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
GeopoliticalQA: Benchmarking Retrieval and Generation over Intelligence Reports Introduces a benchmark of 5,525 QA pairs from intelligence reports to evaluate RAG document- and chunk-level retrieval. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Compares embedding sizes as a finding, but does not contribute model compression methods.
perturber: A Stateless Service for Semantics-Preserving Text Perturbations Presents a stateless service and library providing isolated, semantics-preserving text perturbations for language model robustness testing. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not develop efficient architectures, distillation, or quantization techniques.
Activating the Annotator: Operationalising Lived Experience in Data Annotation Tests whether priming annotators with scenario vignettes and micro-narratives improves the operationalization of lived experience in subjective annotation. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not explore model compression, pruning, or efficient inference.
U4RASD at KnowledgeGraphEval-2026: Ontology-Constrained Encoder and LLM Systems for Arabic Relation Extraction Describes an ontology-constrained ensemble system of encoders and fine-tuned language models for directed Arabic relation extraction. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Uses standard fine-tuning (QLoRA) as an implementation detail rather than advancing model efficiency.
MADS: Multi-Agent Dialog Simulation for Robust Conversational Agents Generates synthetic multi-agent dialogues to benchmark conversational agent robustness against non-cooperative and adversarial user behaviors. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not investigate distillation, pruning, quantization, or hardware efficiency.
DistressBench: Evaluating Multi-Dimensional Crisis Support in Large Language Models Introduces an evaluation benchmark measuring how effectively language models provide supportive responses during self-harm and crisis disclosures. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not study model compression, efficient architectures, or hardware optimization.
Large Language Models, Machine-Assisted Cultural Analytics, and the Agentic Turn Examines methodological and conceptual implications of using large language models and agentic workflows for machine-assisted cultural analytics. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not address model efficiency, quantization, or on-device inference.
People Making Conversation: Evaluating How Naturally Speech Models Converse Evaluates whether real-time speech models decide to speak appropriately during natural pauses and turn ends in unscripted conversations. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not propose architectural compression, pruning, or efficient inference methods.
When Charts Deceive Web Agents: A Paired Benchmark for Misleading Visualization-Grounded Tasks Introduces a benchmark evaluating how misleading visualizations and charts affect the multi-step execution and decisions of web agents. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not introduce techniques for model compression, distillation, or efficient serving.
Fear Is Not Always Fear: Variation in Emotion Label Equivalence Across Datasets Tests cross-corpus emotion classification transfer across three datasets, finding significant loss in discrimination and conceptual disagreement on labels. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not investigate model compression, distillation, quantization, or inference acceleration.
Unreliable Narrator Framing as a Cost-effective Mechanism for Satire Explanation of Australian and American Satirical News Evaluates whether framing explanations around unreliable narrator theory enables open-weight models to explain satirical news cost-effectively. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Only uses open-weights models as a baseline; does not propose compression, quantization, or serving efficiencies.
Temporal Evidence Forests for Long-Horizon Multi-Party Conversational Memory Presents a temporal memory architecture using hierarchical tree structures to trace evolving multi-party conversational histories for agents. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Focuses on memory representations rather than model compression or inference efficiency.
SEEDS: Preview-Guided Evidence Admission for Search Agents Introduces a preview-guided admission framework allowing search agents to selectively incorporate retrieved candidates into the ongoing reasoning context. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Addresses context management in search, not model compression or quantization.
MVGraph: Benchmarking LVLMs on Evidence Composition across Multi-View Visual Graph Introduces a benchmark evaluating vision-language models on graph-theoretic reasoning where evidence is split across multiple visual graph views. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not address model compression or serving optimization.
Forecasting Corpus-Sampling Disagreement in Masked Text Reconstruction Derives a formula to forecast disagreement between masked text reconstruction models fitted on independent samples from the same population. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not involve model pruning, quantization, or inference serving.
Language Models Have Interpretable Narrative Mechanisms Identifies internal attention heads, neurons, and sparse autoencoder features that causally influence narrative representation in language models. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not investigate model quantization or serving efficiency.
Evaluating LLM Creativity through Long-Tail Performance Formalizes language model creativity evaluation as long-tail input-output distribution performance and tests it across multiple benchmarks and models. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not address model compression or inference speed.
Evo$^2$: Evolving the Evolver through Weight-Harness Co-Evolution Proposes a co-evolution framework that coordinates agent harness prompt modifications with model weight updates using self-generated experience. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Deals with agent harness co-evolution, not model quantization or serving efficiency.
CliQ-RRG: Clinical-Knowledge Guided Disease-aware Visual-Textual Alignment for QA-Style Radiology Report Generation Presents a two-stage visual-textual alignment and retrieval-augmented framework for generating structured radiology reports from medical scans. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not focus on model compression or inference efficiency.
From 2.1M Hours of Raw Broadcast Audio to Finnish Speech SSL: Confidence Filtering and Architectural Analysis at Scale Evaluates speech self-supervised learning models on Finnish broadcast audio using ASR confidence filtering and controlled architectural comparisons. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Studies speech SSL data filtering and architecture comparisons, not model compression.
BenCoDiC: A Large-Scale Contextual Cyberbullying Corpus for Bengali Political News Discourse FULL VERSION Presents a multi-dimensional annotated corpus of Bengali political YouTube discussions to support research on contextual cyberbullying detection. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not explore model compression, distillation, or efficient inference.
Les Misérables Dialogue: A film-based network representation of social communication Constructs a temporal and multilayer social dialogue network dataset based on the film adaptation of Les Misérables. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Not related to model efficiency, quantization, or serving.
SOAR: Source-Aware Reinforcement Learning with Precise Credit Assignment for Search Agents Proposes a reinforcement learning framework that tracks external versus internal knowledge sources to reduce search agent fabrication. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Focuses on credit assignment and grounding, not inference acceleration or quantization.
Scaling Agentic Data for Long-Horizon Terminal Intelligence Introduces a synthetic data generation framework to produce long-horizon task trajectories for terminal agents. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Focuses on agentic training trajectories rather than model compression or inference efficiency.
SharP: Soft and Hard Prompt-Guided Augmentation with LLM for Low Resource Fake News Detection Proposes a prompt-guided text augmentation framework to assist fake news detection in low-resource settings. Peer-reviewed	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Refers to low-data regimes, not small model compression or inference efficiency.
What Do Audio-Language Judges of Generated Music Listen For? Four Probes with Controls from Expert Ratings Designs diagnostic probes to determine whether audio-language evaluation models judge music based on audio or prompt content. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not address model compression, pruning, or efficient serving.
Vision Transformers Learn Word Meaning from Images Alone Tests whether self-supervised vision transformers represent semantic meanings of rendered words without paired text data. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not focus on model compression or inference efficiency.
Beyond the Final Embedding: Retrieving Memories from Token-Level Evidence Introduces a retrieval method combining token-level matching and document-level embeddings for memory-augmented language models. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Focuses on retriever ranking performance rather than model compression.
Do Prompt Optimizers Break Requirements They Are Not Scored On? A Preregistered Study Evaluates whether automated prompt optimizers inadvertently cause regressions in unscored prompt constraints and requirements. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not study model compression, distillation, or serving.
Host-Aware Web Crawling for LLM Pretraining Data Studies host-level budget allocation and crawling policies to collect pretraining web data for language models. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not focus on model compression, quantization, or efficient inference.
Exploring Attention Sinks in Text-to-Image Transformers from Diffusion and Autoregressive Models Analyzes the emergence and behavior of attention sinks in text-to-image diffusion and autoregressive Transformers. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Focuses on attention dynamics in generative models rather than compression or serving.
Test-Time Memory Construction Presents a framework that defers memory construction to test time to answer queries across long interaction histories. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Focuses on memory management rather than small models, quantization, or pruning.
From Structure to Validity: Mechanism-Guided Generation and Refinement of LLM Rubrics Analyzes rubric construction mechanisms and introduces a tree-search method to refine rubrics for evaluating language model outputs. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not investigate model compression, distillation, or efficient inference.
MoPrO-SD: Modular Prompt Optimization for Stance Detection Presents a modular prompt optimization framework and benchmark targeting stance detection in text. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Focuses on prompt refinement rather than model compression or efficient serving.
Towards the Roof of the World: NLP Progress in Tibetan Languages Surveys twenty-two years of Tibetan natural language processing literature across corpora, speech processing, and language models. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not focus on model compression or efficient inference methods.
Does Financial Speech Need Continual Learning Regularization? A Study of EWC and mini-KFAC on Earnings Call ASR Investigates continual learning regularization techniques for speech recognition during domain adaptation on financial earnings calls. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Investigates catastrophic forgetting regularization rather than efficiency, quantization, or distillation.
Revisiting Accent Classification with DistilHuBERT: Speaker-Disjoint Evaluation and Cross-Corpus Generalization Evaluates DistilHuBERT-based accent classification under speaker-disjoint and cross-corpus conditions to measure generalization to unseen speakers. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Uses DistilHuBERT as a test subject rather than proposing model efficiency methods.
Towards Enriching Design-Time Models with Evidence from Simulation and Operational Activities Proposes a workflow to enrich design-time cyber-physical system models using empirical evidence collected from simulations and operational runs. Peer-reviewed	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not involve model compression, small models, or efficient serving.
A Better Metric Will Not Save You: What You Compare Beats How You Compare It Analyzes representational similarity metrics and demonstrates that input category selection has a larger impact on comparisons than the choice of metric. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Investigates representational similarity metrics rather than model compression or efficiency.
Proactive Agentic Orchestration Introduces POrchestra, an agentic orchestration framework enabling sub-agents to proactively communicate with the orchestrator during dynamic execution. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Addresses orchestration policy learning, not model compression or small-model serving.
S-Space: Exploring Spatial Workspace in Multimodal Language Models The paper identifies a linear spatial workspace subspace in multimodal language models that encodes continuous 3D coordinate representations. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Explores internal representation geometry without focusing on model compression or serving efficiency.
Personalization Without Stereotyping: Do Speech LLMs Know When to Turn a Deaf Ear? The paper introduces a benchmark to measure demographic robustness, stereotyping, and intentional personalization in speech language models. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Evaluates speech models without addressing small models, distillation, or compression.
Not All Correct Fixes Run Alike: An Empirical Study of Patch Performance in Repository-Level Issue Resolution The paper analyzes the runtime performance differences of test-passing issue resolution patches generated by language models on SWE-bench. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Measures runtime performance of generated patches, not language model compression or serving.
Measuring Artifact Reliance in Machine-Generated Text Detectors The paper demonstrates that machine-generated text detectors rely heavily on document formatting and PDF extraction artifacts rather than text quality. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Focuses on artifact reliance in text detectors, not model compression or efficiency.
BLiMPPI: Benchmark of Linguistic Minimal Pairs for Polarity Items The paper presents a multilingual benchmark of minimal pairs to evaluate language model sensitivity to negative polarity items across 33 languages. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Evaluates existing multilingual models without focusing on model compression or efficiency.
Training User Simulators with Latent Intrinsic Feedback The paper trains conversational user simulators equipped with latent evaluation thoughts and scalar ratings to provide intrinsic feedback. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Studies conversation feedback simulation without addressing model compression or efficiency.
APEX-Harm: Evaluating Agent Safety in Shared Professional Workspaces The paper introduces a benchmark evaluating agent susceptibility to prompt and script injections across multi-tool shared workspace environments. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Assesses frontier model vulnerabilities without focusing on model compression or small models.
AMLworld-Compact: Importance-Weighted Downsampling for Cost-Effective LLM Evaluation and Error Diagnosis The paper presents an importance-weighted downsampling protocol to lower evaluation costs on transaction graph datasets without distorting precision metrics. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Addresses evaluation sample efficiency rather than model compression or efficient model serving.
Retractions Revisited: A Replication, Extension, and Mechanistic Evaluation of Linguistic Obfuscation Effects in Fraudulent Science The paper evaluates linguistic obfuscation patterns and reality-monitoring mechanisms across fraudulent and genuine scientific publications. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Investigates linguistic features of fraud without addressing model efficiency or compression.
Verifiable Retrieval for Legal Reasoning through Structure-driven Knowledge Graph The paper evaluates knowledge-graph-informed chunking and retrieval strategies over German legal code to support verifiable retrieval-augmented generation. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Focuses on retrieval structuring rather than model compression, quantization, or small models.
Enhancing Legal Reasoning in Pre-trained Language Models via Knowledge Graph-Guided Multi-Task Pre-training The paper proposes multi-task pre-training guided by knowledge graphs to improve zero-shot legal reasoning in language models. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Studies domain pre-training techniques without focusing on model compression or inference speed.
OccludeBench: Can VLMs Find What's Missing? The paper presents a benchmark evaluating how effectively vision-language models recover fully occluded objects using remaining visual cues. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Studies benchmark performance of VLMs without focusing on small models or compression.
Not Just an Evaluation: A Case Study of Discourse Particle Sensitivity in Language Models The paper studies language model sensitivity to discourse particles such as 'just' and 'only' in natural conversations. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Investigates internal representations of discourse particles without addressing model efficiency.
REFLECT: Role-Aware Detection of Sentiment, Emotion, Politeness, and Formality The paper introduces a transformer architecture incorporating semantic role labeling into attention mechanisms to detect sentiment, emotion, politeness, and formality. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Proposes role-aware attention mechanisms without addressing model compression or efficiency.
Error Rate is Not Enough: Costs of Augmentation in Low-Resource ASR The paper analyzes phoneme-level error patterns from audio and text data augmentation techniques in low-resource automatic speech recognition. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Studies low-resource training data augmentation rather than model compression or inference efficiency.
Tracing and Taming Hallucinated Terms via Subsequence-Association Filtering and Breadth-Aware Selection for LLM Query Expansion The paper proposes a training-free query expansion method that filters hallucinated terms using corpus-conditioned cross-context reproducibility. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Presents an IR query expansion technique, not model compression or efficient inference.
QTestBench: A Test Framework for Evaluating LLM-Generated Quantum Code The paper introduces a benchmark framework evaluating LLM-generated quantum software unit tests against independent ground-truth checkers. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Evaluates LLM capabilities without investigating model compression or efficient serving methods.
Can Language Models Learn When to Abstain? Policy Learning and Calibrated Answer Selection The paper studies policy learning and answer-selection mechanisms to help language models decide when to abstain on factual questions. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Uses a 14B model but does not study distillation, quantization, or inference acceleration.
OpenPA: Forced-alignment-free Multi-granular Pronunciation Assessment with Mixed-transcript Adaptation The paper presents an automatic pronunciation assessment method using Whisper cross-attention to predict word- and utterance-level scores without forced alignment. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Focuses on speech assessment accuracy rather than model compression or serving efficiency.
Structural Automata for Neural Semantic Parsing The paper proposes accepting-path decoding using structural automata to enforce valid prefix completions within length limits for neural semantic parsing. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Focuses on formal grammar decoding constraints rather than model efficiency or distillation.

2,281–2,340 of 2,430← Previous Next →