MLOps / Systems

Veja os papers deste label, com traduções para PT-BR.

Ver todos os labels

Artigos

🛠️MLOps / Systems • 60 artigos encontrados

Limpar filtro

NLP/LLMs • Score 85

DelegateEval: Evaluating Platforms for Personal External Service Agents

arXiv:2611.07508v1 Announce Type: new Abstract: AI agents that work for an individual over time may also help that individual serve external clients. We define a personal external service agent as an agent authorized and continuously managed by a specific individual to accept client requests, deliver work, and provide ongoing support using that person's knowledge, experience, professional methods, and service boundaries. Such agents could allow individuals to serve more clients with less direct involvement and generate economic value at scale. We propose DelegateEval, a platform-level evaluation framework for agent systems that may combine large language models, reusable skills, protocols such as MCP, and executable tool environments. It covers eight dimensions: ease of creation, service discovery and matching, delivery quality, expertise inheritance, service continuity, learning and improvement, reliability and governance, and efficiency and value. These dimensions assess whether a platform enables people with different backgrounds to turn their expertise into ongoing services that deliver verifiable work, develop capabilities through experience, and remain economically feasible. We present the framework and an experimental design. Complete datasets, executable environments, a unified scoring protocol, and cross-model results remain to be developed and empirically validated.

Fonte: arXiv cs.AI

NLP/LLMs • Score 85

DictLLM: Post-training Compression of Large Language Model with Dictionary Kernels

arXiv:2611.08082v1 Announce Type: new Abstract: Efficient deployment of large language models (LLMs) on mobile devices requires balancing performance with compute and memory constraints. Post-training compression is an effective approach to managing this trade-off, while requiring much less computation than in-training counterparts. We propose DictLLM, a dictionary-based compression scheme in which weight matrices within grouped layers are decomposed into a shared dictionary and layer-specific sparse coefficients, directly obtained from a pre-trained model without further training. Specifically, we cast DictLLM as a four-step post-training optimization. First, we obtain shared dictionaries via singular value decomposition (SVD) of input-scaled weight matrices, amortizing rank across grouped layers; second, we sparsify the layer-specific coefficients to the target compression ratio using a dictionary-aware Hessian; and third, we refine the dictionaries given the resulting sparse coefficients. Finally, we aggressively quantize both components using factor-specific Hessian information: input covariance for the dictionaries and the dictionary-aware Hessian for the coefficients. The resulting representation reduces memory footprint by storing a single shared dictionary per layer group, along with sparse layer-specific coefficients. On language modeling and zero-shot reasoning benchmarks, DictLLM consistently outperforms existing post-training low-rank compression methods and pushes the Pareto frontier between model size and performance. DictLLM compresses the original LLaMA-7B model by 5.1x with only a 2.9-point increase in WikiText-2 perplexity, achieving a 3.6x higher compression ratio than the state-of-the-art low-rank method.

Fonte: arXiv cs.AI

RL • Score 85

MAC-v2: Long-horizon model-based value expansion from pixels

arXiv:2611.07245v1 Announce Type: new Abstract: We study whether \emph{long-horizon model-based value expansion} can serve as an effective framework in pixel-based offline reinforcement learning (RL). Long-horizon model-based value expansion from pixels is substantially harder than state-based environments, as the agent must learn both the representation and its dynamics, and errors from either can accumulate recursively over imagined trajectories. We introduce MAC-v2, which enables long-horizon model-based expansion from pixels by training a reconstruction-free latent world model with action chunking. By learning representations from prediction and control objectives, MAC-v2 does not need to encode control-irrelevant visual details that can hinder the dynamics prediction. Action chunking further mitigates the prediction error by reducing the number of recursive model predictions required for value expansion. Across diverse visual benchmarks including visual OGBench, MMBench2, Robomimic, and LIBERO-10, MAC-v2 achieves strong performance against model-free and model-based offline RL baselines. Our analyses show that long-horizon model-based value expansion is critical for the performance of MAC-v2, and that both action chunking and latent learning objectives are essential for reliable long-horizon latent imagination.

Fonte: arXiv cs.LG

NLP/LLMs • Score 85

Server-Side Sparse Assignment for Federated Mixture-of-Experts

arXiv:2611.07093v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly recognized for their remarkable capabilities; federated learning (FL) offers a promising framework for collaboratively fine-tuning LLMs on edge devices while achieving data privacy. However, edge devices are inherently heterogeneous: clients may exhibit diverse data distributions and resource constraints, including limited memory and computation budgets. Recent federated Mixture-of-Experts (MoE) approaches address these challenges by using sparse expert activation or partial expert loading. However, existing partial-expert methods rely on heuristic full-pool observations or prioritize system-wide load balance, without guarantees for recovering client-specific relevant experts from partial observations. We propose $S^3$-MoE, a server-side sparse expert assignment framework that estimates client-specific expert relevance from partial observations and refines assignments through an exploit--explore mechanism. We also establish relevance-estimation error and exact Top-$K$ recovery guarantees. Experiments using a CNN-based MoE on CIFAR-100 show that $S^3$-MoE recovers task-dependent client--expert preferences and reduces conflicts caused by correlated expert selection across clients. Compared with Full-20 FedAvg, which gives each client access to all 20 experts for local Top-5 routing, $S^3$-MoE achieves 1.31 percentage points higher mean final accuracy while loading only $5$ experts per client.

Fonte: arXiv cs.LG

RL • Score 85

Octopus: One Brain, Many Branches for Latent-KV Multi-Agent Reinforcement Learning

arXiv:2611.06956v1 Announce Type: new Abstract: Multi-agent LLM systems explore in parallel by spawning sub-agents with isolated context windows, whereas single-agent systems append every search and verification step into one growing trajectory. However, existing multi-agent frameworks duplicate prefix KV caches across branches, exchange findings by re-prefilling text summaries, and flatten or drop cross-branch dependencies during reinforcement learning (RL). We introduce Octopus, a single-model multi-agent RL framework in which all threads share one backbone and coordinate over a GPU-resident directed acyclic graph (DAG) of KV states through five primitives (fork, read_thread, wait_threads, rewind, commit). Octopus combines three co-designed mechanisms: (1) Zero-Copy Radix Forking, which spawns branches as $\mathcal{O}(1)$ cursor views over shared physical KV pages; (2) Attention-Only KV Mounting, which mounts a sibling thread's terminal KV blocks directly into full-attention layers while bypassing linear-recurrent states (Gated DeltaNet / Mamba-2), cutting off-task hijacking by $21\times$ ($57.9\% \to 2.7\%$) while retaining $92.2\%$ of full-text mutual information at zero extra KV allocation; and (3) a Packed DAG-GRPO Trainer, which packs all episode branches into one block-sparse attention pass and assigns counterfactual branch credit via leave-one-branch-out marginal rewards. Trained end-to-end with GRPO on Qwen3.5-4B over a sparse BM25 corpus, Octopus reaches $49.6\%$ accuracy on BrowseComp-Plus ($+17.5$ points over single-agent GRPO at 143K context and $+12.5$ points over text-summary multi-agent RL) and $51.2\%$ on OpenResearcher, outperforming dense-retrieval agents up to $250\times$ larger while increasing training throughput $3.3\times$ and cutting peak KV memory $2.8\times$.

Fonte: arXiv cs.LG

Theory/Optimization • Score 85

Loopback Transformer: Towards a Unified View of Residual Looped Transformers

arXiv:2611.07226v1 Announce Type: new Abstract: Looped Transformers reduce parameter count by repeatedly applying shared blocks, but parameter sharing alone does not determine how the results of repeated computation are reused. We take the depth-memory view of residual connections and analyze which representation is stored after each block application and how later reads weight it. A full block output contains both its input and its new update, so weighting that output also reweights the history already carried by the input. Reading individual updates instead, with the shared stack's input kept on a separate path, gives separate controls over input strength, total update weight, and allocation across updates. The Loopback Transformer implements these controls: geometric decay sets the total update weight and supplies a recency prior, and attention learns how to allocate that weight. On ClimbMix, Loopback reaches lower validation loss than standard Transformer baselines with 40-75% of their Transformer parameters (excluding token embeddings and the output head) at reference scales from 191M to 3B, with training tokens and executed depth matched. It also has higher mean accuracy over thirteen downstream tasks at 1.4B and 3B. At 191M, 374M, and 778M, it has the lowest validation loss and the highest thirteen-task mean among the standard Transformer and seven looped residual designs.

Fonte: arXiv cs.LG

Theory/Optimization • Score 85

Bayesian Tensor Train Kernel Machines with Automatic Model Complexity Inference

arXiv:2611.07129v1 Announce Type: new Abstract: Tensor Network (TN) kernel machines accelerate supervised learning by using low-rank TNs to reduce computation and memory. However, most TN-based methods are deterministic with manually tuned ranks and feature dimensions, while the existing Bayesian TN kernel machine is restricted to the CANDECOMP/PARAFAC decomposition (CPD), whose single rank is shared by all modes. We extend this framework to the tensor train (TT) and propose Bayesian Tensor Train Kernel Machines (BTT-KM), whose hierarchical ARD-type priors on the TT cores infer the TT ranks and feature dimensions from data. Since each TT core is coupled to the others only through its neighbors, its closed-form posterior update requires propagating the contributions of all other cores along the chain. Mean-field variational inference yields a Bayesian alternating least squares algorithm with the same computational complexity as its deterministic counterpart and a closed-form predictive distribution. A synthetic experiment illustrates that BTT-KM recovers the true TT ranks and feature dimensions from an over-parameterized initialization.

Fonte: arXiv stat.ML

MLOps/Systems • Score 85

Energy Measurement and Optimization of TinyML in Resource-Constrained Embedded Systems: A Systematic Mapping Study

arXiv:2611.07264v1 Announce Type: new Abstract: The rapid expansion of the Internet of Things (IoT) has increased the volume of data generated at the network edge, where many devices operate under severe memory, processing, and energy constraints. In this context, Tiny Machine Learning (TinyML) enables machine learning (ML) tasks to be executed on microcontrollers, smart sensors, and other resource-constrained platforms. However, the literature remains heterogeneous regarding the algorithms evaluated, energy-measurement methods, reported metrics, influencing factors, and optimization strategies. This study presents a systematic mapping of energy-aware TinyML research. It identifies the main models and learning approaches, classifies energy-evaluation methodologies and metrics, examines the technical factors affecting energy consumption, and organizes the reported optimization techniques into a structured taxonomy. The findings highlight the frequent use of convolutional and lightweight architectures, direct measurements and analytical estimations, and factors related to model architecture, hardware, communication, numerical precision, and input characteristics. The optimization strategies were grouped into model compression, hardware-aware design, system and communication optimization, and data and processing optimization. The results also reveal limited standardization in experimental platforms, measurement procedures, and reporting practices, limiting reproducibility and direct comparison across studies. This mapping provides an updated overview of the field and guidance for developing and evaluating more energy-efficient TinyML solutions.

Fonte: arXiv cs.LG

Evaluation/Benchmarks • Score 85

Closed-Loop Agent Evaluation: Layered Measurement and Self-Calibration

arXiv:2611.07051v1 Announce Type: new Abstract: Once an agent is live, teams want to know whether it is doing its job, not just whether it throws errors. Yet most evaluation is one-off: a benchmark run, a look at traces, a single score. Its conclusions neither check themselves nor reach the agent. We propose closed-loop evaluation, an evaluation method for deployed agents. It models evaluation as performance estimation under biased raters and censored outcomes, and it requires every level's conclusion to trace back to atomic data and to be computed by versioned rules. The method covers four levels of analysis (span, trace, conversation and engagement), measures along a spine of Execution, Substance, Trajectory and Results, treats Conduct as a constraint throughout, and uses Provenance and Adjudication components to certify the evidence and the evaluators. Evaluator agents generated by an Agent Factory carry these components, each with a domain knowledge pack and an adapter harness. Two loops close the evaluation: an inner loop audits the evidence, rules and metrics themselves, and an outer loop uses per-component regression checks to drive agent improvement. We also give a pipeline that runs this process automatically. Using the business-analysis agent SharpInsight as a case and porting the method to public conversation data, we show how problems differ across levels and how evidence-and-rule judgment is more reliable than holistic scoring (ICC 0.86 vs. 0.81; kappa 0.81 vs. 0.63).

Fonte: arXiv cs.AI

NLP/LLMs • Score 85

Stochastic Dynamics of Generative AI: From Autoregressive Models to Interacting LLM Networks and Neural SDEs

arXiv:2611.07065v1 Announce Type: new Abstract: This paper develops a stochastic dynamical-systems perspective on three classes of generative AI models. First, we formulate large language models (LLMs) as high-dimensional nonlinear autoregressive models with attention-based dependencies. This formulation connects pretraining by conditional maximum likelihood to sequential token generation and provides a probabilistic interpretation of LLM operation. Second, we consider networks of interacting LLMs and characterize their collective behavior through mean-field ordinary differential equations. These dynamics provide a framework for understanding emergent preferential attachment and persistent structural biases, including type-dependent glass-ceiling effects. Third, we consider neural stochastic differential equations (SDEs) as continuous-time generative models for stochastic trajectories, including those observed at irregular time intervals. We discuss their training with particular emphasis on Wiener-space cubature methods based on rough-path analysis as an efficient alternative to Monte Carlo estimation. Together, the three settings address probabilistic prediction, information exchange and emergent behavior in interacting networks, and path-space generative modeling.

Fonte: arXiv cs.LG

MLOps/Systems • Score 85

Neural Compression over Equivalence Classes: Exact Gauges and Functional Rate-Distortion

arXiv:2611.07233v1 Announce Type: new Abstract: Neural networks admit many parameterizations of the same or nearly the same function, but deployment typically stores whichever parameter vector training happens to produce. We show that this representational freedom can itself be used for compression. Rather than redesigning the codec, we search over exact function-preserving transformations and controlled functional perturbations, using the unchanged physical compressor itself to select cheaper representations. Exact gauge families provide a zero-distortion reference under the strict audit; allowing small functional changes exposes much larger rate-distortion gains. Across SmolLM2 and Qwen3, our method reduces Zstandard-compressed checkpoint size by up to 23.4\%, with similar relative reductions under gzip, Brotli, and ZipNN, establishing trained-network representational freedom as a practical storage resource.

Fonte: arXiv cs.LG

NLP/LLMs • Score 85

PromptOps: Iterative Prompt Optimization for LLM-as-a-Judge Systems

arXiv:2611.07123v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly used as automated judges for evaluating the behavior of chatbots and other AI systems. In deployment, however, their agreement with the intended human evaluation policy may degrade as user interactions and score distributions change. Existing automatic prompt optimization methods are primarily studied as one-shot procedures on static development sets and therefore provide limited guidance for maintaining LLM judges over time. We frame this problem as judge drift and introduce PromptOps, a closedloop framework for monitoring and iteratively recalibrating LLM-as-a-Judge (LLMaaJ) systems. PromptOps separates four operational decisions: when to optimize, which optimization procedure to apply, which feedback records to use, and how to update the judge prompt. We instantiate the framework with five prompt optimization methods: GEPA, SIMBA, Rubric Synthesis, and exemplar selection using either simulated annealing or a Tree-structured Parzen Estimator. Our experiments demonstrate that iterative prompt optimization can recalibrate LLM judges and recover performance after controlled shifts in the referencescore distribution. However, effective maintenance is inherently contextdependent: neither optimization algorithm choice nor update timing generalizes universally across settings. These findings establish prompt optimization for deployed LLM judges as an inherently iterative maintenance problem and demonstrate that PromptOps provides a practical abstraction for studying and implementing this process.

Fonte: arXiv cs.AI

MLOps/Systems • Score 85

WeightLoop: Weight-Space Programming with Experimental Feedback

arXiv:2611.07319v1 Announce Type: new Abstract: Weight-space programming (WSP) shows that language models can reason about and rewrite neural network weights; recursive self-improvement (RSI) motivates feedback-driven improvement. Models guiding experiments receive concrete feedback: completed assays reveal prediction errors and inform what to test next. We introduce WeightLoop, a framework that turns experimental feedback into persistent updates to deployed predictors. GPT-6 Astra inspects a task specification and the model implementation, then writes a bounded, reusable parameter-editing program. As new measurements arrive, the program identifies prediction discrepancies, validates candidate edits using observed outcomes, and writes the selected changes back to the original model. The editor remains fixed while the predictor evolves, enabling repeated updates without further LLM calls or gradient retraining while preserving the original inference interface. In offline replay, targeted edits improve mean ranking utility in seven of eight endpoint-budget settings (all eight versus random edits) and mean final discovery across three sequential tasks. WeightLoop connects WSP with an RSI-inspired improvement loop, making experimental evidence a direct driver of model change and subsequent decisions.

Fonte: arXiv cs.LG

RL • Score 85

GuidedWAM: Training-Free Energy Guidance for World Action Models in Autonomous Driving

arXiv:2611.07912v1 Announce Type: new Abstract: World action models (WAMs) for autonomous driving couple the driving policy and the visual world model in a single generative network whose objectives are fixed post-training. Conditioning on a new driving objective requires retraining. We study energy guidance as a general mechanism and present GuidedWAM, a model-agnostic framework for injecting driving objectives into frozen WAMs at sampling time. Our framework constructs each objective as differentiable energies on the sampler's intermediate clean-action estimate, which needs only forward access to the frozen model, and is stabilized by sampling-time guidance scheduling, per-energy normalization, and velocity-bounded correction. On the NAVSIM v2 hard-split, guidance improves all three frozen, publicly released WAMs by 1.8 to 3.6 EPDMS, averaged over three sampler seeds, and adds less than 10% to the sampling time per step. These results position energy guidance as a practical post-training control channel for WAMs.

Fonte: arXiv cs.AI

RL • Score 85

Resource-Efficient Reinforcement Learning Post-Training through Elastic Expert Parallelism

arXiv:2611.06984v1 Announce Type: new Abstract: Reinforcement learning post-training improves the capabilities of large language models, but generating training responses is a major computational bottleneck. In mixture-of-experts models, workers that finish their assigned responses may remain occupied because their experts are still needed by workers processing longer responses. Consequently, approximately one-third of occupied worker time in the baseline occurs after local response generation finishes. Here we present FlexFloor, a system that enables earlier resource release during synchronous response generation while preserving the expert computation required by ongoing responses. FlexFloor jointly adapts response scheduling and the active group of workers. It uses historical response lengths and memory constraints to guide batching and prompt placement, then progressively releases workers while maintaining expert availability and correct communication. Compared with the verl baseline, FlexFloor generates approximately 65.40% more response tokens per unit of occupied computing time and increases response generation throughput by 52.74%, with little observed variation in rollout reward. By making reinforcement learning post-training more efficient, FlexFloor supports faster development and iterative improvement of large language models and AI agents.

Fonte: arXiv cs.AI

NLP/LLMs • Score 85

FlashOffloading: Low-Latency KV Cache Offloading for Long-Context LLM Serving

arXiv:2611.07542v1 Announce Type: new Abstract: In large language model (LLM) serving systems, the key-value (KV) cache grows linearly with sequence length and concurrency, so long-context inference easily hits the high-bandwidth memory (HBM) capacity wall and results in low concurrency. Offloading the KV cache to host dynamic random-access memory (DRAM) frees HBM space and raises concurrency, but the additional cache management and host-to-device (H2D) load overhead inflates inference latency, higher concurrency fails to translate into effective throughput. We present FlashOffloading, a host-DRAM-centric inference architecture that fully offloads the KV cache in both the prefill and decode phases. FlashOffloading builds rolling HBM cache for prefill, a hot cache for decode, a reverse sparse index with zero extra miss-lookup overhead, and a compute--load overlapped kernel, optimizing KV offloading latency across the entire pipeline and delivering throughput gains under a time-per-output-token (TPOT) service-level objective (SLO). In experiments deploying GLM-5.2 with multi-token prediction (MTP3) speculative decoding on an Ascend910C prefill--decode (PD)-disaggregated platform, FlashOffloading improves prefill throughput by up to 1.38\(\times\) and decode throughput by up to 1.91\(\times\) under a 30 ms TPOT constraint. FlashOffloading is integrated into an open-source serving framework: https://github.com/vllm-project/vllm-ascend/tree/releases/v0.30.0.

Fonte: arXiv cs.AI

MLOps/Systems • Score 85

Machine Learning for Live-Birth Prediction After IVF/ICSI From Structured Clinical Data: A Systematic Review, Meta-Analysis, and Critical Appraisal

arXiv:2611.07124v1 Announce Type: new Abstract: Machine-learning models predicting live birth after in vitro fertilisation (IVF) or intracytoplasmic sperm injection (ICSI) from routine clinical data report performance from near chance to near perfect. We systematically reviewed these models under PRISMA 2020, searching OpenAlex, Scopus and Web of Science to 22 October 2026, appraised them with PROBAST+AI and TRIPOD+AI, and pooled their discrimination with random-effects meta-analysis on the logit scale. We define what fixes the meaning of a model's output (decision point, endpoint, unit of analysis, data source) and map the 48 included studies by model family, decision point and validation design. Models evaluated only internally pooled to an area under the receiver operating characteristic curve (AUC) of 0.814, against 0.756 (95% prediction interval 0.546-0.889) for models evaluated on a later period or another centre, and 0.764 for studies not at high risk of bias; neither difference was significant, and heterogeneity was extreme. Models whose decision point was not stated pooled to 0.910, the only significant moderator in meta-regression. Against logistic regression fitted to the same data, the best machine-learning model gained a pooled 0.056 AUC (95% CI 0.017-0.090). Of 48 studies, 47 were at high or unclear risk of bias, only four assessed the calibration slope, and no model has evidence of clinical impact. The certainty of the pooled evidence is low to very low; the gains that survive validation came from staging predictions to decision points and local calibration, not deeper architectures. We provide a TRIPOD+AI-mapped checklist, a benchmark specification and a research agenda.

Fonte: arXiv cs.LG

NLP/LLMs • Score 85

Test-Time Self-Play: Scaling without a Verifier

arXiv:2611.07367v1 Announce Type: new Abstract: Test-time scaling methods spend additional compute at deployment to improve a large language model's performance on a hard target problem. Most methods rely on a verifier that scores candidate solutions for correctness, using it to select among or refine the model's outputs. In many deployment regimes, however, a strong verifier on the target problem is unavailable. We introduce Test-Time Self-Play (TTSP), an approach to test-time scaling using asymmetric self-play without reliance on a verifier. On a difficulty-calibrated subset of LiveCodeBench v6, TTSP is the best-performing method at every verifier strength we evaluate, and the gap to the strongest baseline widens as the verifier weakens; in the no-verifier limit, TTSP yields a 35\% absolute improvement in mean@8. The gains hold across a range of verifier strengths, except when the base policy has near-zero coverage of the correct answer.

Fonte: arXiv cs.LG

MLOps/Systems • Score 85

EnvOpt: Automated Optimization for Agent Environments

arXiv:2611.07372v1 Announce Type: new Abstract: An LLM agent's cost and success depend on its model, its harness, and its environment. In addition to the widely studied directions of parameter tuning and harness optimization, we propose the third direction: optimization over stationary environments. Agents are often deployed in rich environments shared by many tasks, such as a company's Slack workspace or a complex collection of data sources, documents or tickets. We call these environments stationary environments. We introduce EnvOpt, a method for automated end-to-end optimization of stationary environments for any given model and harness, using real or synthetic tasks without needing any graders or reference answers. On AutomationBench, EnvOpt reduces the cost of Claude Haiku 4.5 by 72.7% and raises its pass rate by 14.34%. The same environment transfers to Opus 5, cutting its cost by 68.1% with a 1.34% gain. Added on top of Meta-Harness's optimized harness for Haiku 4.5, it lowers cost by a further 83.4% and adds 11.17% accuracy over Meta-Harness alone. Across 5 benchmarks, EnvOpt significantly reduces the cost of agents while improving or maintaining pass rate, and is transferable to different models and different harnesses. We also discuss common successful and rejected environment optimization strategies during automated optimization. More broadly, our results show that how information is represented to an agent shapes both its cost and its success, making the environment worth optimizing in addition to the parameters and the harness.

Fonte: arXiv cs.AI

MLOps/Systems • Score 85

Acoustic Anomaly Detection for Early Identification of Drill Jamming

arXiv:2611.07356v1 Announce Type: new Abstract: In the context of Industry 4.0, predictive maintenance is essential to minimize unplanned downtime and ensure operational safety. In machining processes, drill jamming is a critical failure that compromises both workpiece quality and tool integrity. This work proposes an acoustic anomaly detection system using ultrasonic sensors to identify pre-failure patterns in Computer Numerical Control (CNC) machining. We release a public dataset of time- and frequency-domain acoustic features from 17 drills, and benchmark semi-supervised autoencoder-based normality modelling against supervised XGBoost classification using Leave-One-Drill-Out validation. The results demonstrate that the Autoencoder is highly effective for industrial safety, achieving a Recall of 1.00 and a consolidated F1-score of 0.81. Furthermore, the threshold-independent analysis shows that the Autoencoder better separates normal and pre-failure acoustic patterns, achieving a consolidated AUC of 0.8294, compared with 0.7559 for the supervised model under inter-tool covariate shift. In contrast, the supervised XGBoost achieved an average F1-score of 0.55 and failed to detect failures in some of the evaluated drills. The findings highlight that the ultrasonic spectrum provides a reliable window for identifying incipient mechanical instabilities, and the AE model offers a robust safety envelope by anticipating all jamming events across the tested drill bits.

Fonte: arXiv cs.LG

NLP/LLMs • Score 85

SD-LoRA: Training-Time Structural Distillation into LoRA for Few-Shot Vision-Language Adaptation

arXiv:2611.07063v1 Announce Type: new Abstract: Few-shot adaptation of vision-language models must balance two competing goals: learning fine-grained visual distinctions from very limited supervision and preserving the efficiency that makes frozen CLIP-like models attractive for deployment. Existing lightweight adaptation methods, including prompt tuning, LoRA, and cache-based adapters, typically supervise global image embeddings and therefore underuse token-level structure. Conversely, structure-aware methods exploit local evidence but often introduce additional inference-time computation. We propose SD-LoRA, a structural distillation framework that uses token-level graph reasoning only during training and transfers its effects into a lightweight LoRA-adapted CLIP model. During training, a heterogeneous graph teacher models patch-patch, global-local, and visual-text relations over high-resolution ViT tokens. Rather than retaining this graph at test time, SD-LoRA transfers its structural bias into the shared CLIP-LoRA backbone through implicit distillation and Prototype Predictive Alignment, which aligns graph-conditioned representations with support-derived class prototypes. At inference, the graph teacher is discarded; prediction requires only a single CLIP-LoRA forward pass and a cache-based classifier. Across 11 few-shot benchmarks, SD-LoRA improves over a strong CLIP-LoRA baseline by 2.4/1.8/3.0/2.4/2.0 percentage points in the 1/2/4/8/16-shot settings, respectively, without additional graph computation at test time. Results on base-to-novel generalization, cross-dataset transfer, out-of-distribution robustness, and calibration further show that training-time structural supervision improves both accuracy and transfer reliability. Code is available at https://github.com/MR-Sherif/SD-LoRA.git.

Fonte: arXiv cs.CV

NLP/LLMs • Score 85

Householder Motion Embedding: Noise Modulation for Motion-Controllable Video Generation

arXiv:2611.07294v1 Announce Type: new Abstract: Diffusion models for video generation have achieved remarkable success, yet precise motion control under explicit user guidance remains challenging. Recent noise-warping methods introduce temporal correlations by warping the noise according to the user-provided motion signal. However, standard spatial warping generally disrupts the noise's i.i.d. Gaussian structure, requiring distribution correction, such as Jacobian-based correction, but at a substantial computational overhead. We propose Householder Motion Embedding (HME), which encodes the motion signal into the noise through a sequence of Householder reflections that together form an orthogonal transformation applied to each location's channel vector. HME exactly preserves the spatial Gaussian structure within each frame while introducing cross-frame correlations that promote temporally consistent video. HME is agnostic to the network architecture and integrates directly into standard video diffusion fine-tuning pipelines. Experiments on different scenarios show that HME matches or outperforms the state of the art, while achieving a 2.7x speedup over noise warping. These results demonstrate that channel-space Householder transformations offer a principled and efficient alternative to noise warping for motion-controllable video generation.

Fonte: arXiv cs.CV

Theory/Optimization • Score 85

Greedy Alignment Principle for Optimizer Hyperparameter Selection

arXiv:2512.06370v4 Announce Type: replace-cross Abstract: Recent works have shown that gradient-update alignment is a powerful signal for modulating optimizer updates, often leading to faster training. We promote this update-wise heuristic as a mathematically grounded principle for selecting and tuning optimizer hyperparameters. By treating gradients and updates as signals and an optimizer as a causal filter that maps between them, we formulate optimizer selection as maximizing the expected drop rate in loss over a prescribed family of optimizers. We show that this objective is exactly the inner product between the optimizer filter and the gradient autocorrelation, and prove that a greedy optimum exists and has a stability bound under perturbations of the estimated gradient statistics. Specializing in momentum-based optimizers, the theory yields simple dynamic momentum selection rules for both SGD+Momentum and Adam/AdamW. Experiments across image classification, language model fine-tuning, and vision transformer fine-tuning show that the resulting dynamic momentum rules match or improve upon the best fixed hyperparameters found via manual sweeps, reducing the need for exhaustive momentum sweeps. Code is available at https://github.com/ironjr/gap

Fonte: arXiv stat.ML

NLP/LLMs • Score 85

Perceptual Distortion for Overfitted Video Coding: Sliced Matching in Space, Error Alignment in Time

arXiv:2611.07148v1 Announce Type: new Abstract: Recent works in neural video coding have increasingly been trading off perceptual quality against reconstruction fidelity at a fixed bitrate. Wasserstein Distortion (WD) offers a principled approach to balancing these objectives by relaxing spatial correspondences in feature matching, enabling lightweight overfitted image codecs to compete with computationally intensive generative decoders. Extending vanilla WD to video, however, presents two challenges: its independent Gaussian approximation limits the expressiveness of feature distribution matching, while unconstrained texture resampling across frames introduces temporal inconsistencies. We address these challenges by distinguishing between the correspondences that can be relaxed and those that must be preserved. Spatially, we introduce a local sliced Wasserstein distortion (SWD) that matches empirical feature distributions through one-dimensional optimal transport, capturing cross-channel dependencies without enforcing exact spatial alignment. Temporally, we employ a novel flow residual that constrains coding errors along motion trajectories, preserving temporal consistency without imposing additional reconstruction fidelity constraints. Evaluated on UVG and HEVC Class B with the overfitted codec COOL-CHIC, our framework achieves BD-rate reductions of 9.2% in DISTS and 3.0% in LPIPS relative to frame-wise WD, averaged over test sequences, while lowering tOF by 10.2% at matched rate.

Fonte: arXiv cs.CV

RL • Score 85

Regulating Autonomous Mobility-on-Demand Operators with Fixed and Adaptive Penalties: A Multi-Agent Reinforcement Learning Approach

arXiv:2611.07122v1 Announce Type: new Abstract: Profit-maximizing Autonomous Mobility-on-Demand (AMoD) operators may exploit consumer surplus through excessive prices and underserve less profitable regions by relocating vehicles away from them. We study how penalty-based regulation of an AMoD operator in a multimodal mobility system can limit surge pricing and improve regional service availability. We model regulation as an outer control loop that acts on a learning-based operator. Static regulation with fixed penalties is formulated as a Markov Decision Process for an operator that jointly controls pricing and rebalancing, and dynamic regulation as a Partially Observable Stochastic Game in which a regulator periodically adjusts region-specific penalties in closed loop. Operator and regulator policies are learned with graph-based multi-agent reinforcement learning. In experiments with real-world Manhattan data, fixed surge pricing penalties reduce the share of travelers facing excessive prices from 22% to 1%. For regional service availability, dynamic regulation improves the regulator's reward by up to 15.6% over fixed penalties of equal mean intensity. The gains in service availability are strongest when spare fleet capacity exists, whereas service deficits persist in peak periods with highly utilized fleets. The learned regulator policy responds by reducing penalties where limited fleet capacity restricts compliance. The results identify trade-offs between regulatory targets, operator profitability, and traveler utility.

Fonte: arXiv cs.LG

NLP/LLMs • Score 85

TriAttention: Unified Sparse Mixed-Precision Attention for Long-Context LLM Inference

arXiv:2611.07256v1 Announce Type: new Abstract: Sparsity and quantization reduce the cost of attention by skipping interactions and lowering the precision of those retained. However, useful interactions differ in their precision requirements: some tolerate low-precision scores while remaining sensitive to removal. We therefore introduce TriAttention, which assigns query-key (QK) tiles to skipping, 4-bit, or 8-bit computation, using 4-bit to retain quantization-tolerant interactions at lower cost. To represent compactly, we encode each head's computation regions with two diagonal-distance boundaries, keeping policy storage independent of context length. Offline calibration then evaluates candidate boundary pairs using normalized projected-output errors and complete-tile execution costs. Using these measurements, a model-wide allocator selects each head's boundaries to minimize calibration error under a shared computation budget. During prefill, the selected boundaries generate precision-grouped traversals for a fused kernel that skips omitted tiles and combines contributions from both QK precisions into one normalized attention output. The quantized keys and values prepared during prefill are then reused in a compressed key-value (KV) cache for decoding, reducing repeated reads of prompt data. Experiments show that TriAttention achieves end-to-end generation speedups of up to 7.88x over the Torch attention and 2.65x over the BF16 baseline. Meanwhile, TriAttention maintains quality comparable to the original models on general-purpose and long-context tasks. These results show how a compact retention-and-precision policy translates local sensitivities into attention computation and repeated KV-cache access. The implementation is https://github.com/HanzhiZhang-Ulrica/TriAttention.

Fonte: arXiv cs.LG

NLP/LLMs • Score 85

Align Before Aggregation: Basis-Consistent Federated LoRA under Heterogeneous Ranks

arXiv:2611.07598v1 Announce Type: new Abstract: Federated fine-tuning of large language models (LLMs) under heterogeneous client capacities is increasingly important, and low-rank adaptation (LoRA) with heterogeneous ranks makes it practical. Existing server-side aggregators compose local matrix updates $\Delta W_i$ from client-specific LoRA factors $(A_i,B_i)$ before aggregating them. Although these matrix updates lie in the same weight space $\mathbb{R}^{m\times n}$ and can be averaged directly, the LoRA factors that generate them may expressed client-specific local basis. In the typical compose--aggregate--factorize--dispatch loop, this basis inconsistency can make the rank-$r_i$ reference prefix dispatched to lower-rank clients unstable after server-side refactorization. In this work, we propose \textbf{FedAbA}, a basis-consistent federated LoRA aggregation framework for heterogeneous ranks. FedAbA first extracts each client's local basis, aligns it to the corresponding prefix of the global reference basis, and constructs a basis-consistent reconstruction of each local matrix update before aggregation. It then aggregates these reconstructions with weights that combine client data size and consistency score, followed by exact coefficient-space SVD refactorization and gauge fixing before rank-prefix dispatch. Our theoretical analysis clarifies the role of basis consistency and supports the proposed consistency-aware aggregation mechanism. Extensive experiments show that FedAbA outperforms representative federated LoRA baselines, with the largest gains under the strongest rank heterogeneity.

Fonte: arXiv cs.AI

NLP/LLMs • Score 85

LowRAM: Low-Rank Keys and Amortised Values for Composable KV Cache Compression at Test-Time

arXiv:2611.07025v1 Announce Type: new Abstract: Autoregressive decoding relies on the KV cache, whose memory footprint grows linearly with context length and caps batch size and throughput. Existing post-training methods exploit redundancy across tokens, layers, and feature dimensions, or lower numerical precision. We identify a complementary axis: within each attention head, values are largely predictable from their corresponding keys. We exploit this with value cache amortisation, which fits a lightweight per-head predictor at test time and stores residuals only for the tokens it reconstructs worst, giving $4\times$ near-lossless value compression. The predictor can be initialised by meta-learning for substantially reduced prefill overhead, or fitted from scratch in settings that allow a one-off cost, such as multi-query inference. We pair value amortisation with cross-layer low-rank key compression to produce LowRAM, an asymmetric compression pipeline that stays close to lossless across three model families. By reconstructing only a selected subset of KV pairs at each decoding step, LowRAM can keep decoding overhead minimal and achieve up to $5\times$ higher peak decode throughput at $4\times$ compression. LowRAM further composes with quantisation and eviction, reaching $26.9\times$ compression at $94.5\%$ of baseline performance on long contexts. These results show that treating keys and values as structurally distinct enables efficient, aggressive, and composable KV cache compression.

Fonte: arXiv cs.LG

MLOps/Systems • Score 85

Don't Prompt Your Agents. Equip Them: Skillware as a Modular Standard for Sovereign, Auditable, and Deterministic Capability Execution for Logical Systems

arXiv:2611.07652v1 Announce Type: new Abstract: Current autonomous logical systems rely on a fragile assumption: that language models should dynamically write code, guess API payloads, and memorize operational constraints through prompt engineering. In practice, this burns massive context windows, inflates API bills, and produces brittle agents that fail unpredictably in production. In this work, we present Skillware, an open architectural framework and capability registry that formally separates what an agent wants to do from how that task is executed. Instead of prompting models with loose text instructions or markdown files baptized as 'skills', Skillware equips them with modular, versioned software capability bundles. Much like skill hotkeys in video games that execute deterministically under fixed world rules, each bundle combines a typed JSON input contract, deterministic Python code, task directives, bundled datasets, and automated unit tests. In-process adapters translate these capabilities into native tool declarations for any model, from frontier cloud APIs to local open-weight runtimes with zero network latency. Evaluating 23 frontier and open-weight models across 368 benchmark conditions, we show that small local models (3B to 8B parameters) equipped with Skillware consistently outperform unequipped frontier models on complex multi-step workflows. They cut token usage by up to 84.8%, slash inference costs by up to 97.3%, and eliminate parameter errors entirely. True agent reliability does not come from larger models or longer prompts, but from equipping logical systems with verified, deterministic software. Finally, we articulate the trajectory of capability bundles across an invariant continuum of information processing systems: from digital logical systems today, to IoT, drones, and embodied robotics tomorrow, establishing a foundation for auditable, real-time capability execution.

Fonte: arXiv cs.AI

NLP/LLMs • Score 85

Graph Neural Networks for Resource Allocation in Integrated Sensing and Communication

arXiv:2611.07277v1 Announce Type: new Abstract: In vehicular integrated sensing and communication (ISAC) networks, each service-provider vehicle uses a limited power budget to serve multiple sensing targets and communication receivers. Maximizing the communication sum-rate subject to sensing signal-to-interference-plus-noise ratio (SINR) constraints is nonconvex, and iterative optimization is costly to rerun for every network layout. This paper proposes a graph neural network (GNN)-based power allocation algorithm, termed GNN-ISAC. Concretely, GNN-ISAC addresses the aforementioned problem by considering two distinct edge types in the graph representation, one capturing the inter-provider interference, the other representing the intra-provider power-budget sharing. GNN-ISAC is trained without labels using a sensing-penalized sum-rate loss. We also adapt the weighted mean-square error minimization (WMMSE) algorithm to this problem by introducing a sensing-power reservation step. Simulation results show that GNN-ISAC achieves communication sum-rates close to WMMSE with low sensing-link violation rates and much less computation. It also outperforms an existing heterogeneous GNN and a channel-blind baseline, and generalizes to larger networks without retraining.

Fonte: arXiv cs.LG

NLP/LLMs • Score 85

BlockLock: Reproducible LLM Inference with Fixed Batch Geometry

arXiv:2611.06977v1 Announce Type: new Abstract: Reproducible LLM inference requires that a request produces the same output regardless of the other requests in the batch. Existing solutions pay for this in different ways: batch-invariant kernels constrain every operator, and isolating prefill with fixed-shape decode ({our CoRun-style baseline}) gives up shared prefill compute and pads decode. We present BlockLock, which keeps shared batched prefill and pins only the numerical path of the protected request: it fixes the geometry that the request sees, with {two modes: BlockLock-Rows uses fixed-width rows and retains batched decode of all requests; BlockLock-Pack uses terminal-anchored packing with an isolated dummy segment, then extracts the protected request's KV state for target-only decode. Under a stated per-deployment contract, both modes preserve the target's output bitwise when its companions change.} Across five {dense} models (Qwen3-8B/14B, Llama-3.1-8B-Instruct, and Gemma-3-4B/12B-IT) and three single-GPU deployments (NVIDIA A6000, A100, and H100), we evaluate our approach on both tinyBenchmarks and BFCL{; we also include initial tensor-parallel experiments on two NVIDIA A6000 GPUs. Both modes preserve target sequences and Top20-logit bits in the reported tests.} At equal delivered work, {BlockLock-Rows} reduces latency by 53.5--55.0\% and improves throughput by 115.2--122.2\% relative to vLLM Batch-Invariant; compared with batch-invariant kernels, it needs no custom kernel implementations and constrains only the protected requests.

Fonte: arXiv cs.AI

NLP/LLMs • Score 85

TileQuant: Tile as the Precision-Allocation Unit for LLM Mixed-Precision Post-Training Quantization

arXiv:2611.07217v1 Announce Type: new Abstract: Post-training quantization (PTQ) reduces large language model (LLM) deployment costs, but preserving quality requires accounting for uneven weight sensitivity. Uniform PTQ uses one weight bit-width, importance-aware PTQ adjusts scales or reconstructs weights within that width, and column-wise mixed precision retains selected columns at higher precision. However, sensitivity varies within columns, so whole-column retention also protects less sensitive weights, while runtime outlier detection adds decoding overhead. We therefore introduce TileQuant, a training-free framework that assigns precision to hardware-aligned 2D weight tiles. Offline calibration provides input-activation-aware scoring from forward statistics for lower calibration cost and gradient-informed scoring from loss sensitivity for stronger quality preservation. Within each projection's retention budget, the highest-scoring tiles remain at 16-bit and the rest use 8-bit. The resulting map guides reconstruction into compact weight buffers, which the kernel reuses for native integer and 16-bit computation with local accumulation to reduce memory traffic. Experiments show that TileQuant reaches 2.66x the original model's generation throughput. Across language modeling, reasoning, long-context retrieval, and question answering, it preserves quality and surpasses quantized and original models on multiple tasks. These results support selecting precision units that capture local sensitivity while enabling efficient hardware execution. The implementation is: https://github.com/HanzhiZhang-Ulrica/TileQuant

Fonte: arXiv cs.LG

Theory/Optimization • Score 85

BUMP: Brief Unexpected Mid-Interpolation Peak

arXiv:2611.08182v1 Announce Type: cross Abstract: We identify an intriguing phenomenon when linearly interpolating between permutation-aligned neural networks: test accuracy briefly improves near the interpolation endpoints before dropping toward the midpoint, which we term BUMP (Brief Unexpected Mid-interpolation Peak). We observe that BUMP appears consistently across datasets, architectures, training regimes, and remains robust under substantial perturbations such as early checkpoints and extreme pruning. Moreover, controlled ablations suggest that it cannot be attributed to trivial regularization effects. While similar near-endpoint gains are visible in prior interpolation plots, they have not been explicitly documented or systematically and thoroughly analyzed. Leveraging this effect, we propose a simple, novel ensembling scheme: weight-prediction mixed ensembling that combines weight mixing with prediction-space mixing. Across our experiments, the resulting ensembling method consistently outperforms standard logit ensembling in test accuracy and yields improved calibration trade-off, highlighting complementary benefits of weight- and prediction-space mixing.

Fonte: arXiv stat.ML

MLOps/Systems • Score 85

The Two-Ledger Model: Transaction-Grain Cost Accounting for Multi-Agent AI Systems Across Clouds, Models, and Systems of Record

arXiv:2611.06989v1 Announce Type: new Abstract: Enterprises run business transactions through multi-agent AI systems spanning clouds, large language models (LLMs), orchestration harnesses, and systems of record. Finance receives AI cost only at settlement grain, aggregated by SKU and charge period. The business question (what did this order, claim, or case cost in AI?) cannot be answered from billing data, because providers never see enterprise transaction boundaries. The FinOps Open Cost and Usage Specification (FOCUS) gives billing data one schema across providers but remains settlement-grain even as it adds AI token and requester attribution. We formalize this grain mismatch, derive seven requirements, and propose the Two-Ledger Model: a settlement ledger of FOCUS-conformant billing data joined to an attribution ledger of invocation-grain usage records built on OpenTelemetry GenAI telemetry. A layered correlation spine (trace context, A2A/MCP metadata, business-key correlation, probabilistic matching) assigns each record to its business transaction with recorded method and confidence; a versioned rate card prices usage in near real time; period reconciliation ties both ledgers to the invoice, and the unexplained residual identifies shadow AI. Costs are reported per order, claim, or outcome with agentic cost metrics. A simulation with known ground truth shows the spine sustaining cost coverage above 99% as trace propagation falls to zero, the reconciliation residual recovering planted unmetered spend within 0.24 percentage points while flagging provider billing errors, and mid-flight gating cutting p99 transaction cost by up to 61% at a 2-4% false-pause rate. We propose the attribution ledger as an invocation-grain FOCUS profile for standardization.

Fonte: arXiv cs.AI

Theory/Optimization • Score 85

Implicit Regularization via Spectral Neural Networks and Non-linear Matrix Sensing

arXiv:2402.17595v2 Announce Type: replace-cross Abstract: The phenomenon of implicit bias has attracted interest in recent years as a fundamental aspect of the remarkable generalizing ability of neural networks. In a nutshell, it entails that gradient descent dynamics in many neural nets, even without any explicit regularizer in the loss function, converges to the solution of a regularized learning problem. In this work, we show that this phenomenon extends to non-linear spectral parameterizations in matrix sensing. More precisely, we propose a matrix network architecture called Spectral Neural Networks (abbrv. SNN) where each neuron applies nonlinear activation to the singular values of the input matrix. Conceptually, this entails coordinatizing the space of matrices by their singular values and singular vectors, as opposed to by their entries, a potentially fruitful perspective for matrix learning. We prove the exponentially fast decay of squared loss during the training dynamics of an SNN parameterized matrix sensing model under gradient flow (which is a continuous approximation of the discrete gradient descent), and also that the training implicitly selects an approximate solution that minimizes nuclear norm. We also demonstrate the effectiveness of the SNN architecture in the context of matrix sensing by empirical investigations. We believe that the SNN architecture has the potential to be of wide applicability in a broad class of matrix learning scenarios.

Fonte: arXiv stat.ML

MLOps/Systems • Score 85

God Does Not Play Dice: Enforcing Batch Invariance in Agentic Kernel Generation

arXiv:2610.36623v1 Announce Type: new Abstract: Large language model (LLM) serving systems dynamically batch user requests for GPU execution, so the same request may be processed alongside different requests across runs. If the GPU kernels lack batch invariance (BI), this changing batch context can alter the numerical output for an unchanged input, undermining reproducibility. Existing kernel generation agents focus on correctness and efficiency but \emph{overlook BI}. We find 19 confirmed BI violations among 83 batch-applicable released artifacts. To address this limitation, we propose \tool, which combines contract-guided static dependency analysis with dynamic validation of bitwise consistency across batch contexts. \tool is designed as a plug-in component for general kernel agents, allowing their original optimization capabilities to be reused while adding BI enforcement. We integrate it as successive acceptance gates in the optimization loop, rejecting structurally risky or dynamically non-invariant candidates before performance benchmarking and returning feedback to the agent. Experiments across three agent frameworks under baseline and BI-aware generation on real world LLM serving tasks show that \tool improves the BI pass rate from 67.6\% to 91.9\% while retaining 91.8\% of the baseline geometric mean speedup. These results demonstrate that \tool can be incorporated into kernel optimization by agents as an additional reliability objective alongside correctness and speed.

Fonte: arXiv cs.AI

RecSys • Score 90

MAREval: A Multi-Agent Framework for Evaluating Natural Language Recommendation Explanations

arXiv:2610.36389v1 Announce Type: new Abstract: Evaluating natural language explanations in recommender systems is essential for fostering user trust, transparency, and engagement. However, existing evaluation approaches like human evaluations, while accurate, are resource-intensive and impractical at the scale required by modern recommendation platforms. Also, automated methods using single-agent LLMs suffer from prompt sensitivity and inconsistent outputs. To address these challenges, we propose MAREval, a structured multi-agent framework for evaluating recommendation explanations using large language models. MAREval orchestrates (i) a planner agent that uses a novel Chain of Debate (CoD) prompting strategy to coordinate agent roles and enforce logically consistent evaluation plans; (ii) a moderator agent that regulates discussions by mitigating prompt drift; and (iii) an arbitrator agent that aggregates outputs from multiple evaluation rounds and (iv) a Monte Carlo sampling method, improving robustness and alignment with human judgment. We conduct comprehensive evaluations on both public (TopicalChat) and proprietary recommendation datasets, demonstrating that MAREval outperforms state-of-the-art baselines. Comprehensive experiments on a public benchmark and a proprietary e-commerce dataset show that MAREval improves alignment with human judgments over strong singleand multi-agent baselines. Stability analyses indicate substantially lower variability across repeated trials. In a large human-annotation gate, MAREval meets production quality thresholds where prior evaluators fall short, and online A/B testing demonstrates statistically significant improvements in engagement and revenue metrics. These results establish MAREval as a scalable and reliable solution for human-aligned evaluation of recommendation explanations in real-world systems

Fonte: arXiv cs.AI

Theory/Optimization • Score 85

Data-driven Discovery of Lie Symmetries and First Integrals with Latent ODEs

arXiv:2610.35967v1 Announce Type: cross Abstract: We present LatentLieODE, a novel neural ODE method to discover linear Lie symmetries and first integrals of the learned dynamics from timeseries observations such as video inputs. It starts with an autoencoder, which learns a low dimensional representation of the observed system and then maps trajectories to decorrelated latent coordinates $z \in \mathbb{R}^d$. A bank of linear infinitesimal generators $\mathbf{A}_k \in \mathbb{R}^{d \times d}$ is fitted jointly against a differentiable Lie-bracket residual $\mathbf{J}_f(z)\mathbf{A}z - \mathbf{A} f(z)$, while an MLP optimises the latent ODE $f(z) = \mathrm{d}z/\mathrm{d}t$ to reconstruct the observation-space timeseries. After suitable generators and $f$ are found, we fit a third network $H_\phi$ to generate the first integral of the system. Across six benchmark systems (harmonic oscillator, simple and twin pendulums, Lotka-Volterra, pixel pendulum, sphere geodesics) we find each system's relevant symmetries, and the discovered invariants match ground-truth conserved quantities at $|\rho| \geq 0.978$. On the sphere, the pipeline recovers a non-abelian $\mathfrak{so}(3)$, certified by the negative-definite Killing form of its recovered structure constants. This work presents significant advances in data-driven discovery of symmetries and provides a new pathway for neural surrogates of dynamical systems.

Fonte: arXiv stat.ML

Theory/Optimization • Score 85

Confidence Is Not a Promise: A Reference Audit of Typed Decision Models

arXiv:2610.36311v1 Announce Type: cross Abstract: Typed decision models return a probability distribution over options supplied by the caller, and software tends to act on that number as if it were a guarantee. We measure what it licenses, comparing the open-weight Laya, in five checkpoints, with Jev 1.13, available only through an API. A probability licenses a statement about groups of decisions, not about one decision, and a gate that promises an accuracy among the cases it decides alone keeps that promise only with a lower bound checked on data like its traffic. Across 190 exchangeable evaluations, a threshold chosen by point estimate breaks the promise in 147, Learn then Test keeps it in all of them, and a Beta lower-bound rule decides more items on average than Learn then Test in 128 and fewer in none, without its guarantee. Calibrated in English, Jev keeps a promise of 95% in 8 of 51 locales. Ordinary typing noise breaks a promise calibrated on clean text, and, in exploratory runs, training with noise raises the noisy accuracy with no loss on clean text. In four further domains, a fine-tuned Laya needs from 250 to 8,000 labels to be non-inferior to Jev, and does not get there on LIAR2. Under drift, the preregistered claim for a Bayesian retraining loop is not supported: the loop avoids needless retraining where the change does not degrade the model, at a cost equivalent to recalibration alone, and re-qualifying the threshold in every window keeps the promise more often where it does. On the monthly series of M3, neither decider predicts the direction of the next value better than a constant answer. Confidence separates right from wrong answers with an AUROC from 0.80 to 0.96, but it does not select faithful explanations, and exact coalition games of single decisions show what an explanation can still say through an API that rounds to two decimals. We release the frozen answers with a datasheet, a promise check and the code.

Fonte: arXiv stat.ML

MLOps/Systems • Score 85

Collective Collaboration of Recursive Autonomous Agents

arXiv:2610.33515v1 Announce Type: new Abstract: Smart factories must translate order-level objectives into production tasks at multiple scales and combine the capabilities of individual units into an overall delivery capability. To address repeated adaptation of organizational mechanisms across equipment, workshops, and factories, as well as coordination across levels under changing production conditions, this paper develops a framework for the collective collaboration of recursive autonomous agents (RAAs). A common task and capability interface supports the recursive decomposition of production objectives from the top down and the aggregation of conditional capabilities through peer collaboration from the bottom up. Composite units can therefore participate in higher-level organizations as agents of the same kind. We formulate organizational models for leaf and composite units, specify interface closure under composition and conditions for finite task expansion, and develop operational mechanisms linking task decomposition, resource commitments, result aggregation, and coordination within the scope affected by change. A production network integrating machining, inspection, and logistics across multiple workshops illustrates how a common organizational kernel is instantiated at the equipment-service, workshop, and factory scales. The framework connects recursive organization with explicit task responsibilities, capability conditions, and execution interfaces for manufacturing systems.

Fonte: arXiv cs.LG

NLP/LLMs • Score 85

MARQ: A Modular, LLM-Agnostic Requirement Quality Assessor Infused with Systems Engineering Knowledge

arXiv:2610.36234v1 Announce Type: new Abstract: Requirements engineering (RE) is foundational to systems engineering (SE), as requirement defects can propagate downstream and adversely affect development outcomes. Although large language models (LLMs) are promising for requirement quality assessment, prior work shows that off-the-shelf models miss a large share of expert-identified defects, perform unevenly across quality criteria, and vary across executions. This study investigates whether infusing expert-derived SE knowledge can improve LLM-based requirement quality evaluation beyond off-the-shelf capability. We present MARQ, a modular, LLM-agnostic prototype that supplements LLMs with a criterion-specific knowledge stack operationalizing INCOSE requirement quality criteria and evaluated against off-the-shelf baselines on an expert-labeled NASA requirement set using Claude Opus 5, GPT-5.6 Sol, and the open-weight Gemma 3 12B, with 100 executions per model. MARQ achieves median F1 scores of 0.80 for Claude Opus 5, 0.62 for GPT-5.6 Sol, and 0.53 for Gemma 3 12B, compared with off-the-shelf baselines of 0.55, 0.52, and 0.18, respectively. However, the nature of these gains is model-dependent. Claude Opus 5 improves defect detection and false-alarm control; GPT-5.6 Sol gains detection and better execution-level consistency at the cost of more false alarms; and Gemma 3 12B reaches the performance of off-the-shelf proprietary models, although it remains highly variable. Criterion-level analysis shows gains concentrated in Correctness and Conformance, trade-offs between misses and false alarms within individual criteria, and model-dependent responses across several criteria. These findings position SE knowledge infusion as a promising model-agnostic pathway for strengthening trustworthy LLM-based RE support while underscoring the need for criterion-specific interventions and human-in-the-loop review.

Fonte: arXiv cs.AI

Theory/Optimization • Score 85

R\`enyi-Regularized Flow-Based Variational Inference for Bayesian Calibration

arXiv:2610.37169v1 Announce Type: cross Abstract: Variational inference with normalizing flows provides a flexible posterior approximation for Bayesian model calibration, but struggles when the posterior has several modes separated by regions of low probability mass. We study this behavior in a two-parameter twin experiment and regularize the variational objective. Under Gaussian observation errors and a Gaussian prior, the corresponding minimization objective is the expected Tikhonov calibration objective minus the differential entropy of the approximating distribution. However, the expected Tikhonov term favors concentration near one of the modes, while entropy maximization does not guarantee multimodality. We introduce a regularization term based on R\'enyi divergence that encourages the approximate posterior to cover the prior support. In the connected bimodal case, a flow approximation represents both modes; in the case with well-separated modes, the flow and Gaussian mixture frequently miss one mode. The introduced regularization reduces the near-collapse fraction from 50% to 25%. While the regularization improves mode coverage in this experiment, it changes the Bayesian variational objective, introduces a trade-off in approximation accuracy, and does not provide a general solution to mode collapse.

Fonte: arXiv stat.ML

MLOps/Systems • Score 85

Feature Scaling Beats Architecture Choice in Offline Respiratory Screening for Low-Resource Clinics

arXiv:2610.33030v1 Announce Type: new Abstract: Deep models are the usual choice for respiratory sound classification, but in clinics without GPUs or connectivity that choice carries a cost. We compare classical and deep classifiers on a five-class respiratory-sound corpus of 1,211 clips (asthma, COPD, pneumonia, Bronchial, healthy) from 201 patient groups, with a 3.86:1 class imbalance, using pre-registered patient-grouped cross-validation (25 runs per configuration). An SVM over 30 summary features (MFCC, ZCR and RMS statistics) reaches macro recall 0.78 and 0.71 recall on the rarest class. A 1.13M-parameter convolutional recurrent network over log-Mel spectrograms reaches 0.75 to 0.77, and no difference between model families is resolved. The largest single effect we measure is a preprocessing effect rather than an architectural one. Standardizing the SVM's features, a routine step for RBF kernels, moves its macro recall from 0.51 to 0.78, a 27-point swing wider than any gap between model families here. Augmentation did not help: SMOTE, waveform augmentation and SpecAugment have no measurable effect. Latency does not separate the models (51 to 74 ms per clip on a CPU-only laptop), but the classical models avoid a 449 MB PyTorch runtime. Clip-level splits, which let a patient's recordings fall on both sides, inflate macro recall by 7.5 to 9.8 points. Recording source is strongly tied to label (Cramer's V = 0.65 between native sampling rate and class), and in an exploratory analysis clips recorded at a class's minority rate are mostly assigned to classes typical of that rate; we state this confound as the paper's main limitation.

Fonte: arXiv cs.LG

RL • Score 85

CASEL: Consequence-Aware Selective Execution for Reliable Long-Horizon Tool-Using Agents

arXiv:2610.36570v1 Announce Type: new Abstract: Tool-using language agents increasingly modify external systems, yet most runtime safeguards allocate scrutiny by generic uncertainty scores or coarse mutation heuristics. We introduce CASEL (Consequence-Aware Selective Execution Layer), a model-agnostic runtime allocation policy that combines calibrated failure probability, state-dependent consequence, reversibility, ambiguity, and evidence freshness to choose, for each proposed tool call, among direct execution, verification, sandboxing, clarification, approval, and abstention by minimizing a per-mode cost. We formalize direct execution as selective prediction and state and test the validity conditions of a finite-sample threshold and a trajectory risk ledger. In a controlled simulation over 20 seeds with 120,000 test actions each, in which every method has the same intervention menu and the same compute cost, CASEL reduces harmful-side-effect trajectories from 14.75 percent for an uncertainty-ranked gate to 7.02 percent and mean normalized harm from 0.184 to 0.123. A gate ranked by the product of failure probability and consequence reaches similar harm (5.82 percent, 0.120) but sends 45 percent more actions to human approval, and removing reversibility-aware routing raises the rate to 15.53 percent. A trajectory-level risk budget, in contrast, increases harm at equal cost (11.82 percent). Intervention effectiveness is assumed, not measured: the results rank allocation rules under the simulator's assumptions and do not establish deployment effectiveness.

Fonte: arXiv cs.AI

Vision • Score 85

Vision Token Pruning via Query-Vision Interaction Decomposition

arXiv:2610.33215v1 Announce Type: new Abstract: Large vision-language models (VLMs), which process both visual inputs and text queries, incur high inference costs due to the large number of visual tokens, making visual token pruning a natural approach for improving efficiency. Recent work has leveraged query information to guide token selection, often through heuristics derived from attention patterns or token dynamics. However, these approaches do not explicitly exploit the underlying structure of query-vision interactions. We address this gap by observing that the query-vision interaction matrix, naturally induced by the query and key projections already present in transformer attention, is effectively low-rank, with a small number of dominant latent interaction modes capturing query-relevant semantics. Motivated by this observation, we propose Query-Vision Decomposition (QViD), a novel training-free query-aware visual token pruning method that exploits this structure. Extensive experiments on both image and video understanding benchmarks show that QViD consistently improves performance under matched token budgets, with the largest gains appearing under aggressive compression regimes.

Fonte: arXiv cs.LG

MLOps/Systems • Score 85

Skill Dreaming: Evolving Repository Skills through Pull Request Streams

arXiv:2610.36188v1 Announce Type: new Abstract: Large language model (LLM) agents increasingly rely on reusable skills for complex software engineering tasks, making skill libraries essential for repository development and maintenance. However, repository-specific skills must evolve with changing code structures, implementation conventions, and maintenance requirements. Pull requests (PRs) provide a natural source of such knowledge by recording how repositories evolve through implementation and review. Meanwhile, new PRs must follow repository-specific implementation patterns, modify related components consistently, and ensure that existing tests continue to pass while adding coverage for new functionality. This creates a mutually reinforcing cycle: PRs provide knowledge for updating skills, while these skills improve future PRs, maintain repositories, and support further skill evolution. To enable this co-evolution of PRs, repositories, and skills, we propose Skill Dreaming, a framework that learns from PR streams to guide continuous repository development and maintenance. Given a repository and its PR history, Skill Dreaming builds a repository-specific skill library by extracting executable knowledge from validated PRs and continuously refines the skills as new PR information becomes available. These skills guide coding agents not only in implementing requested features but also in maintaining associated tests, documentation, compatibility, and repository conventions. By integrating PR analysis, skill management, code execution, and continuous integration, Skill Dreaming provides an end-to-end workflow for applying learned skills to repository changes and incorporating subsequent PR outcomes into future skill updates.

Fonte: arXiv cs.AI

MLOps/Systems • Score 85

LoRA on a Budget: Function-Aware Post-Hoc Compression for Adapter Fleets

arXiv:2610.36386v1 Announce Type: new Abstract: Low-Rank Adaptation (LoRA) has become a key mechanism for efficiently specializing foundation models into many task- and user-specific variants. Serving many LoRA adapters under a shared resource budget turns post-hoc compression into a two-level allocation problem: rank must be distributed across layers within each adapter and across adapters within the fleet. Although trained LoRAs contain substantial spectral redundancy, spectral reconstruction alone does not determine how much compression an adapter can functionally tolerate. We formulate fleet compression through a functional rank-distortion frontier for each adapter, followed by fleet-level minimax allocation under a shared rank budget. Given exact frontiers, this decomposition is equivalent to joint optimization over all layer ranks. We introduce Functional Rank Allocation (FRA), an unlabeled method for constructing practical frontiers. FRA measures isolated module truncations end to end, uses these measurements as an additive search surrogate solved by dynamic programming, remeasures every complete candidate, and takes the lower-risk envelope with a complementary spectral proposal before exact minimax allocation over the constructed frontiers. Across 78 adapters from three public fleets, FRA improves worst-case and tail utility over uniform spectral compression in constrained-budget regimes. Fixed-budget probes show that isolated costs rank candidates well, but their calibration shifts across proposal shapes; end-to-end verification and the spectral proposal therefore limit functional-search error.

Fonte: arXiv cs.AI

RL • Score 85

Reinforcement Learning for Controller Parameters in the Auto-Bidding System at Pinterest

arXiv:2610.33248v1 Announce Type: new Abstract: Auto-bidding systems in online advertising must balance competing objectives-maximizing budget delivery while minimizing cost per action-across heterogeneous campaigns under non-stationary market conditions. This trade-off evolves over campaign lifespans: early on, cost optimization dominates when budgets roll over; near deadlines, budget delivery becomes urgent. While PID controllers are widely used in production for their stability and interpretability, fixed tunings cannot adapt to diverse campaigns and shifting traffic patterns. We present RLBP (Reinforcement Learning for Bidding Parameters), a hybrid architecture deployed at Pinterest that uses RL as a meta-controller to select campaign-specific PID parameters daily while retaining PID for high-frequency bid adjustments. RLBP trains a multi-head Q-network using a relative-lift reward formulation normalized against the production baseline, and performs inference-time fusion that adapts the cost-delivery trade-off to campaign states without retraining. Through online A/B testing at Pinterest, RLBP achieves a statistically significant reduction in effective cost-per-click (eCPC) with no revenue degradation, and two months of post-rollout measurements against a holdout also show consistent cost reductions. These results demonstrate that meta-controller RL can successfully augment production control systems at scale.

Fonte: arXiv cs.LG

MLOps/Systems • Score 85

WAKMan: Workload Aware KVCache Manager

arXiv:2610.36355v1 Announce Type: new Abstract: Prefix caching enables a Large Language Model (LLM) servers to reuse the Key-Value (KV) states and accelerate prefill computation stage for requests that share a prefix with earlier prompts, thus reducing latency and serving cost. With limited memory resource, the KV cache eviction policy becomes a decisive factor in determining the overall benefits of prefix caching. Production traffic may have single-turn, multi-turn, agentic, or a mixture of any combination of these request types. Our findings indicate that agentic steps return within seconds and favor recency, while popular prompts and long conversations return after many other requests and favor frequency. Across 20 mixes, no single policy is always best: vLLM's default strategy LRU provides 12% higher cache hits as compared to the frequency-oriented S3FIFO on agentic-heavy mixes but trails it by up to 22% elsewhere, and even a single workload parameter can flip the winner even for any particular mix. We present WAKMan, a workload-aware KV cache manager that picks its eviction policy from live traffic without knowing request types or reading prompts. It improves hit rate over LRU by 5 percentage points on average, with the lowest mean and worst-case optimality gap of all policies.

Fonte: arXiv cs.AI

NLP/LLMs • Score 85

LM-DSB: A Lightweight Multi-agent System for Automated Daily Safety Bulletin Generation in Construction

arXiv:2610.36433v1 Announce Type: new Abstract: Construction safety management still struggles to produce daily safety bulletins (DSBs), structured daily communications that identify site-specific hazards, convey safety protocols, and set preventive measures for workers. Manual DSB creation takes 2-3 hours daily and yields inconsistent quality, with 25% of crews never conducting pretask planning (Olson et al., 2016), diverting safety personnel from field supervision. This study proposes the Lightweight Multi-Agent DSB System (LM-DSB), which integrates retrieval-augmented generation (RAG) with a locally deployable Large Language Model (LLM) to automate DSB generation on consumer-grade personal computers. LM-DSB coordinates specialized planner, researcher, and writer agents in a hierarchical architecture, enhanced by domain-specific fine-tuning on construction safety documentation and RAG grounding in 37 health, safety, and environment (HSE) documents. A two-phase validation assessed feasibility. Phase 1 comparative evaluation showed the multi-agent RAG system (Llama 3-8B) outperformed all alternative configurations, with strong lexical alignment and coherence (TF-IDF: 0.70, BLEU: 0.39, ROUGE-1: 0.66) and near-complete document generation (length ratio: 1.01 vs. 0.39 for single-agent RAG). Phase 2 expert evaluation (N=40) confirmed practical acceptability (mean scores 3.50-4.20 on a 5-point scale), particularly in regulatory compliance (4.20) and factual accuracy (4.10). The findings show that architectural design enables lightweight models to handle safety-critical documentation on consumer-grade hardware, advancing technology democratization in construction safety management. Limitations include knowledge base scope, language constraints, and lack of field deployment validation; future work should address multimodal integration and workflow adoption.

Fonte: arXiv cs.AI

MLOps/Systems • Score 85

Plenty of Room in the Middle: Revealing Opportunities on a GPU Cluster via Fine-Grained Monitoring

arXiv:2610.36329v1 Announce Type: new Abstract: GPU clusters have become essential computing resources for AI and scientific research in academia and industry. However, their high cost and growing demand often result in long queueing delays. While previous monitoring efforts have operated at second- or minute-level frequencies, they primarily informed job-level scheduling for clusters and lacked the detail needed to optimize GPU-level utilization. This paper presents an in-depth study of a university research GPU cluster, analyzing resource usage at 100 ms intervals. We have developed a lightweight monitoring tool that leverages Nvidia's DCGM library to capture fine-grained "compute utilization" metrics within GPUs. Deployed on a cluster with 89 nodes and 319 GPUs, our tool collected data over several months. This paper focuses on a 10-day period with 7,399 jobs submitted by 150 users from 12 departments. Our study reveals three key findings: (1) the GPU cluster exhibits significant underutilization, with over 70% of jobs using less than 20% of GPU memory and 50% averaging below 5% Streaming Multiprocessor (SM) utilization; (2) 80% of jobs have a drastic mismatch between their requested and actual durations, significantly limiting the scheduler's effectiveness; and (3) the multi-GPU jobs we examined show highly periodic communication patterns. These findings suggest there is plenty of room in the middle: substantial utilization gains are available through better job scheduling and GPU multitasking, independent of hardware or application improvements. Our monitoring solution is available as open-source software at https://github.com/als244/gpu_server_monitoring.

Fonte: arXiv cs.AI

NLP/LLMs • Score 85

Developing a Speech Language Model for Basque

arXiv:2610.36911v1 Announce Type: new Abstract: Although the recent development of LLMs has paved the way for multi-modal interaction, minority languages face difficulties keeping pace. This work presents the first open Speech Language Model for Basque, designed for low-latency, end-to-end spoken interaction. Based on the LLaMA-Omni architecture, the system integrates a Latxa model as its backbone, transferring its Basque capabilities to the spoken modality by fine-tuning several modules. To address data scarcity, the English VoiceAssistant-400K corpus was translated into Basque, and the audio was synthesized using a Basque TTS. Evaluation results confirm the system's viability, achieving 3.86/5 in content quality, a 9.94% WER for spoken intelligibility, and a 3.07 UTMOS for audio quality. This work demonstrates that advanced spoken interaction systems can be developed using modular architectures and synthetic data even in low-resource languages.

Fonte: arXiv cs.CL

NLP/LLMs • Score 85

Look Only When Needed: Certificate-Guided Visual Routing for Efficient Diffusion VLMs

arXiv:2610.36671v1 Announce Type: new Abstract: Diffusion vision-language models (DVLMs) process their entire input at every denoising step. Since visual tokens make up 89-97% of this input on the benchmarks we study, most of the decoding computation is spent repeatedly processing the same image. Visual token pruning methods partially relieve this burden by removing part of the visual tokens, but they lose the removed information and still process the remaining image at every step. To measure how often the image is actually needed, we compare LLaDA-V's predictions with and without the image at every decoding step and find that removing the image changes a token being unmasked at only 22.8% of steps on average and at as few as 7.9%. We therefore propose LOWN, a decoding method that decides at every step whether to process the image and how many tokens to unmask. LOWN predicts the masked tokens without the image and processes the image only when this prediction disagrees with the last image-conditioned prediction. Agreement certifies the earlier prediction for reuse, and the number of certified tokens sets how many are unmasked in that step. We further observe that the need for the image differs across benchmarks, so LOWN's adaptive decision is needed rather than a fixed schedule. Without training or a tuned threshold, LOWN retains 99.6% of LLaDA-V's score on ten benchmarks at 35.3% of the decoding FLOPs on average. LOWN also generalizes across LLaDA-based DVLMs, composes orthogonally with visual token pruning, and runs under key-value caching for further efficiency gains. Code and project page are available at https://hyeon9698.github.io/LOWN/

Fonte: arXiv cs.CV

MLOps/Systems • Score 85

COPE: Composable Performance Emulation for Mixture-of-Experts Offloading Policies

arXiv:2610.33404v1 Announce Type: new Abstract: As LLMs scale, mixture-of-experts (MoE) architectures increase model capacity while keeping per-token computation manageable through sparse expert activation. This sparsity also enables memory-efficient deployment by retaining most expert weights in CPU memory and transferring or executing only the routed experts on demand. Existing MoE offloading systems typically target specific model-platform configurations, yet differences in model sparsity, CPU and GPU throughput, VRAM capacity, and host-device bandwidth can change the policy ranking. Because implementing and exhaustively evaluating all candidate policies incurs substantial engineering cost, efficiently selecting the best-performing policy for each model-platform configuration remains a key challenge for practical MoE offloading. We present COPE (Composable Offloading Policy Emulator), which selects the best-performing MoE offloading policy for a target model-platform configuration before runtime integration. COPE comprises two components. Its Policy Matrix factors the design space into two independent axes, expert residency and MoE execution strategy, enabling mechanism reuse across candidate policies and exposing four new candidates. Its Virtual Heterogeneous Machine profiles combined-operator schedules on the target platform and combines these measurements with offline routing traces to estimate candidate performance. Compared with the evaluated baselines, COPE improves fastest-policy selection accuracy by 24%-100% and reduces performance-prediction error by $1.7\times$ to $6.8\times$. By identifying the preferred policy before integration, COPE further reduces policy-specific implementation effort by up to 74%.

Fonte: arXiv cs.LG

MLOps/Systems • Score 85

ORBITAL: Optimization of Robust Brain--Computer Inference Through Architecture-Aware Low-Precision Compression

arXiv:2610.36411v1 Announce Type: new Abstract: Compact electroencephalography (EEG) decoders enable embedded brain--computer interfaces (BCIs), but compression can alter adversarial robustness without proportional hardware savings. We present ORBITAL, a robustness-constrained framework combining hardware-realizable structured compression, recovery-aware operating-point selection, multi-attack evaluation, and native edge validation. Across nine subjects and three seeds on BCI Competition IV-2a, S25-TRADES achieves the highest observed mean true-union robustness among compressed EEGNet models (23.06% vs 21.27% for standard Full), with 26.27% fewer parameters and 25.10% fewer MACs. Matched controls identify TRADES as the main source of robustness recovery, improving S25 union accuracy by 6.74 percentage points over clean recovery (p=0.0039), while per-example analysis reveals boundary redistribution rather than uniform margin expansion. On Jetson Orin Nano, FP16 compression reduces board energy by up to 16.20% and latency by 8.00%. Despite 99.3% prediction agreement, native INT8 Q/DQ incurs about twice the FP16 latency and energy, demonstrating the need for joint robustness and hardware-aware optimization.

Fonte: arXiv cs.CV

NLP/LLMs • Score 85

FedSEAM: Federated Continual Learning of LLMs via Subspace Expert Agreement Merging

arXiv:2610.33496v1 Announce Type: new Abstract: Federated continual learning trains a model on a sequence of tasks across several clients without sharing their data while retaining knowledge of earlier tasks. Sparse mixture-of-experts models support this setting by adapting selected parameters on a frozen backbone. However, clients learn different expert content as tasks evolve, making it difficult to share knowledge across clients without disrupting task-specific and client-specific specialization. Matching tile coordinates alone does not establish whether their learned content is compatible. We present FedSEAM, a federated continual learning framework that organizes task knowledge into sparse experts and selectively shares their content across tasks and clients. On each client, task-aware tile selection allocates capacity to a new task, while a subspace-based compatibility test determines which existing expert tiles to reuse and where to introduce fresh capacity. On the server, Subspace Expert Agreement Merging (SEAM) compares client updates as subspaces and moves each client towards consensus only along agreed-upon directions, leaving the rest of its update unchanged. We identify two remaining sources of forgetting, routing drift and merge erasure, and address them with three protections, one of which exactly preserves tile components in a selected prior-overlap subspace at every merge. We evaluate FedSEAM on TRACE using Qwen3-4B and Llama-3.2-3B, with component-wise ablations and studies of data size to assess its balance between learning new tasks and retaining prior knowledge.

Fonte: arXiv cs.LG

Vision • Score 85

SAGE: Saliency-Annealed Gated Encoder for Universal Vision Recognition

arXiv:2610.36719v1 Announce Type: new Abstract: When adapting self-supervised Vision Transformers (ViTs) to dense prediction, existing methods often treat the ViT as a passive feature extractor and may underuse the rich priors encoded in its representations, including saliency, global context, and high-frequency details. We propose SAGE (Saliency-Annealed Gated Encoder), an enhanced ViT-Adapter backbone that explicitly exploits these priors through three complementary innovations: (1) a saliency-guided architecture that extracts DINO-style class-token (CLS) saliency to guide feature gating and high-frequency enhancement, together with dynamic multi-layer aggregation and a local-detail branch; (2) an annealing-based training paradigm designed to improve convergence and feature quality during early training while reducing feature drift later in training; and (3) a two-stage, quality-aware training mechanism that smoothly transitions from focal loss to Varifocal Loss (VFL) within a single training run. Across the evaluated tasks (Fig.~\ref{fig:fig_com}), SAGE achieves state-of-the-art (SOTA) or competitive results on COCO (62.8 box AP and 55.3 mask AP), COCO keypoint estimation (78.5 AP), ADE20K (60.3 mIoU for semantic segmentation and 44.9 mask AP for instance segmentation), TinyImageNet classification (94.76\% Top-1 accuracy), and DIOR-R oriented object detection (78.8 AP$_{50}$). The same backbone is used across tasks, with task-specific decoders. To the best of our knowledge, SAGE is the first framework to achieve multi-task state-of-the-art results on Huawei Ascend NPUs. Code and pretrained models will be released after internal review.

Fonte: arXiv cs.CV

NLP/LLMs • Score 85

One Call Ahead: Packed Self-Speculative Decoding with Diffusion Drafting

arXiv:2610.33379v1 Announce Type: new Abstract: Autoregressive (AR) language models generate tokens sequentially, so generating long outputs is costly. Speculative decoding mitigates this by using a lightweight drafter to propose multiple tokens for the target AR model to verify in parallel, while a diffusion-based drafter can produce an entire draft block in one forward pass. In models with AR and diffusion modes, one network both drafts and verifies, but drafting still waits for verification, so the standard schedule needs two sequential model calls per iteration. Existing one-call schedules avoid the wait but incur input size quadratic in the speculative width. Motivated by this efficiency bottleneck, we introduce Packed-SS, a self-speculative decoding schedule that verifies and drafts in a single model call with input size linear in the speculative width, while preserving the target model's distribution. The key idea is to let the diffusion drafter run one call ahead: while verifying the current draft, the model simultaneously constructs the next draft from an older verified prefix and keeps its unresolved suffix for the next iteration. Under a mild monotonicity assumption on the drafter, we further show that Packed-SS requires no more model calls than the standard two-call decoding in the greedy setting. Experiments on Nemotron 3B/8B and SetDLM 1.7B show higher throughput and lower mean wall time than both baselines under greedy decoding.

Fonte: arXiv cs.LG

Evaluation/Benchmarks • Score 85

CrowdCC: Jointly Benchmarking Corruption Robustness and Deployment Quantization in Crowd Counting

arXiv:2610.36340v1 Announce Type: new Abstract: Crowd-counting models deployed on edge devices face two shifts: inputs are degraded along the capture and transmission pipeline, while models are quantized to satisfy resource constraints. Yet corruption robustness and deployment quantization are typically evaluated separately. We introduce CrowdCC, a controlled corruption benchmark that jointly evaluates these effects using 23 deployment-motivated corruptions at three severity levels, four crowd-counting datasets, and 26 models using A100 TensorRT engines at FP32, FP16, and post-training INT8 precision, paired with throughput and energy measurements from corresponding Jetson Orin NX configurations. The corruptions span six families, with head annotations adjusted when geometry or visibility changes. Our evaluation shows that capture-chain corruptions consistently produce the largest degradation across the evaluated models, while crowd density affects absolute and relative error differently. Clean accuracy does not fully predict corrupted performance. INT8 alters corruption robustness substantially more than FP16, with no systematic direction, and does not consistently improve latency or energy. INT8 inference efficiency is strongly associated with reformat overhead in the compiled engine, while slowdowns are also observed under limited INT8 coverage. Precision changes corrupted-error rankings and robustness-cost frontiers, showing why model selection must consider corruption robustness and measured deployment cost together. The code is available at https://github.com/Tulane-Computer-Architecture-Systems/CrowdCC-Release.

Fonte: arXiv cs.CV

MLOps/Systems • Score 85

Hierarchical Low-Rank Reparameterization for Parameter-Efficient

arXiv:2610.37068v1 Announce Type: new Abstract: Parameter-efficient fine-tuning (PEFT) methods adapt large models by introducing a small number of trainable parameters. However, we find that these parameters still contain redundancy and can be further compressed while maintaining strong task performance. To this end, we propose a novel hierarchical low-rank reparameterization method. Its core idea is to reorganize the trainable parameters of PEFT methods into an approximately square matrix and then express it via a low-rank decomposition, which can be applied recursively. This forms a multi-level structure with factors compressed level by level, drastically reducing the number of trainable parameters. In our experiments, we combine this reparameterization with LoRA, DoRA, MoSA, and RaSA. We evaluate commonsense reasoning on LLaMA-3.1-8B and Qwen2.5-7B, with additional GSM8K math reasoning experiments on LLaMA-3.1-8B. Under comparable trainable-parameter budgets, our method combined with LoRA or DoRA achieves better performance. Furthermore,when the parameter count of the original methods is compressed by approximately an order of magnitude, our method can still maintain competitive task performance and even surpass the corresponding baselines in average accuracy across the eight commonsense reasoning benchmarks.

Fonte: arXiv cs.CL