9 papers
The Mirrored Influence Hypothesis: Efficient Data Influence Estimation by Harnessing Forward Passes
Myeongseob Ko, Feiyang Kang, Weiyan Shi +3
Large-scale black-box models have become ubiquitous across numerous applications. Understanding the influence of individual training data sources on predictions made by these model…
Can Generalist Agents Automate Data Curation?
Feiyang Kang, Hanze Li, Adam Nguyen +5
Curating training data is among the most consequential yet labor-intensive parts of modern AI development: practitioners iteratively propose, implement, evaluate, and revise data p…
Confidence-Aware Alignment Makes Reasoning LLMs More Reliable
Kejia Chen, Jiawen Zhang, Yihong Wu +5
Large reasoning models often reach correct answers through flawed intermediate steps, creating a gap between final accuracy and reasoning reliability. Existing alignment strategies…
Characterizing Model-Native Skills
Feiyang Kang, Mahavir Dabas, Myeongseob Ko +1
Skills are a natural unit for describing what a language model can do and how its behavior can be changed. However, existing characterizations rely on human-written taxonomies, tex…
Demystifying Synthetic Data in LLM Pre-training: A Systematic Study of Scaling Laws, Benefits, and Pitfalls
Feiyang Kang, Newsha Ardalani, Michael Kuchnik +7
Training data plays a crucial role in Large Language Models (LLM) scaling, yet high quality data is of limited supply. Synthetic data techniques offer a potential path toward sides…
Quagmires in SFT-RL Post-Training: When High SFT Scores Mislead and What to Use Instead
Feiyang Kang, Michael Kuchnik, Karthik Padthe +4
In post-training for reasoning Large Language Models (LLMs), the current state of practice trains LLMs in two independent stages: Supervised Fine-Tuning (SFT) and Reinforcement Lea…