Use this view to inspect the underlying evidence corpus. For ranked developments, decision posture and interpretation, use Signals.
Raw feed or Signals?
Raw feed is chronological evidence. Signals ranks and interprets material change.
Researchers propose SynH-Rank, a code search system that incorporates non-functional qualities like execution speed and memory usage into reranking.
Research addresses temporal leakage in legal precedent retrieval, evaluating citation context methods against a temporally-fenced benchmark.
Research paper proposes a continuous geometric framework, modeling Transformer architecture components as an integro-differential equation on a semantic fiber bundle.
SlotGuard is a proposed local transcript boundary solution to prevent LLM agents from leaking sensitive data like API keys or file paths into provider-bound transcripts.
New research proposes A-MESS, a defender-centric, Shapley-based evaluation for jailbreak attacks to improve LLM safety alignment through red-teaming.
New research introduces DRNOISE, a benchmark for evaluating deep research agents' ability to maintain sound evidential standards in environments with misleading documents.
Research introduces WAR, a workload-aware rollout system, to accelerate synchronous agentic reinforcement learning by optimizing decoding and scheduling.
Research demonstrates that removing refusal behaviors from LLMs via 'abliteration' causes off-target effects on unrelated decision-making tasks.
Research proposes a semantic map to bridge language model token probabilities to interpretable, declared state uncertainty for professional decisions.
Research explores hyperbolic geometry in AI, finding placement at the loss layer, not trainable adapters, prevents training collapse (HySAT).
Research explores "Thinking in Video," a paradigm where video generative models simulate and reason about real-world dynamics, but its promise remains unverified.
New research identifies 'salience induction' as a third attack vector for Multi-Hop RAG agents, exploiting fact position and framing.
D-NOVA, a research paper, proposes an in-storage accelerator for Retrieval-Augmented Generation (RAG) to reduce latency and energy overhead in vector retrieval.
Research explores using progressive disclosure via 'Agent Skills' for long-document Q&A, allowing agents to dynamically read relevant document parts.
PoLoRA introduces a preconditioned, orthogonalized optimizer for LoRA fine-tuning, aiming to improve efficiency and performance over standard Adam.
Research explores rhetorical patterns beyond directive verdicts for AI-assisted information evaluation, suggesting debate-style interactions improve critical assessment.
Mobius Learning introduces cyclic depth folding in Transformer architectures, enabling different data streams to use cyclically shifted block orders.
AlphaOracle uses a multi-stage deep learning framework to decipher 3,000 previously undeciphered oracle bone script characters by replicating expert workflows.
Research evaluates five small LLMs for generating biomedical research topic ontologies, assessing their capacity for domain-specific classification.
MADA-RL is a post-training framework optimizing compact models (<=4B parameters) for reasoning by specializing them into generator and critic roles.
Research proposes L1 augmented attention, a modification to scaled dot product attention that incorporates L1 distance to improve vector similarity metrics.
FinSAgent, a multi-agent RAG framework, improves evidence-grounded Q&A on SEC filings by addressing prior-corpus misalignment in retrieval.
Researchers propose Experiential Learning (EL) for LLMs to generate and distil rich textual feedback as 'coach' knowledge for open-ended tasks.
O-VAD, an object-centric tracking and reasoning system, improves industrial video anomaly detection where general VLM-based methods struggle.
Research explores LLMs' capability for structure-based drug design, benchmarking performance under spatial constraints for molecule generation.
Research tests how logical judgments in LLMs respond to learned contextual soft prefixes, diagnosing syllogistic stability.
Octopus introduces fine-tuning for 2B, 3B, and 7B parameter on-device LLMs to enhance software API function calling using a new dataset.
Octopus v3 introduces a sub-billion parameter multimodal AI agent designed for on-device inference, processing language, visual, and audio inputs.
Octopus v4 research introduces a 'Graph of Language Models' to orchestrate multiple niche LLMs, aiming for more efficient and capable systems.
Researchers propose Octo-planner, an efficient on-device Planner-Action framework that separates planning and action execution for AI agents.
© 2026 OneBench: AI Insights. All rights reserved.
Evidence before opinion