Category
9,568 papers
VGBench is introduced, a 1,018-item diagnostic benchmark for action-level addressedness across side-talk, self-talk, and speaker-switch scenarios that measures multi-cue acoustic-context gating rather than isolated speaker identity.
Activation Flow (ActFlow), which manufactures activations from correct labels without fine-tuning, scores higher than the smallest-norm rule in 16 of the 18 combinations of locked model and $k, and its steering direction is nearly orthogonal to the honest difference-in-means direction.
Intent drift is established as a measurable multi-turn failure mode and explicit state maintenance as a partial mitigation and IntentFlux is introduced, an executable benchmark that converts verifiable tasks into dialogues with controlled intent changes while preserving their original graders.
Experiments on the TRACE benchmark across three Qwen3 model scales show that LAPS consistently outperforms distillation-based, prompt-based, and joint prompt--parameter adaptation baselines, improving average performance while reducing forgetting.
Reach audiences
Advertise in front of researchers, engineers, and readers.
The results show that the proposed task remains challenging even for hosted proprietary LLMs, and that their contribution to theorem proving is moderate: generated relations are often only partially sound and may be tailored to the specific NLI problem rather than representing generally valid lexical knowledge.
The rapid advancement of Artificial Intelligence (AI) and Natural Language Processing (NLP) has revolutionized the way humans interact with machines, but these advances have disproportionately benefited a small fraction of the world's languages.
NarraWorld is presented, a structured memory system for long-form writing that treats memory construction as world creation and achieves the strongest aggregate results across three writing benchmarks.
This work proposes ER-Audit, a two-stage black-box auditing framework that compares frozen verifier checkpoints before and after adaptation, and introduces two evaluation benchmarks pairing monitored and hidden task prompts grounded in shared contexts.
This work introduces compositional controllability to address compositional controllability in pipeline performance, and bounds the gap between observed and controlled score differences using only nuisance outside the window.
TRAP, a one-sided penalty on tokenwise TRA that acts only where the target model pulls ahead of its reference, brings memorization near the level of an untrained model at little utility cost, where generic regularizers barely move and differential privacy gives up most of what fine-tuning bought.
This framework is adopted to systematically analyse assumptions in recent research on culture in NLP along the full pipeline of task selection, data collection, system design and evaluation.
This work reviews 28 papers posted between 19 and 24 September and relates their findings to earlier work on label-probability classification, constrained decoding, reranking, calibration, and model cascades and uses recurring weaknesses in these studies to derive a 14-item evaluation checklist for future TDM work.
Natural Language Processing
Our most capable vision-language-action model enhances robot dexterity, reliability, and safety
A Blog post by Ai2 on Hugging Face
A Blog post by IBM Research on Hugging Face
Flint-interactive gives AI agents a language for creating interactive visualizations. Click a chart, filter data, highlight related content, and explore beyond the view the author originally chose. The post Flint-interactive: A language for creating interactive data visualizations appeared first on Microsoft Research.
We’re on a journey to advance and democratize artificial intelligence through open source and open science.
We use cookies to run the site and, with your consent, for analytics and to show ads. See our Cookie Policy.
