João A.
Leite
Senior Applied Scientist at Microsoft
I develop methods to evaluate and improve the safety and reliability of language models.
Selected publications
Tailored untruths: How personalisation challenges LLM safeguards
Weakly Supervised Veracity Classification with LLM-Predicted Credibility Signals
A Cross-Domain Study of the Use of Persuasion Techniques in Online Disinformation
EUvsDisinfo: A Dataset for Multilingual Detection of Pro-Kremlin Disinformation in News Articles
Toxic Language Detection in Social Media for Brazilian Portuguese
Show all publications
Build it, Break it, Repeat: Benchmarking and improving LLM-manipulated disinformation detection in social media posts
A Survey on Automatic Credibility Assessment Using Textual Credibility Signals in the Era of Large Language Models
Comparison Between Parameter-Efficient Techniques and Full Fine-Tuning
SheffieldVeraAI at SemEval-2023 Task 3
Noisy Self-Training with Data Augmentations for Offensive and Hate Speech Detection Tasks
NELL’s subcategories from a question answering environment
Writing
Parameter-Efficient Fine-Tuning with LoRA
Transformers
Updates
Joined Microsoft as a Senior Applied Scientist.
Graduated with a PhD in Computer Science from the University of Sheffield.
Led an in-house research hackathon with four invited researchers.
Earlier updates
Attended the Lisbon Machine Learning School, sponsored by a Google and Zendesk scholarship.
Became a Research Associate at the University of Sheffield.
Began my PhD in Computer Science at the University of Sheffield, sponsored by an EPSRC DTP Scholarship.
Joined PicPay as a Data Scientist.
My AACL paper received the Best Paper award at Universidade Federal de São Carlos.
Graduated with a bachelor's degree in Computer Science from Universidade Federal de São Carlos.
Joined Birdie.ai as a Data Scientist.