Every paper checked — nothing hidden

All papers

Every paper checked in the week of 12–18 Oct 2026 · Small and efficient models — nothing is hidden. Brief papers first.

All2,430 In brief3 More (on-topic)455 Not relevant1,957 Not assessed6 Older27 Could not confirm6 Blogs with no paper16 All blogs19

Paper	Topic	Relevance
Backward Compatibility in Zero-Bit Neural Audio Watermarking Establishes a training approach to ensure backward compatibility and prevent false positives across versions of neural audio watermarking. Preprint	Small and efficient models via arxiv · 17 Oct, 06:18	Not relevant Does not involve model pruning, distillation, quantization, or efficient serving.
Robust Reinforcement Learning for Dynamic Treatment Regimes under Model and Measurement Uncertainty Develops a robust reinforcement learning framework for dynamic treatment policies under patient measurement and model uncertainty. Preprint	Small and efficient models via arxiv · 17 Oct, 06:18	Not relevant Does not address model compression, quantization, or efficient inference techniques.
Lean Graph Search for Formal Theorem Proving Introduces a graph-based retrieval method that traces multi-hop library dependencies to find lemmas for formal theorem proving. Preprint	Small and efficient models via arxiv · 17 Oct, 06:18	Not relevant Addresses premise retrieval algorithms, not model quantization or efficient inference.
PathFM: Fast Generative Forecasting of Multivariate Trajectories Uses flow matching and distillation to generate full multivariate time-series trajectory forecasts rapidly. Preprint	Small and efficient models via arxiv · 17 Oct, 06:18	Not relevant Distills time series trajectory generators rather than small language models.
Learning Topology-Aware Representations for Mechanics-Relevant TPMS Metamaterials Applies persistent homology to encode global structural connectivity into learned representations of mechanical metamaterials. Preprint	Small and efficient models via arxiv · 17 Oct, 06:18	Not relevant Focuses on topology-aware representation learning rather than model compression or efficient inference.
Purify Before You Align: Improving Representation Alignment for Diffusion Models Introduces local appearance purification to filter alignment targets and improve representation learning during diffusion model training. Preprint	Small and efficient models via arxiv · 17 Oct, 06:18	Not relevant Studies visual representation alignment rather than model compression or efficient inference.
Pixels to Tokens: Token Space Efficient Active Learning for Low-Budget Semantic Segmentation Selects maximally informative pixel tokens directly from vision transformers to train segmentation heads under tight annotation budgets. Preprint	Small and efficient models via arxiv · 17 Oct, 06:18	Not relevant Covers active learning annotation budgets rather than model compression or efficient inference.
Audio-Visual Dog Identification and Tracking in Real-World Videos Combines visual tracking and vocal fingerprinting to identify and monitor individual dogs across real-world multi-dog video recordings. Preprint	Small and efficient models via arxiv · 17 Oct, 06:18	Not relevant Focuses on multi-modal tracking, not model compression or serving.
Spectral-Spatial Interpretation Introduces a Shapley-based attribution method that jointly explains spatial and spectral channel importance in multispectral imagery. Preprint	Small and efficient models via arxiv · 17 Oct, 06:18	Not relevant Focuses on feature attribution, not model compression or efficiency.
Semantic Latent Alignment for Flow Matching with Invertible Bi-Lipschitz Adapters Proposes invertible bi-Lipschitz adapters to align pretrained latent spaces with semantic representations for flow matching without retraining encoders. Preprint	Small and efficient models via arxiv · 17 Oct, 06:18	Not relevant Adapts generative latent spaces, not efficient serving or quantization.
When Should a Low-Light Pipeline Generate? Auditable Risk Control for Selective Generative Enhancement Presents an auditable framework that uses risk budgeting to selectively deploy generative enhancers over base restoration models in low-light images. Preprint	Small and efficient models via arxiv · 17 Oct, 06:18	Not relevant Focuses on selective generative enhancement, not model compression or serving.
L2GSSL: Label-Efficient Local-to-Global Self-Supervised Learning for Volumetric Healthcare Data Introduces a label-efficient self-supervised learning framework that splits volumetric CT scans into local slice and global context representations. Preprint	Small and efficient models via arxiv · 17 Oct, 06:18	Not relevant Focuses on label-efficient CT pretraining rather than model compression or inference efficiency.
Rethinking Classifier-Free Guidance for TI2I Flow Matching Identifies early-stage issues with classifier-free guidance in flow matching models and proposes a correction for controllable image editing. Preprint	Small and efficient models via arxiv · 17 Oct, 06:18	Not relevant Focuses on flow guidance mechanics rather than model compression.
International Multi-Centre Validation of Unsupervised Domain Adaptation for Precise Discrimination between Normal and Abnormal Retinal Development Validates an unsupervised domain adaptation model to classify normal and abnormal retinal development across diverse optical coherence tomography systems. Preprint	Small and efficient models via arxiv · 17 Oct, 06:18	Not relevant Focuses on domain adaptation, not model compression or efficiency.
Single-stage joint diffusion of sky images and clear-sky index for ultra-short-term irradiance forecasting: what the frame-to-irradiance stage discards Proposes a single-stage joint diffusion method to forecast solar irradiance directly from sky images without requiring a separate regression model. Preprint	Small and efficient models via arxiv · 17 Oct, 06:18	Not relevant Not about language model pruning, quantization, or serving.
Temporal Periodic Image Registration with Implicit Neural Representations The paper applies implicit neural representations with temporal consistency constraints to register periodic medical image sequences across time. Peer-reviewed	Small and efficient models via arxiv · 17 Oct, 06:18	Not relevant Not about language model compression, distillation, or serving.
Beyond the Dominant Label: Recovering Coexisting Lesion Evidence with Multiple Instance Learning The paper introduces a multiple instance learning method to recover coexisting secondary lesions from whole-slide pathology images labeled only by dominant severity. Preprint	Small and efficient models via arxiv · 17 Oct, 06:18	Not relevant Does not focus on model compression, quantization, or efficient serving.
FutureLift: Lifting Trajectory Proposals into Bidirectional Future Graphs for Motion Forecasting The paper presents a refiner that models motion proposals as heterogeneous temporal graphs to improve trajectory forecasting for autonomous vehicles. Preprint	Small and efficient models via arxiv · 17 Oct, 06:18	Not relevant Does not focus on model compression, quantization, or efficient inference architectures.
How Much Physics Can Vision Learn from Audio? Probing and Distilling Damping Structure The paper investigates learning physical damping properties from impact audio and distills this knowledge into vision models. Preprint	Small and efficient models via arxiv · 17 Oct, 06:18	Not relevant Addresses cross-modal distillation for physics probing, not model compression or serving.
Privacy-Preserving Text Inference for Health-Misinformation Classification and Semantic Retrieval Designs a homomorphic encryption framework for privacy-preserving text classification and semantic retrieval on health misinformation queries. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Addresses cryptographic privacy preservation rather than model compression or inference efficiency.
Efficient Agnostic Learning of Univariate Densities Under Local Differential Privacy Proves theoretical sample bounds and presents algorithms for learning univariate probability distributions under local differential privacy. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Covers theoretical distribution learning algorithms rather than small, efficient language models.
Clean Data Can Still Carry Backdoors: Support-Persistent Backdoors for Model Reuse Introduces a clean-data backdoor method that persists through standard knowledge distillation by targeting naturally occurring feature subspaces. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Investigates backdoor attacks surviving distillation rather than designing efficient or small models.
Federated Art Chase: Privacy Leakage and Data Poisoning in Federated Fine-Tuning Studies label privacy leakage and data poisoning risks during head-only federated fine-tuning of vision models across distributed clients. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Studies federated learning attacks rather than efficient model inference.
CascadeLedger: Counterfactual Auditing of Information and Resource Integrity in LLM Cascades Introduces an auditing framework that checks counterfactual execution traces to ensure language model routing uses only legitimate information. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Not focused on model compression, pruning, or efficient inference design.
APRiL: An Open-ended Benchmark for Chinese Social Awareness in Large Language Models Presents a benchmark with open-ended tasks and rubrics to assess social awareness and response generation in Chinese. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Does not evaluate model compression, distillation, or inference efficiency.
PRISM: Transfer and Limits of Activation Interventions in Tool-Calling Language Models Studies the transferability of internal activation interventions to reduce unsafe tool execution across model precisions and languages. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Uses quantization only as an auxiliary tool, not efficient model design.
Gibbs-Gap Policy Optimization for Off-Policy RL with Verifiable Rewards Introduces an off-policy reinforcement learning algorithm for language models using verifiable rewards without importance sampling weights. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Focuses on RL optimization rather than small models or efficient inference.
PEDBench: Can Speech Models Detect Phonetic Errors? Introduces a benchmark combining natural, scripted, and synthetic audio to evaluate phonetic error detection in speech models. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Does not evaluate model compression, distillation, or serving efficiency.
Does Morphological Correctness Matter for Tokenization? A Scrambled Control Study in Tagalog Tests whether linguistically accurate morpheme boundaries in tokenizers improve translation performance compared to scrambled control segmentations. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Studies morphological tokenization rather than model compression or inference efficiency.
CogDrift-4D: Hierarchical Multimodal Modeling of Temporal Behavioral Decay for MCI Detection Proposes a hierarchical multimodal transformer to model long-term behavioral changes across multi-session interactions for cognitive impairment detection. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Does not investigate model compression, distillation, or efficient serving.
Task-Centric Multimodal Learning for Cognitive Impairment Detection Evaluates a multimodal screening method using voice, video, and text from voice-assistant interactions to detect cognitive impairment. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Does not address model efficiency, quantization, or serving compression.
VELD-QA: A Large-Scale Dataset for Visual Evidence Grounding and Question Answering in Long Documents Presents a benchmark with human-annotated visual evidence to evaluate retrieval and question answering across long documents. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Does not cover model compression, distillation, or efficient inference serving.
One Bias Score, Two Factors: How Mitigations Affect Option Bias in Multiple-Choice Evaluation Deconstructs multiple-choice option bias in language models into two factors tracking answer flips and positional skew. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Not about model compression, distillation, or serving efficiency.
Can LLMs Anticipate Their Cruxes? Explores methods to extract the underlying decisive sub-questions that drive language model decisions across diverse generated outputs. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Does not explore model quantization, pruning, or small models.
INCONVENIENT: A Benchmark to Assess LLM Generalization Capabilities from Convenience Samples Introduces a benchmark to test model generalization under systematic convenience sampling biases in tabular data. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Does not focus on model compression, quantization, or efficient deployment.
SHIP: Structured and Hierarchical Insight with Privacy The paper proposes SHIP, an architecture for extracting structured differentially private representations from text for analytical queries. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Does not address model pruning, distillation, or efficient inference.
Multimodal AI Detection for Math-Heavy Scientific Text The paper examines AI-generated text detection in scientific papers by combining visual formatting, PDF text, and LaTeX representations. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Not about model compression, quantization, or lightweight architectures.
From Observation to Intervention: Distilling Self-Generated In-Context Demonstrations via Executable Causal Intervention The paper introduces a causal intervention framework to distill transferable self-generated in-context learning demonstrations. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Not about parameter pruning, quantization, or efficient model serving.
Stop Training Against Soft Users: Mentalized Adversarial Self-Play for Proactive Dialogue The paper introduces an adversarial self-play framework modeling user mental states to improve proactive dialogue agent planning. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Does not explore model compression, distillation, or inference optimization.
Automatic Sentence Completion Using Bengali Audio Speech with a Transformer-Based Early Speech Recognition System The paper presents an automated speech recognition and sentence completion system to assist dementia patients speaking Bengali. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Does not focus on general model compression, quantization, or serving.
Beyond Retrieval Failure: Understanding Why LLMs Miss Evidence in Multi-Hop RAG The paper introduces ExecutionRAG to show how search drift causes apparent retrieval failures in multi-hop question answering. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Does not study model compression, parameter reduction, or quantization.
KBS: A Korean Benchmark Suite for Foundation Model Evaluation The paper introduces the Korean Benchmark Suite to evaluate foundation models across six natural language capabilities. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Does not study model efficiency, distillation, or quantization techniques.
What Should We Remove? Evaluating the Reliability and Costs of RLVR Data Decontamination The paper audits data decontamination methods on math corpora to quantify false removals and benchmark leakage risks. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Does not focus on model compression, quantization, or efficient architectures.
FGSS-BO: Feedback-Guided Hypothesis Management for LLM-Based Bayesian Optimization The paper proposes FGSS-BO, adding structured hypothesis management to language-model-guided Bayesian optimization. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Does not examine model compression, pruning, or low-cost serving.
Isolating a Language Model's Judgment with a Label-Free Subspace The paper proposes isolating internal model judgments from instruction formatting variations using a label-free hidden state subspace. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Not focused on model compression, distillation, or efficient serving.
TimeCap: A Scalable Open-Source Pipeline for Captioning Wearable Health Time-Series Data The paper introduces TimeCap, an open deterministic pipeline for generating signal-grounded captions from wearable health time-series data. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Not about model quantization, distillation, or efficient model serving.
Making every stage of medical device document assessment with large language models deterministic The paper evaluates output non-determinism in commercial language model deployments during medical device regulatory compliance assessments. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Focuses on determinism and reproducibility rather than model compression or distillation.
VALCON: Benchmarking and Enhancing Value Consistency in Large Language Models via Dual-Role Variation The paper introduces VALCON to evaluate and improve value consistency when language models alternate between generator and evaluator roles. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Addresses behavioral alignment consistency rather than model compression or efficient serving.
ZoomCap: Evidence-Routed Visual Zooming for Recoverable Visual Information Transfer The paper introduces ZoomCap, an adaptive visual zooming and evidence-routing framework to improve fine-grained image captioning supervision. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Focuses on visual captioning quality rather than model compression or inference efficiency.
VoteGEO-Bench: Measuring the Impact of Generative Engine Optimization on Electoral Visibility The paper presents VoteGEO-Bench to measure how rewriting web content influences information visibility in generative search engine election queries. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Focuses on generative engine optimization rather than model compression or efficient serving.
QuasiLeak-Bench: Evaluating LLM-Based Privacy Detectors on Policy-Conditioned Quasi-Evidence Group Recovery The paper introduces QuasiLeak-Bench to evaluate whether privacy detectors can identify distributed quasi-identifiers governed by disclosure policies. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Examines privacy detection benchmark performance, not model compression or efficient serving.
OpCredit: Operation-Aware Policy Learning for Agentic Knowledge Graph Question Answering The paper proposes OpCredit, a policy learning framework that assigns individual operation credits for agentic knowledge graph question answering. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Focuses on reinforcement learning credit assignment rather than model compression or inference efficiency.
Emotion Is Not External Knowledge: Internal Attention Signals for Affective State Reasoning The paper presents an internal attention boosting method to improve how language models infer implicit conversational affective states. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Studies attention steering rather than model compression, distillation, or serving efficiency.
RenoLayout: Unified Floorplan Renovation and Furniture Layout for Constraint-Aware Residential Design The paper introduces RenoLayout to generate residential renovation floorplans and furniture layouts under structural and physical constraints. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Does not address model compression, quantization, distillation, or serving efficiency.
Beyond Invariance: Factor-wise Selective Treatment of Shifts for Multimodal Learning Proposes a factor-wise representation framework to selectively treat different distribution shifts in multimodal machine learning. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Addresses out-of-distribution multimodal learning rather than model compression or inference efficiency.
Identifying Effective Strategies in Prompt Evolution for Stress and Emotion Prediction Evaluates prompt optimization methods and identifies effective prompting strategies for social media stress and emotion classification. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Analyzes prompting strategies rather than model compression, distillation, or efficient inference.
SITAR: A Situational Distortion Metric for Automatic Speech Recognition Evaluation Develops a lightweight situational distortion metric to evaluate speech recognition errors based on practical conversational severity. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Builds a lightweight metric regressor rather than studying model compression or efficient serving.
OPRD: On-Policy Reminder Distillation for Long-Horizon Coding Agents Uses multi-scope reminders and on-policy distillation to help coding agents maintain constraints over long software engineering interactions. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Uses distillation for agent training, not for model compression or efficient serving.
Writability and the Pathway: A Controlled Study of Knowledge Injection into Addressed Memory Conducts a controlled evaluation of direct knowledge injection into addressed conditional memory modules across small and production models. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Tests post-hoc parameter updates rather than model compression, quantization, or distillation.
ContextLift: Breaking the Spurious Correlation Behind Conditional Emergent Misalignment Introduces a data augmentation method that breaks spurious contextual correlations causing emergent misalignment during domain finetuning. Preprint	Small and efficient models via arxiv · 17 Oct, 06:17	Not relevant Does not involve model quantization, pruning, or efficient serving.

601–660 of 2,430← Previous Next →