This study reveals how LLMs evolve their capabilities during training.
― 9 min read
Computer Science - Computation and Language
RSSResearchers propose a new method to classify fake news using deep learning.
― 5 min read
A new method to assess accuracy in language model outputs.
― 4 min read
Rasa dataset advances text-to-speech for Indian languages with neutral and expressive speech.
― 6 min read
DocBench benchmarks LLM-based systems for reading and responding to various document formats.
― 4 min read
A new method enhances the retrieval of scientific documents.
― 5 min read
New techniques make RAG systems faster and more accurate in answering questions.
― 5 min read
Researchers enhance formula writing in spreadsheets using language models for better user support.
― 5 min read
This article discusses using Reinforcement Learning to reduce bias in classification tasks.
― 6 min read
A method that translates plain language into executable programs for non-programmers.
― 6 min read
Using language models to improve tasks in underrepresented languages through data generation.
― 6 min read
SongCi enhances forensic examinations through visual and textual analysis.
― 6 min read
An overview of NLG progress, challenges, and future research directions.
― 6 min read
A framework to assess LLMs' abilities in data-related tasks with code interpreters.
― 5 min read
This paper studies how training influences the predictions of large language models.
― 6 min read
A look into how CLIP processes negation in language.
― 6 min read
Evaluating the impact of decoding strategies on LLM performance.
― 5 min read
Research shows how MBR decoding enhances translation quality in smaller models.
― 5 min read
A new model enhances inductive reasoning in knowledge graphs by using multiple starting entities.
― 5 min read
New methods enhance cache management for large language models.
― 5 min read
Research reveals varying effects of whitening techniques on sentence embeddings in NLP tasks.
― 5 min read
A new framework for verifying authorship with clear explanations.
― 7 min read
BinaryAlign improves word alignment accuracy for both high and low-resource languages.
― 5 min read
A method using text and images to better detect fake news.
― 6 min read
A new framework enhances fake news detection using text and images.
― 4 min read
This article explores how LLMs express values similar to humans.
― 7 min read
Examining the dynamics of AI-generated content on Civitai.
― 5 min read
A new method allows language models to generate their own training data for better performance.
― 5 min read
A new method enhances systems for answering image-related questions using external knowledge.
― 7 min read
Exploring how in-context probing and influence functions enhance data selection for models.
― 6 min read
Neural models help restore damaged Coptic manuscripts with missing text.
― 6 min read
An overview of practices undermining trust in machine learning model assessments.
― 6 min read
This study compares models for predicting emotions in Polish political texts.
― 5 min read
Exploring new methods for accurate medical translation.
― 5 min read
Dartboard improves retrieval in language models by maximizing relevant information.
― 6 min read
New models enhance the identification of speakers in dialogue content.
― 6 min read
GoldFinch offers efficient memory and processing for long text tasks.
― 5 min read
New attention methods improve transformer models in efficiency and performance.
― 5 min read
Elliptical Attention improves focus and performance in AI tasks.
― 5 min read
HaluQuestQA helps identify and correct errors in long-form question responses.
― 4 min read