Modern surgical AI increasingly requires multimodal understanding rather than isolated perception, yet existing datasets remain largely limited to single-task supervision such as recognition or templated question answering. Current medical VQA resources often rely on synthetic templates, lack natural descriptive superv...
M. Hamdy, I. Zouari, Fatmaelzahraa Ahmed et al.· 0 citations
Supervised Fine-Tuning (SFT) of LLMs suffers from catastrophic forgetting, and prevailing mitigation strategies struggle to reconcile the plasticity-stability dilemma. To address this, we propose Entropic Control for Homeostatic Optimization (ECHO), a closed-loop control framework. By decomposing the KL divergence, we...
Shangxi Wu, Xiaorong Yan, Xiaojun Meng et al.· 0 citations
Latent reasoning performs intermediate computation in continuous space rather than through explicit tokens, which is particularly appealing in chemistry, where errors in explicit reasoning can propagate to the final answer. Yet if latent states truly constitute an implicit reasoning chain, they should be causally impor...
Tool-using language model agents are usually evaluated by whether they finish the user's task. That metric misses a common failure: an agent can complete the requested work while also taking an unauthorized side-effecting action, such as writing to the wrong object, expanding the scope of the request, or contacting an...
Jia-Ju Huang, Shao-Bin Chen, Xing-Long Liang et al.· 0 citations
Reach audiences
Advertise in front of researchers, engineers, and readers.
While Multimodal Large Language Models (MLLMs) excel at zero-shot reasoning, their spatial capabilities are primarily evaluated through static, passive observation. In complex, human-centric scenes, occlusion often hides crucial geometric evidence, and passive benchmarks force models to guess rather than recognize thei...
Tianchen Guo, Tengyue Zhang, Chen Liu et al.· 0 citations
Reinforcement learning with verifiable rewards (RLVR) enables large language models (LLMs) to continuously self-improve their reasoning capabilities. Compared to RLVR's scalar rewards, recent self-supervised on-policy distillation methods leverage denser supervision signals. However, both paradigms suffer from explorat...
Wenhang Shi, Yiren Chen, Zhe Zhao et al.· 0 citations
Large language models increasingly collaborate with humans and peer AI agents whose contributions may be incomplete, inconsistent, or incorrect. Effective collaboration requires calibrated trust, whereby decisions to rely on, verify, or assist a partner adapt rationally as empirical evidence accumulates. We investigate...
Coding agents are given embedding search and code-graph tools on the premise that a repository's structure is something the model lacks and the tool supplies. We test that premise on SWE-bench Verified with small open-weight models, stratifying tasks by a deterministic hop depth (the shortest code-graph path from code...
Saurav Bellam (The University of Texas at Austin)· 0 citations
Continual learning is the ability of LLM agents to improve over time by acquiring, refining, and applying knowledge from experience while retaining useful prior knowledge. It is essential for agents learning over long time horizons, yet current evaluations largely assess individual task performance rather than improvem...
Haoyu Zhao, Sitian Tu, Yuxuan Huang et al.· 0 citations
Pipelines that require structured outputs from hosted language models often specify a Pydantic schema in the prompt, validate the response, and retry when validation fails. Although native structured-output modes provide an alternative, application-side validation remains important for enforcing custom constraints. Yet...
This work proposes a family of counterfactual test-time augmentation (CTTA) methods, and proves that the optimal shortcut-invariant predictor is a member of this CTTA family, and describes its weights.
Katie Matton, R. Picard, John V. Guttag et al.· 0 citations
Hybrid post-training often mixes supervised fine-tuning (SFT) and reinforcement learning (RL), using training-side signals to set each sample's balance. Such signals characterize the anchor data and policy, but not whether the emphasized behavior helps in the target task and evaluation setting. This mismatch can reacti...
Leiyue Chen, Meng Wei, Xuefeng Hu 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.
