Experiment-as-Code Labs proposes a paradigm where scientific experiments are encoded as declarative configurations compilable to instrument APIs. AI agents formulate hypotheses, a systems layer performs program analysis and orchestration, then experiments execute via physical equipment control. General-purpose stack independent of science domain, lab type, or instrument.
MemOCR is a multimodal memory agent that compresses long interaction histories into structured images with adaptive information density. Trained via RL with budget-aware objectives, it outperforms text-based baselines on multi-hop and single-hop QA benchmarks under extreme context constraints.
Study on non-uniform replay effectiveness in off-policy RL. Authors identify three key factors: replay volume, recency of transitions, and entropy of sampling distribution. They propose Truncated Geometric replay, which biases toward recent experience while maintaining high entropy, improving sample efficiency in low-volume regimes.
PersonaDual is an LLM framework that adaptively switches between objective and personalized reasoning based on context. Trained via SFT then optimized with RL using DualGRPO, it preserves personalization benefits while maintaining factuality and reducing interference.
Researchers propose UALM (Unified Agent Lifecycle Management), a governance framework for agentic AI systems in healthcare. The model addresses agent sprawl through five control-plane layers: identity registry, orchestration, PHI-bounded context, runtime policy enforcement with kill-switch triggers, and lifecycle management linked to credential revocation and audit logging.
PriHA is a RAG-enhanced LLM system for primary healthcare assistance in Hong Kong. The framework combines a query optimizer and novel Dual Retrieval Augmented Generation (DRAG) architecture to retrieve fragmented clinical guidelines and generate accurate, localized responses, outperforming ChatGPT and DeepSeek on accuracy metrics.
OptBias, a meta-learning framework, addresses offline black-box optimization with scarce data. It generates synthetic tasks via Gaussian process to learn reusable optimization bias, then fine-tunes the surrogate model on small target datasets. Outperforms baselines on continuous and discrete optimization benchmarks.
R2M (Real-Time Aligned Reward Model) introduces a lightweight RLHF framework to mitigate reward overoptimization. Instead of relying solely on semantic representations, R2M leverages evolving hidden states from the policy model to align with real-time distribution shifts during reinforcement learning training.
Comparative study of protein representations (amino acid composition, k-mers, ProtBERT, physicochemical descriptors) for Parkinson's disease classification. ProtBERT + MLP achieves F1=0.704±0.028, ROC-AUC=0.748±0.047. Primary sequences alone show limited discriminative power; structural/functional features required.
Researchers show that combining acoustic and expectation-related ANN representations improves music identification from brain activity (EEG). Models pretrained on both representation types outperform baselines, with complementary gains exceeding seed ensembles. This approach enables general-purpose EEG models grounded in cortical encoding principles.
PCFJudge, an inference-time method, evaluates factuality by rerunning a listwise prompt across multiple candidate orderings and aggregating scores. On RewardBench 2 Factuality with K=7 permutations, top-1 accuracy improves from 86% to 91.33% (GPT-5.4) and 86.33% to 89.67% (Claude Sonnet 4.6).
Asynchronous RL pipelines for LLM agents lose historical old logits required for PPO off-policy correction, entangling discrepancy repair with staleness correction. The paper proposes three acquisition strategies (snapshot, dedicated model, interruption) and a revised PPO-EWMA method to preserve decoupled correction semantics.
Causal inference framework for financial stress testing in panel data with continuous treatment and no control group. Proposes closed-form confounding envelope parameterized by two sensitivity parameters, combines partial identification with importance-weighted conformal prediction. Shows standard predictive models remain causally biased on US unemployment data.
Complementary-label learning (CLL) methods remain limited to 10-class classification. This paper proposes BICL, a framework that deliberately uses biased (non-uniform) transition matrices to restrict complementary labels to class subsets. On CIFAR-100 and TinyImageNet-200, BICL achieves 7× accuracy improvements over traditional methods.
Study showing that two protocols for evaluating layer redundancy in transformers (replacement and interchange) yield divergent results for identifying layers to prune. On Pythia, Qwen3-8B, and Llama-3.1-8B, the protocol gap dramatically changes which layers appear safe to remove, even under the same KL evaluator.
Computational tool to classify manner and result verbs at scale. Uses linguistically informed prompts with LLMs to generate annotations over MASC and InterCorp data (436 VerbNet classes). RoBERTa-based classifier achieves 89.6% accuracy on three held-out gold-standard datasets. Applicable to developmental research on verb semantics.
Researchers propose LAD-inspired PPT, pre-pretraining on MP-STRUCT, a formal language encoding hierarchical composition and long-distance displacement. After 500 steps, this approach matches formal-language baselines in token efficiency while imparting LLMs human-like resistance to structurally implausible languages.
RTI-Bench is a structured dataset of 1,516 Indian Central Information Commission (CIC) decisions with outcome labels, exemption citations, and IRAC-style reasoning components. Mistral 7B achieves 57.3% accuracy on outcome prediction (baseline 14.3%). First publicly released structured dataset for Indian RTI administrative decision analysis.
Early Pruning accelerates RAPTOR routing algorithms for public transport by pre-sorting transfer connections and pruning longer transfers that cannot improve the best arrival time. Tested on Swiss and London transit networks, the technique reduces query times by up to 57% without compromising optimality.
ARROW extends DreamerV3 with a dual-buffer replay system (short and long-term) for continual reinforcement learning. Tested on Atari and Procgen, the algorithm reduces catastrophic forgetting while maintaining knowledge transfer, inspired by experience replay in the brain.
Explicit Logic Channel method runs parallel to MLLMs to validate and enhance multimodal models on zero-shot tasks. Combines LLM, VFM and probabilistic logical reasoning (factual, counterfactual, relational) with Consistency Rate metric for model selection without ground-truth labels. Evaluated on MC-VQA and HC-REC with 11 open-source MLLMs.
arXiv paper introducing spatiotemporal robustness (STR) for temporal logic specifications over discrete-time signals. Formalized as multi-objective reasoning problem via partial order, STR captures joint spatial and temporal perturbations. Applicable to multi-agent systems, smart cities, air traffic control. Monitoring algorithms proposed.
JSPG introduces a dynamic filtering framework for Chinese contextual ASR combining semantic, pinyin, and glyph features. The approach uses an extended Smith-Waterman algorithm to score N-best hypothesis sequences against keywords. Experiments on Aishell-1 and RWCS-NER datasets show significant improvements in keyword recognition accuracy.
Large Reasoning Models generate traces aligned with human reaction times, but this alignment persists regardless of inference-time reasoning budget. Study across GPT-OSS-20B and GPT-OSS-120B: token allocation tracks human difficulty patterns and remains invariant across effort levels, suggesting cognitive cost alignment is crystallized at training time.
SkillTTA synthesizes task-specific textual skills by retrieving relevant training trajectories, with no parameter updates to the solver model. Evaluated on SpreadsheetBench, ALFWorld, and BigCodeBench: SpreadsheetBench improves from 0.397 to 0.505 Pass@1, BigCodeBench from 0.517 to 0.651.
Self-evolution loops in LLMs plateau when they fail to generate learnable information. This study identifies three roles (Proposer, Solver, Verifier) and three system designs (asymmetric co-evolution, capacity growth, proactive information seeking) to sustain information gain across iterations on coding tasks.
PluRule is a multimodal multilingual benchmark for moderating pluralistic communities on social media. It covers 13,371 rule violations across 1,989 Reddit communities (9 languages, 2,885 rules). State-of-the-art vision-language models, including GPT-4.5 with advanced reasoning, only marginally outperform a trivial baseline.
HeteroFusion merges heterogeneous language models (Llama, Qwen, Mistral) by aligning functional module structures rather than raw weights, and suppressing incompatible transfer signals. Outperforms fusion, merging, and ensemble baselines on heterogeneous transfer, multi-source fusion, and cross-family generalization.
LEAF is a living benchmark to evaluate LLM forecasting capabilities using multidimensional events. The system uses recursive retrieval agents and dual-agent cross-validation to provide relevant auxiliary text. Testing shows LLMs leverage signals from complex events to improve predictions, particularly on equities.
AgentRevive introduces a Markov state-aware framework for resilient multi-agent LLM system evolution. Instead of aggressively pruning failing agents, the method uses soft state transitions (Active/Standby/Terminated) with a hallucination risk estimator. Results: outperforms baselines on general reasoning, domain-specific tasks, and hallucination challenges while reducing token consumption.
EmergentBridge improves unified multimodal embedding models for unpaired modality pairs (audio↔depth, infrared↔audio). The method learns a mapping producing a 'noisy bridge anchor' and enforces alignment in the orthogonal subspace, preserving existing anchor-alignment structure. Results across 9 datasets: outperforms baselines on zero-shot classification and retrieval.
AMATA is an adaptive multi-agent trajectory alignment framework for knowledge-intensive question answering. Six specialized agents collaboratively perform structured actions to improve factual consistency and reduce hallucinations. The system formalizes multi-agent collaboration as a trajectory preference alignment problem with intra-trajectory and inter-agent dependency learning.
MiniGPT is a compact from-scratch PyTorch implementation of GPT-style autoregressive language modeling in a single notebook. The 10.77M-parameter model achieves validation loss of 1.4780 on Tiny Shakespeare with character-level tokenization and generates text with recognizable Shakespeare-style dialogue structure.
SurgUn, an unlearning method for text-to-image diffusion models, treats forgetting as controlled competition rather than direct deletion. Using target-gradient ascent and descent over semantically diverse distractors, it reduces erase-retain imbalance and limits collateral damage. Tested on Stable Diffusion v1.5, SDXL, and SANA-1.5.
Comparative study of 10 annotation workflows for conversational speech summarization. Audio-based summaries are less informative than transcript-based ones, but iterative peer-editing with audio mitigates this gap. Validates this approach for creating benchmarks incorporating lexical and prosodic information.
Novel topic modeling approach (SBTA) assigns topics to text segments rather than entire documents, reducing topic contamination. Authors construct SemEval-STM, a dataset annotated via LLM + human refinement, and validate improved clustering quality and interpretability across multiple models.
Pilot benchmark for translating natural language to First-Order Logic (FOL) in planetary exploration. Dataset built from NASA mission documentation (2003-2013), manually annotated with FOL representations capturing temporal structure, agent roles, and operational dependencies. Structured predicate vocabularies provided.
UxSID introduces a framework for modeling ultra-long user sequences using Semantic IDs and dual-level attention. The architecture balances computational efficiency with semantic awareness, achieving state-of-the-art performance and 0.337% revenue lift in large-scale advertising A/B testing.
MarsTSC, an agentic framework for VLMs, addresses few-shot multimodal time series classification. Three collaborative roles (Generator, Reflector, Modifier) iteratively refine a knowledge bank via reflective reasoning. Tested on 12 benchmarks across 6 VLM backbones: consistent gains, interpretable rationales.
TruthMarketTwin, a multi-agent simulation framework, studies strategic behavior of LLM agents in e-commerce markets. Agents autonomously exploit weaknesses in reputation-based governance, while warrant enforcement reduces deception and reshapes strategic reasoning.