Experiments on seven heterogeneous handwriting benchmarks show that complementary supervision consistently improves training with page transcription alone, while joint multi-source training provides further gains on most datasets, and the proposed sparse expert model further improves the dense baseline on six of the se...
Nam Hoai Dang, H. Nguyen, Quang Huu Hieu et al.· 0 citations
It is argued that the right resolution is state-dependent, and GACA, a critic-free estimator whose granularity follows an uncertainty-based criticality proxy is proposed, improves task success over GRPO and GiGPO at both 1.5B and 7B scales.
Tao Liang, Yang Liu, Shang Luo et al.· 3 citations
This paper introduces Tasks over Application Manuals (TAM), a benchmark for evaluating long-horizon procedural reasoning and shows that current benchmarks may overestimate LLM reasoning ability and miss a key challenge: reliably following long, rule-based procedures.
Utkarsh Soni, S. Murtaza, Yi-Fan Nie et al.· 0 citations
DualMLC is proposed, a dual-branch framework that processes the same document through an autoregressive decoder-only language model and a bidirectional encoder, achieving state-of-the-art results on three widely used large-scale multi-label text classification benchmarks.
Hui Ye, Jing Zhang, Xiu-Long Yang et al.· 0 citations
Reach audiences
Advertise in front of researchers, engineers, and readers.
This work introduces context segmentation, a two-level agentic framework that divides complex exploitation tasks into manageable, contextually isolated sub-problems and demonstrates that for the E4B model, the strategy acts as an intelligent search, achieving competitive rewards with superior token efficiency compared...
We present ORQA, a method for testing occupation-level knowledge in large language models. Prior methods either map abstract LLM skills to occupations via task definitions or utilize expert knowledge which is difficult to obtain at scale and expensive. ORQA complements both of these methods by connecting O*NET occupati...
Shreyas R. Krishnan, Serina Chang, Abhishek Nagaraj· 0 citations
Small models are made more capable through distillation from a larger one that shares their tokenization scheme. However, do distilled byte and token models behave similarly in terms of scaling trends as compute and data increases? To enable this comparison, we introduce two variants to efficiently convert token logits...
Kalyani Marathe, Artidoro Pagnoni, Tomasz Limisiewicz et al.· 0 citations
This paper presents an open and modular transformer-based framework that detects and structures social tipping point evidence at the passage level, outperforming both a climate-focused model and untuned language models.
Kavindu Perera, Mohammad Abaeiani, Ekaterina Gilman et al.· 0 citations
The results indicate that front-loading code understanding into a reusable, codebase-specialised vector store enables small local models to deliver grounded and repository-specific answers, making the agent well suited as a privacy-preserving development tool for in-house scientific codebases.
Aaron Nobile, Andreas Adelmann, Mohsen Sadr· 0 citations
Assistants built on large language models are expected to write as their user would, and the dominant approach is single-channel: preferences summarised from conversation history and reinserted into context. This inverts the order of inference. Preferences are the task-dependent surface of a comparatively stable person...
B. Sankar, S. Deepthika, Pawni Yadav et al.· 0 citations
Blinded physician evaluation has been considered by many to be the gold standard for assessing clinical reasoning in large language models (LLMs). This is difficult to scale; thus, prior studies typically rely on small physician panels, often from a single institution or specialty, which both limits the scientific ques...
Thomas A. Buckley, Zahir Kanjee, Peter G. Brodeur et al.· 0 citations
The Graph Theory Agent (GTA), which pairs a preference-trained representation selector with plan-and-decompose scaffolding around a frozen executor LLM, is proposed, which lifts Phi-4 from 53.5% to 69.1% on the benchmark's easy split and from 33.0% to 41.5% on its hard split.
Zi-Xiang Xu, Yan-Bo Wang, Chenxi Wang et al.· 4 citations· ⚡1
