This work finds that utilizing chat templates renders the model more compliant with harmful queries compared to a non-chat template, and shows that using a non-chat template during distillation better preserves the base student's internal representations, while chat template distillation induces a larger representation...
Anjila Budathoki, Manish Dhakal, Benjamin M. Ampel et al.· 0 citations
Tandem speech-to-speech architectures couple a responsive speech frontend with an asynchronous text backend. In KAME, a large language model (LLM) serves as the backend, supplying candidate responses as guidance to the speech frontend while the user is still speaking. Ordinary conversation recordings capture the eventu...
Manato Yaguchi, Yotaro Kubo, Hikaru Asano et al.· 0 citations
SEA-CLIP-Tiny is introduced, a compact multilingual text-vision embedding model for Southeast Asia with fewer than 50M parameters that adapts a CLIP-KD-style framework to Southeast Asian multilingual settings through regional data curation and multilingual teacher guidance.
Puja Ahmad Habibi, Faiz Assabil Firdaus, S. Ashvanth et al.· 0 citations
It is found that nearly identical QA accuracy can mask substantial differences in completion, answer validity, and generation workload, while proactive performance separates into response quality, response delay, false alarms, and missed target windows.
Yi-Bo Ma, Qian-Qian Zhang, Peng Liu et al.· 0 citations
Reach audiences
Advertise in front of researchers, engineers, and readers.
DCE+SRCL outperforms OPSD across multiple model scales and four competition-level mathematics benchmarks, and its comprehensive evaluations show that DCE+SRCL outperforms OPSD across multiple model scales and four competition-level mathematics benchmarks.
Audience sizing is a critical component of digital marketing. It enables precise resource allocation, campaign planning, and performance optimization. Traditional approaches using skeleton audiences, sampling, or predictive modeling suffer from significant delays, estimation errors, and poor scalability over high-dimen...
Hai-Xu Ma, Aditya Bansal, Shubham Lohiya et al.· 0 citations
It is found that training on code-switched data aligns the representations of parallel text, particularly across different scripts, and that this alignment persists through training on monolingual documents.
Dries Rooryck, Alex Cai, Yonatan Belinkov et al.· 0 citations
Analysis reveals that scaling model size, adding reasoning effort, expanding to authoritative web sources, and applying medical fine-tuning do not resolve these failure modes, demonstrating that they represent fundamental limitations of the retrieve-then-verify paradigm in open-ended medical settings rather than artifa...
He-Yuan Huang, Ji Dai, Alexandra DeLucia et al.· 0 citations
We introduce target-speaker unlearning ASR (TSU-ASR) task in a fully end-to-end framework for multi-speaker ASR and diarization. Given a multi-speaker utterance and a set of opt-out speakers who do not wish to have their speech transcribed, the task requires an ASR system to transcribe all speakers except the opt-out o...
Large Language Models (LLMs) have shown strong performance in low-resource offline translation; however, extending them to simultaneous speech-to-speech translation (Simul-S2ST) remains challenging due to the scarcity of causally aligned training data with high cross-lingual speaker fidelity. In addition, existing appr...
A. Hussein, Enas Albasiri, Travis M. Bartley et al.· 0 citations
An evaluation framework that accounts for class imbalance is proposed, i.e., the natural prevalence of excluded articles relative to included articles in SRs, and PromptSR, a tool designed to support prompt experimentation, experiment management, and result analysis for LLM-based screening are introduced.
G.Aravind Kumar, Luciano Marchezan, G. Genois et al.· 0 citations
HiCoMER is proposed, a framework for hierarchical collaborative memory management and validity-aware retrieval in LLM agents that consistently outperforms strong baselines by reducing outdated retrieval, preserving current team consensus, and improving downstream QA quality.
Yu-Fei Shi, Ru-Jing Yao, Ang 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.
