Probability-floor samplers substantially narrow the blind spot, demonstrating that the choice of truncation criterion matters for retaining human-used tokens and motivate decoding methods that preserve contextually appropriate low-probability choices while maintaining generation quality.
E. Arias, Nurzhan Sapargali, Christian Heumann et al.· arXiv.org· 2 citations
VMMU, a Vietnamese Multitask Multimodal Understanding and Reasoning Benchmark, is introduced to evaluate how vision-language models (VLMs) interpret and reason over visual and textual information beyond English, showing that the primary source of failure is not OCR, but instead multimodal grounding and reasoning over t...
V. Dang, A. Vo, Emilio Villa-Cueva et al.· 0 citations
This paper applies various NLP techniques to demonstrate feature de-biasing, feature compression, and industry classification in the commercial insurance context to show that NLP is not merely a supplementary tool but a foundational element of modern, data-driven insurance analytics.
Pan-Yi Dong, Z. Quan· North American Actuarial Jou...· 2 citations
It is proved that coherence optimization is equivalent to description-length regularization, and that among all such regularization schemes, coherence regularization with a prior derived from a pretrained model optimizes a lower bound of worst-case accuracy for semi-supervised learning.
Tianyi Alex Qiu, A. Ismail, Zhonghao He et al.· Trans. Mach. Learn. Res.· 1 citation
Reach audiences
Advertise in front of researchers, engineers, and readers.
A ModernBERT Bi-Encoder model is developed that achieves 98.4% accuracy on the PAN21 authorship verification task and identifies loss function, batch size, training duration, pre-trained model, input context length, and random text span data augmentation as important factors of model performance.
This work proposes the Agent-Editing World Model (AEWM), which models how reasoning and actions shape future task progress rather than simulating tool responses, and combines AEWM with EditAct, which integrates these capabilities with real execution.
Shuang Sun, Guo-Xin Chen, Fan-Zhen Meng et al.· 0 citations
Elastic Weight Consolidation preserves general capabilities best in both stages; on the 8B Spanish model the average score on general benchmarks drops 1.7 points versus 11.0 for standard fine-tuning, yet its scores for formality and grammatical gender control remain close to standard fine-tuning.
Niklas Scholz, David Thulke, Abdallah Nasir et al.· 0 citations
Weight-space post-training quantization is formulated as pre-deployment configuration selection using priced layer-output error and a trace reduction then yields a calibration-time price table and a budgeted price-guided selector, making fixed-geometry bit allocation a special case rather than the organizing problem.
Jun-Bin Qiu, Jian Mu, Wei-Tong Zhang et al.· 0 citations
The key-value (KV) cache of autoregressive transformers can be viewed as a fourth-order tensor spanning attention heads, tokens, features, and grouped layers. We measure the singular-value spectra of all four mode unfoldings on Mistral-7B-v0.3 and LLaMA-2-13B and compare four standard tensor decompositions: Tucker, CP,...
It is demonstrated that model selection, retrieval behaviour, and hardware limits must be evaluated separately when deploying open-weight LLMs for Turkish domain documents.
Imtiaz Ul Hassan, Öykü Akbulut, Onur Kaya et al.· 0 citations
This work reformulates eviction as a deployment risk-control problem: a material degradation occurs when eviction lowers task utility by more than a deployment-specified tolerance relative to full-KV inference on the same request, and deployment risk is the population frequency of such events.
Beomgu Kang, Sojin Yun, Hojoon Kim et al.· 0 citations
This work presents an approach that ranks encoder layers by the leave-one-layer-out change in Word Error Rate (WER), and presents the pruned model, which is simply a more shallow encoder with fewer layers.
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.
MIT News · Artificial Intelligence· news.mit.eduSep 24, 2026
