>), [ai](<https://devfeed.tech/tags/ai.md>), [algorithms](<https://devfeed.tech/tags/algorithms.md>), [algorithms-theory](<https://devfeed.tech/tags/algorithms-theory.md>), [cloud](<https://devfeed.tech/tags/cloud.md>), [cloud-computing](<https://devfeed.tech/tags/cloud-computing.md>), [distributed-systems-parallel-computing](<https://devfeed.tech/tags/distributed-systems-parallel-computing.md>), [efficiency](<https://devfeed.tech/tags/efficiency.md>), [google](<https://devfeed.tech/tags/google.md>), [packing](<https://devfeed.tech/tags/packing.md>), [provisioning](<https://devfeed.tech/tags/provisioning.md>), [research](<https://devfeed.tech/tags/research.md>), [resource](<https://devfeed.tech/tags/resource.md>), [scale](<https://devfeed.tech/tags/scale.md>), [server](<https://devfeed.tech/tags/server.md>), [servers](<https://devfeed.tech/tags/servers.md>), [software-systems-engineering](<https://devfeed.tech/tags/software-systems-engineering.md>), [virtual-machines](<https://devfeed.tech/tags/virtual-machines.md>)

### AI overview

Google Research presents LAVA, a scheduling system that uses AI models to continuously predict and adapt to virtual machine lifetimes. Its NILAS, LAVA, and LARS algorithms improve VM allocation and rescheduling in large cloud data centers, reducing stranded resources and improving server efficiency.

### Source excerpt

Algorithms & Theory

## Packing Matrix-Vector Multiplication in Fully Homomorphic Encryption

DevFeed: [Packing Matrix-Vector Multiplication in Fully Homomorphic Encryption](<https://devfeed.tech/articles/packing-matrix-vector-multiplication-in-fully-homomorphic-encryption-40487.md>)

Original publisher: [Read original article](<https://www.jeremykun.com/2024/09/06/packing-matrix-vector-multiplication-in-fhe/>)

Published: 2024-09-07T04:18:09Z

Content type: tutorial

Language: en

Sources: [Jeremy Kun](<https://devfeed.tech/sources/jeremy-kun.md>)

Topics: [homomorphic encryption](<https://devfeed.tech/topics/homomorphic-encryption.md>), [FHE](<https://devfeed.tech/topics/fhe.md>), [Encryption](<https://devfeed.tech/topics/encryption.md>), [layout](<https://devfeed.tech/topics/layout.md>), [Parallelism](<https://devfeed.tech/topics/parallelism.md>), [Python](<https://devfeed.tech/topics/python.md>)

Tags: [arithmetic](<https://devfeed.tech/tags/arithmetic.md>), [code](<https://devfeed.tech/tags/code.md>), [cryptography](<https://devfeed.tech/tags/cryptography.md>), [data](<https://devfeed.tech/tags/data.md>), [encryption](<https://devfeed.tech/tags/encryption.md>), [fhe](<https://devfeed.tech/tags/fhe.md>), [github-repository](<https://devfeed.tech/tags/github-repository.md>), [homomorphic-encryption](<https://devfeed.tech/tags/homomorphic-encryption.md>), [layout](<https://devfeed.tech/tags/layout.md>), [linear-algebra](<https://devfeed.tech/tags/linear-algebra.md>), [lwe](<https://devfeed.tech/tags/lwe.md>), [mathematics](<https://devfeed.tech/tags/mathematics.md>), [packing](<https://devfeed.tech/tags/packing.md>), [parallel](<https://devfeed.tech/tags/parallel.md>), [programming](<https://devfeed.tech/tags/programming.md>), [python](<https://devfeed.tech/tags/python.md>), [rlwe](<https://devfeed.tech/tags/rlwe.md>), [simd](<https://devfeed.tech/tags/simd.md>), [strategies](<https://devfeed.tech/tags/strategies.md>)

### AI overview

This article explains packing for SIMD-style fully homomorphic encryption. It describes how to arrange plaintext data in RLWE ciphertexts so matrix-vector multiplication requires fewer alignment multiplications and rotations, then introduces two basic packing techniques and a computational model.

### Source excerpt

In my recent overview of homomorphic encryption, I underemphasized the importance of data layout when working with arithmetic (SIMD-style) homomorphic encryption schemes. In the FHE world, the name given to data layout strategies is called "packing," because it revolves around putting multiple plaintext data into RLWE ciphertexts in carefully-chosen ways that mesh well with the operations you'd like to perform. By "mesh well" I mean it reduces the number of extra multiplications and rotations required merely to align data elements properly, rather than doing the actual computation you care about.

## Converting Between Packings in SIMD-Style FHE

DevFeed: [Converting Between Packings in SIMD-Style FHE](<https://devfeed.tech/articles/shift-networks-40486.md>)

Original publisher: [Read original article](<https://www.jeremykun.com/2024/09/02/shift-networks/>)

Published: 2024-09-02T21:01:03Z

Content type: tutorial

Language: en

Sources: [Jeremy Kun](<https://devfeed.tech/sources/jeremy-kun.md>)

Topics: [FHE](<https://devfeed.tech/topics/fhe.md>), [homomorphic encryption](<https://devfeed.tech/topics/homomorphic-encryption.md>), [data](<https://devfeed.tech/topics/data.md>), [Code](<https://devfeed.tech/topics/code.md>)

Tags: [compilers](<https://devfeed.tech/tags/compilers.md>), [encryption](<https://devfeed.tech/tags/encryption.md>), [fhe](<https://devfeed.tech/tags/fhe.md>), [github](<https://devfeed.tech/tags/github.md>), [graph-coloring](<https://devfeed.tech/tags/graph-coloring.md>), [heir](<https://devfeed.tech/tags/heir.md>), [homomorphic-encryption](<https://devfeed.tech/tags/homomorphic-encryption.md>), [mathematics](<https://devfeed.tech/tags/mathematics.md>), [optimization](<https://devfeed.tech/tags/optimization.md>), [packing](<https://devfeed.tech/tags/packing.md>), [permutation](<https://devfeed.tech/tags/permutation.md>), [programming](<https://devfeed.tech/tags/programming.md>), [rlwe](<https://devfeed.tech/tags/rlwe.md>), [simd](<https://devfeed.tech/tags/simd.md>)

### AI overview

This article explains packing in SIMD-style fully homomorphic encryption and focuses on converting between established packings. It introduces a computational model involving RLWE ciphertext vectors, elementwise operations, cyclic rotations, and differing operation costs.

### Source excerpt

In my recent overview of homomorphic encryption, I underemphasized the importance of data layout when working with arithmetic (SIMD-style) homomorphic encryption schemes. In the FHE world, the name given to data layout strategies is called "packing," because it revolves around putting multiple plaintext data into RLWE ciphertexts in carefully-chosen ways that mesh well with the operations you'd like to perform. By "mesh well" I mean it reduces the number of extra multiplications and rotations required merely to align data elements properly, rather than doing the actual computation you care about.

## Improving Hugging Face Training Efficiency Through Packing with Flash Attention 2

DevFeed: [Improving Hugging Face Training Efficiency Through Packing with Flash Attention 2](<https://devfeed.tech/articles/improving-hugging-face-training-efficiency-through-packing-with-flash-attention-2-7433.md>)

Original publisher: [Read original article](<https://huggingface.co/blog/packing-with-FA2>)

Author: Rhui Dih Lee; Arthur Zucker; Achintya Kundu; Laura Wynter; Raghu Ganti; Mayank Mishra

Published: 2024-08-21T00:00:00Z

Content type: article

Language: en

Sources: [Hugging Face](<https://devfeed.tech/sources/hugging-face-blog.md>)

Topics: [hugging face](<https://devfeed.tech/topics/hugging-face.md>), [Transformers](<https://devfeed.tech/topics/transformers.md>), [Machine learning](<https://devfeed.tech/topics/machine-learning.md>), [GPU](<https://devfeed.tech/topics/gpu.md>), [dataset](<https://devfeed.tech/topics/dataset.md>)

Tags: [datasets](<https://devfeed.tech/tags/datasets.md>), [efficiency](<https://devfeed.tech/tags/efficiency.md>), [flash-attention-2](<https://devfeed.tech/tags/flash-attention-2.md>), [gpu](<https://devfeed.tech/tags/gpu.md>), [hugging-face](<https://devfeed.tech/tags/hugging-face.md>), [models](<https://devfeed.tech/tags/models.md>), [packing](<https://devfeed.tech/tags/packing.md>), [padding](<https://devfeed.tech/tags/padding.md>), [training](<https://devfeed.tech/tags/training.md>), [transformers](<https://devfeed.tech/tags/transformers.md>)

### AI overview

The article explains how Hugging Face improves training throughput by packing instruction-tuning sequences without padding while using Flash Attention 2. A new data collator preserves sequence boundaries and can provide up to 2x higher throughput while maintaining convergence quality, with measured gains varying by dataset and sequence-length variance.

### Source excerpt

We're on a journey to advance and democratize artificial intelligence through open source and open science.