Vision-Language Models (VLMs) are widely deployed, yet the internal mechanisms behind their safety behavior remain poorly understood. We present a lightweight inference-time framework for localizing and manipulating safety-sensitive representations in frozen VLMs, without retraining or modifying parameters. On Qwen3-VL...
Muhammad Zeeshan Karamat, Syed Talal Hassan, Christiana Chamon Garcia· 0 citations
Proactive streaming video understanding requires assistants to track unfolding events and respond when a user's requested condition is met. To make these decisions, assistants must accumulate relevant evidence and reassess response conditions as new observations arrive. Existing work offers complementary capabilities f...
Xueyun Tian, Bingbing Xu, Jiaming Liu et al.· 0 citations
We report a systematic dissociation in modern text embedders: sentiment is almost perfectly decodable from their embeddings, yet almost entirely absent from their metric. Across eleven frozen encoders spanning 2021 to 2026, including 2026-generation models and the 4B-parameter LLM embedder Qwen3-Embedding-4B, a linear...
Large language models (LLMs) often exhibit sycophancy, altering their answers to align with user suggestions even on objective factual and reasoning tasks. Yet the quantitative form of this influence--- and whether it is calibrated to source reliability--- remains poorly understood. In this work, we investigate this be...
An-Xin Guo, Chirag Pabbaraju, Sara Ahmadian et al.· 0 citations
Reach audiences
Advertise in front of researchers, engineers, and readers.
Safety guards check content before an LLM application acts on it, and most paste that content into a plain-text prompt. Content that imitates the prompt can append a second task, and the guard answers that one instead. We call this target rebinding. Our explanation is that the guard grades the last complete task in its...
Conversation success is important for communication, social bonds, and well-being and is often reduced in autistic-neurotypical conversations. Previous studies have predicted perceived conversation success using acoustic and lexical features in pediatric autism-enriched samples, or using nonverbal behavior in non-autis...
Thanushi Withanage, Desi Jones, Noah Sasson et al.· 0 citations
Learned representations often encode sensitive attributes, such as gender or race, that may be undesirable or unsafe to retain. Concept erasure aims to remove this information while preserving task-relevant content. Existing methods remain incomplete: linear methods provide guarantees only against linear probes, while...
Idan Simai, Shauli Ravfogel, Uri Shaham· 0 citations
Large language models serve applications with diverse inference budgets, but training and maintaining separate models for each budget incurs substantial costs. Recent Looped Transformers enable flexible inference computation through parameter reuse, but maintaining strong performance across different inference budgets...
Yudong Zhang, Wei Ding, Kaili Zheng et al.· 0 citations
Several formal theories have been developed to characterize the capabilities and limitations of the transformer architecture, each employing different abstractions that capture different features of the architecture. Due to the differences in their underlying assumptions and conceptual frameworks, these theories can di...
Looped language models apply one block of layers repeatedly, and the design pays off only if depth is reusable: loop steps beyond the trained budget must remain usable at inference. We ask whether that rests on normalising the state between loop steps, as the released Ouro models do, or on agreement between the loop a...
Ivan Viakhirev, Kirill Borodin, Maxim Abramov et al.· 0 citations
Synthetic rewrites, where a language model rewrites a natural document e.g. into a target style, are now a standard ingredient of pretraining corpora. However, critical aspects of their use remain poorly understood, such as how much compute should be spent rewriting training data and whether a model trained on rewrites...
Nicholas Roberts, Ahmed Ahmed, Suhas Kotha et al.· 0 citations
This article provides an introduction to the mathematical foundations and algorithmic frameworks used to align Large Language Models (LLMs) with human intentions, preferences, and values. We discuss standard alignment techniques, such as fine-tuning (SFT), reinforcement learning with human feedback (RLHF), and direct p...
Siliang Zeng, Luca Viano, Chenliang Li et al.· 0 citations
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.
A new computational approach reveals how cells communicate to assemble into tissues. This could accelerate stem cell engineering, improve disease models, and inform therapies.
Microsoft Research Blog· microsoft.comOct 15, 2026
Zero-knowledge proofs can verify computing results without rerunning workloads or exposing sensitive data. Magnetar helps bring them into the mainstream with support for standard programming languages, major speed gains, and quantum-resistant security. The post Proof-based general-purpose computing with Magnetar appeared first on Microsoft Research.
We use cookies to run the site and, with your consent, for analytics and to show ads.
See our Cookie Policy.
