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
Designing Protein Structures from Natural Language Introduces Prompteina, an all-atom generative model and dataset for generating de novo protein structures conditioned on natural language prompts. Preprint	Small and efficient models via arxiv · 12 Oct, 16:29	Not relevant Applies text conditioning to protein design models, not small language model efficiency.
DyssomniaNN: Multi-Scale Spiking Neural Modeling of Human Sleep Disorders and Perceived Restedness Develops a multi-scale biophysical spiking neural network model to simulate sleep-wake regulation and discrepancies between objective sleep and subjective restedness. Preprint	Small and efficient models via arxiv · 12 Oct, 16:29	Not relevant Biophysical spiking neural network model for sleep disorders, not efficient LLM serving.
Goal-Oriented Bundle Management and Flow Control for Deep-Space Communications via Reinforcement Learning Applies reinforcement learning to manage bundle scheduling and flow control at relay nodes in deep-space communication networks. Preprint	Small and efficient models via arxiv · 12 Oct, 16:29	Not relevant Addresses network bundle protocol scheduling, not small or efficient language models.
KL Before and After Mixing: A Round-Complexity Separation in Parallel Sampling Analyzes round complexity in parallel coordinate samplers, proving a separation between schedule-averaged and output-law Kullback-Leibler divergences. Preprint	Small and efficient models via arxiv · 12 Oct, 16:29	Not relevant Focuses on theoretical sampling complexity, not small language models or efficient deployment.
Constructive Bounds for the Topological Expressivity of Neural Networks Proves upper bounds on the topological expressivity of neural networks measured by the Betti numbers of their output spaces. Preprint	Small and efficient models via arxiv · 12 Oct, 16:29	Not relevant Pure theory of ReLU network expressivity; not about small model efficiency or distillation.
Open-Vocabulary Audio-Visual Event Localization via Complex-Valued Fusion The paper proposes a complex-valued neural network to fuse audio and visual similarity streams for open-vocabulary audio-visual event localization. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not focus on model compression, quantization, or inference acceleration.
Revisiting Identity and Spectra Dispersion in Media-Bridged Time Series Forecasting: Linking Multivariate Signals and Narrative Flows The paper proposes MIDAPN, a unified spatiotemporal network using identity-aware graphs and temporal learning across numerical and narrative-flow time series. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not address model compression, quantization, or efficient serving.
Perception Test 2026: Challenge Summary and Extension to City-scale Audio-Visual Reasoning The paper introduces benchmarks and summarizes winning solutions for city-scale audio-visual reasoning and spatial intelligence at the Perception Test 2026. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not investigate model efficiency, compression, or lightweight serving.
DVLA-RL++: Dual-Level Vision-Language Alignment with Reinforcement Learning Gating for Few-Shot Learning The paper presents DVLA-RL++, combining complementary semantic purification and reinforcement learning gating to reduce context contamination in few-shot learning. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Addresses few-shot learning prototypes, not model compression or efficient serving.
Just Weather Scoring: Efficient End-to-end Nowcasting with Distributional Diffusion Introduces an end-to-end diffusion model operating directly in radar space for few-step probabilistic precipitation nowcasting. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Focuses on domain-specific diffusion forecasting rather than general small/efficient LLMs.
Beyond Spatio-Temporal Priors: A Generalizable Approach for Dense Correspondence Matching Presents a dense correspondence framework combining generative and semantic representations to handle non-rigid image editing and generation. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not address model efficiency, distillation, or quantization for small models.
Fly-by-Code: Embodied Coding Agents for Aerial Manipulation with Active Visual and Physical Feedback Presents an embodied coding agent framework using active physical and visual probing to refine aerial manipulation policies. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not focus on model compression, quantization, or efficient model serving.
Bridging Human and Model Attention: Explainability Analysis of CNN, Mamba, and ViT Architectures with Gaze-Based Validation Evaluates explainability methods across CNN, ViT, and Mamba models using eye-tracking gaze data as ground truth. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Focuses on interpretability and attention maps, not model efficiency or compression.
StereoEngine: Scaling Stereo Bimanual Robot Data without Stereo Cameras Synthesizes stereo manipulation data from monocular robot demonstrations using estimated depth and video diffusion models. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Focuses on synthetic data generation for policy training, not model compression or efficient inference.
Kernelized Parameter Sensitivity Atlases for Amortized Conditional Interpretability Develops a kernelized parameter sensitivity atlas to amortize input-specific gradient attribution across vision, forecasting, and biology models. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Focuses on parameter sensitivity attribution, not model compression or efficient serving.
BLEND: Balancing Personalization vs. Generalization in Federated Vision-Language Models Introduces a federated vision-language model tuning framework combining global shared adapters with client-specific personalized projection adapters. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Proposes parameter-efficient adapters for federated learning, not core model compression, quantization, or efficient serving.
Golden Anchor for Video Generation Studies how text-to-image anchor properties correlate with downstream video generation quality in image-to-video diffusion pipelines. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Focuses on generative anchor optimization rather than small models, quantization, or inference acceleration.
The Predictability Bias of Joint-Embedding Predictive Learning in Medical Imaging Audits joint-embedding predictive learning in medical CT scans, showing a bias that degrades performance on unpredictable localized pathology. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Analyzes pretraining representation dynamics rather than model compression, quantization, or efficient serving.
Decoding Register Tokens Preserves Global Scene Details Investigates decoding register tokens alongside patch tokens in vision transformers to preserve global scene appearance details. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Focuses on token decoding in autoencoders, not model compression or efficiency.
Sonic-AMG: Streaming Human Motion Generation from Spatial Audio Introduces a streaming causal transformer framework for generating human motion continuously from spatial audio inputs. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Addresses streaming generative modeling rather than model compression or efficient serving.
RARA: Robust Policy Fine-Tuning through Retrieval-Anchored Representation Alignment for Robotic Manipulation Aligns robot policy representations with retrieved pretraining examples to maintain generalization during low-data fine-tuning. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Addresses out-of-distribution policy adaptation rather than model quantization, distillation, or efficient serving.
Shortcut Learning in Geospatial Foundation Model Embeddings for Cross-Disaster Damage Assessment Evaluates shortcut learning and out-of-distribution failure in geospatial foundation models on cross-disaster damage assessment tasks. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Studies shortcut learning in remote sensing, not small model optimization.
Spanning Diffusion and Flow-based Posterior Sampling via Exact Gaussian Conditioning Presents a unified plug-and-play posterior sampling framework combining diffusion and flow models via exact Gaussian conditioning. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Studies trajectory priors for diffusion models, not model compression or efficient serving.
Dream to Remember: Continual Steering of Generalist Robot Policies with World Models Uses a frozen world model to continually steer generalist robot policies onto new tasks without catastrophic forgetting. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Addresses policy steering via world models, not small model distillation or quantization.
Could LLM Watermark Detection be Public? Evaluates public LLM watermark detectors and introduces a split-key mechanism combining public detection with private verification. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Focuses on watermarking security against tampering, not model efficiency, pruning, or quantization.
Verification with Transfer: Exact Information Frontiers and Their Price in Calls Establishes information bounds and query complexity frontiers for verifying answers when transferring hints from related source tasks. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Analyzes abstract verification query complexity, not model compression or efficient LLM inference.
PVBRF-ID: Private, Verifiable and Byzantine-Robust Federated Intrusion Detection for Oilfield Enterprise Networks The paper presents a two-server federated learning protocol using secret sharing and public verification to defend against poisoning attacks in industrial networks. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Studies federated learning protocols for network security, not efficient or small language models.
Byzantine-Resilient Federated Large Language Model via Hypergraph Message Passing The paper introduces a hypergraph autoencoder framework to filter malicious model updates and preserve Byzantine resilience in federated large language models. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Addresses federated aggregation defense, not model compression, distillation, or efficient inference serving.
Space Over Statistic, Retraining Over Ranking: Auditing Training Labels Under Label-Only Backdoors The paper evaluates label-only backdoor defenses and demonstrates that inspecting representations in unpoisoned feature spaces improves poison detection accuracy. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Studies backdoor defenses, not small or efficient model compression or serving.
The Limits of Query Monitoring against Distributed Model Extraction The paper evaluates the theoretical and empirical limits of using query monitoring to detect distributed model extraction campaigns among ordinary users. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Investigates defending against unauthorized model extraction, not techniques for model efficiency or compression.
AI Models Can Provably Hide Arbitrary Capabilities The paper shows that backdoor mechanisms can cryptographically hide arbitrary neural network capabilities until activated by secret triggers. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Focuses on hiding capabilities cryptographically rather than model efficiency or compression.
Structured Sentiment Analysis Using Sequence Labeling as Dependency Graph Parsing The paper models structured sentiment analysis as dependency graph parsing implemented through sequence labeling across five languages. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Uses simpler sequence labeling architectures, but does not study model quantization, distillation, or serving.
From Sparse Representations to Behavioral Insights for Multimodal Depression Assessment The paper proposes a sparse factor framework to map multimodal behavioral data to interpretable concepts for depression assessment. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not address model compression, quantization, distillation, or efficient serving.
Detecting Spin in Clinical Trials with Large Language Models The paper develops an LLM-based pipeline to automatically detect outcome switching and presentation spin in clinical trial reports. Peer-reviewed	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not study model compression, efficient inference, or edge deployment techniques.
Can Decision Models Understand Stance? Evaluating Jev Against General-Purpose LLMs The paper benchmarks a specialized decision model against general-purpose LLMs on English and Chinese stance detection tasks. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not investigate model pruning, quantization, efficient serving, or small models.
Not Every Change Is Necessary: Recoverable Drift in Large Language Model Unlearning The paper introduces a method to unlearn target knowledge while recovering collateral capability loss in non-target tasks. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Deals with machine unlearning rather than model efficiency, quantization, or distillation.
Event-Centric Memory with Query-Aware Graph Augmentation for Long-Term Conversational Agents The paper proposes an event-centric memory framework that converts dialogue history into query-relevant graph working memory for conversational agents. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Focuses on memory representations rather than model compression, distillation, or quantization.
When History Helps and Hurts: Selective History Use across Multimodal Turns The paper introduces a 7,000-task benchmark to evaluate how multimodal models selectively use or ignore conversational context. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not explore model compression, quantization, distillation, or efficient serving.
MiMo-V2.6: Scaling Reinforcement Learning Towards Self-Improvement The report presents an omni-modal model family trained by scaling reinforcement learning throughput, environments, and agentic reward compute. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Scales compute up to large context lengths rather than optimizing for small, efficient models.
Specialized Decision Models vs. General-Purpose LLMs: Benchmarking Jev Across Knowledge, Reasoning, and Multilingual Tasks The paper benchmarks a specialized multiple-choice decision model against 19 general-purpose LLMs across knowledge, reasoning, and multilingual tasks. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Studies decision architectures rather than model quantization, pruning, or efficient inference techniques.
Natural Language to First-Order Logic LLM-based Autoformalization The paper reviews LLM-based first-order logic autoformalization, distinguishing ontology extraction from logical translation and outlining key benchmarking challenges. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Surveys logic autoformalization rather than model compression or efficient serving.
ILM: An AI-Powered Storytelling Educational Tool The paper introduces an educational tool combining Arabic knowledge graphs and passage retrieval to generate narrative-based questions. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Presents an educational platform rather than model compression or efficient serving.
A persistent accuracy ceiling in automated verbal deception detection The paper meta-analyzes 25 years of automated verbal deception detection models and identifies a persistent accuracy ceiling around 70 to 75 percent. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Surveys deception classification models rather than model compression or inference efficiency.
Language-Specific Effects of Tokenizer Choice in Multilingual Language Models The paper trains 123 models across 54 tokenizers to show tokenizer choice matters more for languages with less training data. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Focuses on multilingual representation quality rather than model compression or efficient serving.
When KL Regularization Misfires in Group Policy Optimization Analyzes interaction failures between KL regularization and rewards in group policy optimization and proposes a calibrated surrogate objective. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Focuses on RL loss formulations rather than model compression or inference efficiency.
SteerablePlex: Can We Steer Full-Duplex Models? Introduces an instruction-following benchmark and training recipe to improve goal adherence in full-duplex conversational models. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Focuses on steerability and RL training rather than model compression or efficient serving.
SciTBERT: A family of chronologically consistent language models for scientific and technological language processing Releases BERT models pre-trained with yearly chronological cutoffs to reduce lookahead bias in scientific and patent text mining. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Focuses on chronological pretraining rather than model compression or inference efficiency.
Language Models as AI Research World Models Evaluates whether language models can act as world models to predict empirical experimental outcomes across AI research workflows. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Studies predictive models for experiment selection rather than model compression techniques.
NativeScope: Relation-Localized Retrieval over Native Topology with a Correct Anchor Proposes a scope-then-rank retrieval method using document topology and relational structure rather than pure semantic similarity. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Focuses on retrieval algorithms rather than model compression or efficient inference.
EgoVoice: Proactive Spoken Assistance from Egocentric Multimodal Streams Develops a framework for egocentric multimodal assistants to decide both when to speak and what guidance to provide. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Focuses on multimodal interactive assistance rather than model compression or efficient serving.
HarnessSQL: Harness-Native Training for SQL Agents in Realistic Database Environments Introduces a post-training framework that trains text-to-SQL agents directly inside interactive database execution environments. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Addresses agent training for SQL tasks rather than model compression or serving efficiency.
Long Text to Predictive Features: LLM-Guided Blockwise Feature Engineering via Executable Program Search Presents a framework that uses language models offline to generate executable feature-engineering code for tabular risk models. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Focuses on program synthesis for feature extraction rather than model compression or efficient serving.
SpaceCast-Bench: Evaluating Predictive Spatial Reasoning in Vision-Language Models Introduces a benchmark evaluating how well vision-language models predict unseen physical outcomes following scene interventions. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Focuses on multimodal benchmark evaluation rather than model compression, quantization, or efficient serving.
ViSkill: Reinforcing VLM Agents with Evolving Visual-Native Skills Proposes a framework that encodes visual trajectories into reusable skill cards to reinforce vision-language agent planning and rewards. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Focuses on reinforcement learning sample efficiency rather than model compression or inference latency.
WOVEN: Weaving Visual World Modeling into Multimodal LLMs Presents a benchmark and dataset using generative video rollouts to train and evaluate visual transition reasoning in multimodal models. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not investigate model pruning, quantization, or efficient deployment serving.
FastBench: Can Streaming VLMs Perceive High-Dynamic Real-World Streams? Introduces a benchmark evaluating streaming vision-language models on fast-moving, high-dynamic events in real-world video streams. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Investigates frame-rate trade-offs rather than model compression, quantization, or distillation.
SWE-Duel: Towards a Self-Scaling Arena for Coding Agents Presents an arena benchmark where pairs of coding agents dynamically generate, test, and debug code against each other. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not study model compression, distillation, or inference optimization.
CLUE: A Contrastive Learning Framework for Unsupervised Event Type Induction and Annotation Proposes an unsupervised framework using contrastive learning and clustering to induce, name, and annotate event types from text. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not explore model compression, quantization, or efficient inference serving.
Reflective Prompt Optimization of a Deep-Research Agent: A Topic-Disjoint, Multi-Seed Protocol for Measuring Transferable Gains and Prompt-Level Overfitting Studies reflective prompt optimization in research agents, demonstrating that gains on held-out tasks can stem from prompt-level memorization. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Explores prompt evolution rather than model distillation, pruning, or efficient inference.
Knowing Is Not Telling: Identity Ossification in LLM Representations of Trans-border Communities Proposes a diagnostic framework to study how language models compress fluid trans-border ethnic identities into rigid nation-state categories. Preprint	Small and efficient models via arxiv · 12 Oct, 16:28	Not relevant Does not address model compression, quantization, or efficient deployment.

2,221–2,280 of 2,430← Previous Next →