End-to-end speech parsing, a task recently proposed, consists in predicting both the transcription and the syntactic tree for a spoken utterance. Existing architectures for speech parsing often utilise intermediate neural networks. In this work, we examine the effectiveness of intermediate neural networks (NN) for pars...
Minnie Kabra, Benjamin Lecouteux, Maximin Coavoux· 0 citations
Reinforcement learning with verifiable rewards (RLVR) and on-policy distillation (OPD) have become two main paradigms for post-training reasoning models. RLVR gives each response a single outcome label, leaving the steps inside it without separate credit. OPD provides token-level guidance at student-visited prefixes, b...
Large language models (LLMs) often default to Modern Standard Arabic (MSA) when generating Arabic, even when prompted with dialectal Arabic. A natural explanation is that their internal representations are dominated by MSA. We test this hypothesis by adapting the language-dominance framework of Shani and Basirat (2025)...
Abdu Sallouh, Nicholas Popovic, Michael Färber· 0 citations
Despite the adoption of large language models (LLMs) in recommendation systems, prevailing approaches mostly model single-type behaviors (e.g., views or purchases). Even when incorporating multiple behaviors, existing methods flatten heterogeneous actions into homogeneous token sequences, ignoring their distinct decisi...
Lei Liang, Guo-Shuai Wang, Xin He et al.· 0 citations
Reach audiences
Advertise in front of researchers, engineers, and readers.
This work introduces SAIL, an open model with 35B total and 3B active parameters for literature research, scientific coding, and multi-step research workflows that achieves competitive performance across scientific research tasks with substantially fewer parameters than leading open-weight models.
Sai-Shuai Sun, Bryan Dai, Che Liu et al.· 0 citations
An answer judge instructed to grade the final commitment should reject an explicitly wrong final value even when an earlier value matches the reference. We show that adding one colon to a candidate can violate this requirement depending on the presentation of the structured judging request. Numeric references certify t...
A framework for reference-free hallucination detection in Sinhala-to-English, with a semantic rescue mechanism that uses character-level similarity to distinguish hallucinations from morphological variants is introduced.
Contemporary decoder-only large language models (LLMs) have demonstrated strong capabilities across a wide range of domains. However, existing pretraining paradigms for gloss-free sign language translation (SLT) are largely designed around conventional encoder-decoder pretrained language models, which limits their dire...
Zhi Rao, Yucheng Zhou, Qian-Ran Sun et al.· 0 citations
Rule-governed contextual decision tasks require models to apply specified rules to case-specific context and evidence. Written rules can leave gaps in decision guidance and process evaluation, while reference judgments vary in their support from the rules and evidence. To address these challenges, we introduce AdaptEvo...
Shi-Jun Wan, Jian-Cong Xie, Han-Rui Xu et al.· 0 citations
Large language models (LLMs) increasingly act as agents with little human oversight, so potential mistakes they make can go unnoticed. Users then depend on the model to report what went wrong. An honest model discloses its mistakes, while a deceptive one conceals them. However, it is unclear how current LLMs behave in...
L. Florin, Amelie Knecht, Ulysse Schaller et al.· 0 citations
Structured pruning reduces the deployment cost of reasoning language models, but the resulting capability degradation can hinder subsequent on-policy distillation (OPD) recovery. Because OPD relies on student-generated trajectories, pruning damage that persists after offline distillation can limit its effectiveness. We...
Houcheng Jiang, Mao Zheng, Mingyang Song et al.· 0 citations
Large Language Models (LLMs) serving as long-term dialogue agents require memory systems that support reliable reasoning over extended interactions. However, existing Retrieval-Augmented Generation (RAG) frameworks typically treat memory as passive storage, making it difficult to distinguish source-attributed beliefs f...
Zi-Rui Liao, Zheng-Xian Wu, Zhuo Chen et al.· Proceedings of the 2026 Conf...· 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.
