LMSpell: Neural Spell Checking for Low-Resource Languages

JITServe has been introduced as the first SLO-aware serving system for Large Language Models (LLMs), addressing the challenges posed by diverse workloads and unpredictable request information. This system aims to optimize service goodput by effectively scheduling requests to meet specific service-level objectives (SLOs) across various applications, including chatbots and multi-agent systems.

RoboNeuron has been introduced as a modular framework designed to enhance the adaptability and efficiency of embodied AI systems by integrating Large Language Models (LLMs) and Vision-Language-Action (VLA) models with the Robot Operating System (ROS). This framework aims to address existing challenges such as rigid inter-module coupling and fragmented inference acceleration, thereby improving real-time execution capabilities.

A novel approach to Named Entity Recognition (NER) for Portuguese has been introduced, utilizing a three-step ensemble pipeline of locally run Large Language Models (LLMs). This method demonstrates superior performance over individual models across multiple datasets, particularly in zero-shot scenarios, where minimal annotated data is available.

This chapter discusses the use of Large Language Models (LLMs) in the legal field, highlighting their ability to enhance traditional legal tasks such as interpreting statutes, contracts, and case law. It also addresses the challenges posed by these technologies, including algorithmic monoculture and compliance with regulations like the EU's AI Act and U.S. initiatives.

The newly introduced framework, Text2Graph, integrates lightweight large language models (LLMs) with graph neural networks (GNNs) to enhance text classification, particularly in scenarios with limited labels. This open-source Python package allows for flexible component swapping, including feature extractors and sampling strategies, and has been benchmarked across five datasets for zero-shot classification tasks.

SparseSwaps introduces a scalable method for refining pruning masks in large language models (LLMs), addressing the computational challenges associated with traditional pruning techniques that often lead to performance degradation. This approach enhances the efficiency of LLMs by optimizing the selection of pruning masks without the need for full retraining, which is typically resource-intensive.

An open, machine-readable dataset of Greek government decisions has been introduced, sourced from the national transparency platform Diavgeia, comprising 1 million decisions with high-quality raw text extracted from PDFs. This dataset is released with a reproducible extraction pipeline and includes qualitative analyses to explore boilerplate patterns and a retrieval-augmented generation (RAG) task to evaluate information access and reasoning over governmental documents.

A recent study highlights the importance of safety alignment in large language models (LLMs) as they are increasingly adapted for various tasks. The research identifies safety degradation during fine-tuning, attributing it to catastrophic forgetting, and proposes continual learning (CL) strategies to preserve safety. The evaluation of these strategies shows that they can effectively reduce attack success rates compared to standard fine-tuning methods.