We introduce BUST, a comprehensive benchmark designed to evaluate detectors of texts generated by instruction-tuned large language models (LLMs). Unlike previous benchmarks, our focus lies on evaluating the performance of detector systems, acknowledging the inevitable influence of the underlying tasks and different LLM generators. Our benchmark dataset consists of 25K texts from humans and 7 LLMs responding to instructions across 10 tasks from 3 diverse sources. Using the benchmark, we evaluated 5 detectors and found substantial performance variance across tasks. A meta-analysis of the dataset characteristics was conducted to guide the examination of detector performance. The dataset was analyzed using diverse metrics assessing linguistic features like fluency and coherence, readability scores, and writer attitudes, such as emotions, convincingness, and persuasiveness. Features impacting detector performance were investigated with surrogate models, revealing emotional content in texts enhanced some detectors, yet the most effective detector demonstrated consistent performance, irrespective of writer’s attitudes and text styles. Our approach focused on investigating relationships between the detectors’ performance and two key factors: text characteristics and LLM generators. We believe BUST will provide valuable insights into selecting detectors tailored to specific text styles and tasks and facilitate a more practical and in-depth investigation of detection systems for LLM-generated text.

This document describes our solution to the MEDIQA-M3G: Multilingual & Multimodal Medical Answer Generation. To build our solution, we leveraged two pre-trained models, a Visual Language Model (VLM) and a Large Language Model (LLM). We fine-tuned both models using the MEDIQA-M3G and MEDIQA-CORR training datasets, respectively. In the first stage, the VLM provides singular responses for each pair of image & text inputs in a case. In the second stage, the LLM consolidates the VLM responses using it as context among the original text input. By changing the original English case content field in the context component of the second stage to the one in Spanish, we adapt the pipeline to generate submissions in English and Spanish. We performed an ablation study to explore the impact of the different models’ capabilities, such as multimodality and reasoning, on the MEDIQA-M3G task. Our approach favored privacy and feasibility by adopting open-source and self-hosted small models and ranked 4th in English and 2nd in Spanish.

This paper describes our submissions to the Social Media Mining for Health Applications (SMM4H) shared task 2022. Our team (mattica) participated in detecting stances and premises in tweets about health mandates related to COVID-19 (Task 2). Our approach was based on using an in-domain Pretrained Language Model, which we fine-tuned by combining different strategies such as leveraging an additional stance detection dataset through two-stage fine-tuning, joint-learning Stance and Premise detection objectives; and ensembling the sentiment-polarity given by an off-the-shelf fine-tuned model.