Socrates or Smartypants: Testing Logic Reasoning Capabilities of Large Language Models with Logic Programming-based Test Oracles

Reddit has initiated testing of a new verification system designed to differentiate notable figures from regular users, addressing the shortcomings of previous verification methods that were often exploited. This initiative aims to enhance user experience and foster community trust on the platform.

Reddit has announced that it will begin verifying public figures on its platform, a move aimed at enhancing the authenticity of accounts and reducing misinformation. This initiative is part of a broader effort to improve user trust and engagement within the community.

Researchers have developed a method to fine-tune language models, specifically Qwen2.5-3B-Instruct, to utilize Prolog for verifiable computation. This approach employs Group Relative Policy Optimization (GRPO) and has shown improved performance in reasoning tasks, achieving zero-shot MMLU results comparable to larger models.

A new method called TopiCLEAR has been introduced for topic extraction from social media posts, addressing challenges posed by the informal nature of platforms like X, Facebook, and Reddit. This method utilizes Sentence-BERT for embedding text and Gaussian Mixture Models for clustering, refining the clusters iteratively to improve topic modeling accuracy.

A new study introduces a Confidence-Aware Fine-Grained Debate (CFD) framework that utilizes multiple open-source large language models (LLMs) to enhance data enrichment for mental health and online safety. This framework simulates human annotators to reach consensus on labeling real-world indicators, addressing the challenges of dynamic life events. Two expert-annotated datasets were created, focusing on mental health discussions on Reddit and risks associated with sharenting on Facebook.