Νικόλαος Αλέτρας

Analyzing Political Parody in Social Media

2020

Parody is a figurative device used to imitate an entity for comedic or critical purposes and represents a widespread phenomenon in social media through many popular parody accounts. In this paper, we present the first …

Paragraph-level Rationale Extraction through Regularization: A case study on European Court of Human Rights Cases

2021 · White Rose Research Online (University of Leeds, The University of Sheffield, University of York)

Interpretability or explainability is an emerging research field in NLP. From a user-centric point of view, the goal is to build models that provide proper justification for their decisions, similar to those of humans, by …

Flexible Instance-Specific Rationalization of NLP Models

2022 · Proceedings of the AAAI Conference on Artificial Intelligence

Recent research on model interpretability in natural language processing extensively uses feature scoring methods for identifying which parts of the input are the most important for a model to make a prediction (i.e. explanation or …

Improving the Faithfulness of Attention-based Explanations with\n Task-specific Information for Text Classification

2021 · arXiv (Cornell University)

Neural network architectures in natural language processing often use\nattention mechanisms to produce probability distributions over input token\nrepresentations. Attention has empirically been demonstrated to improve\nperformance in various tasks, while its weights have been extensively used as\nexplanations …

HashFormers: Towards Vocabulary-independent Pre-trained Transformers

2022

Transformer-based pre-trained language models are vocabulary-dependent, mapping by default each token to its corresponding embedding. This one-to-one mapping results into embedding matrices that occupy a lot of memory (i.e. millions of parameters) and grow linearly …

Active Learning Principles for In-Context Learning with Large Language Models

2023

The remarkable advancements in large language models (LLMs) have significantly enhanced predictive performance in few-shot learning settings. By using only a small number of labeled examples, referred to as demonstrations, LLMs can effectively perform the …

Adapting Chat Language Models Using Only Target Unlabeled Language Data

2024 · arXiv (Cornell University)

Vocabulary expansion (VE) is the de-facto approach to language adaptation of large language models (LLMs) by adding new tokens and continuing pre-training on target data. While this is effective for base models trained on unlabeled …

Progressive Depth Up-scaling via Optimal Transport

2025 · arXiv (Cornell University)

Scaling Large Language Models (LLMs) yields performance gains but incurs substantial training costs. Depth up-scaling offers training efficiency by adding new layers to pre-trained models. However, most existing methods copy or average weights from base …