Yenting Lin

I am currently taking a break to fulfill my military duty in Taiwan. During this time, I have been building scalable and efficient environments for reinforcement-learning (RL) agents.

At Meta, I co-developed a scalable RL pipeline that improves long-form reasoning capabilities. I completed my Ph.D. in 2025 at National Taiwan University, advised by Professor Yun-Nung (Vivian) Chen.

Internship Experience

• Meta GenAI: Enhanced reasoning with stepwise feedback and long-form reasoning
• NVIDIA Research: Developed error correction methods for multimodal language models
• Amazon Alexa AI: Built factuality evaluation agent and synthetic data techniques

Selected Publications

• Step-KTO: Optimizing Mathematical Reasoning through Stepwise Binary Feedback (arXiv 2025)
• Measuring Taiwanese Mandarin Language Understanding (COLM 2024)
• LLM-Eval: Unified Multi-Dimensional Automatic Evaluation for Open-Domain Conversations with Large Language Models (NLP4ConvAI 2023)
• Selective In-Context Data Augmentation for Intent Detection using Pointwise V-Information (EACL 2023)
• Knowledge-Grounded Conversational Data Augmentation with Generative Conversational Networks (SIGDIAL 2022)
• SalesBot: Transitioning from Chit-Chat to Task-Oriented Dialogues (ACL 2022)

Featured Projects

• Taiwan-LLM: Traditional Mandarin LLMs for Taiwan
• Mistral-Small-Reasoning: Open-weight reasoning model

Fun Facts

• Proud owner of two cats
• Completed a full marathon
• Part-time (trance) DJ during college