2 papers
cs.LG2025
Discovering Global False Negatives On the Fly for Self-supervised Contrastive Learning
Vicente Balmaseda, Bokun Wang, Ching-Long Lin +1
In self-supervised contrastive learning, negative pairs are typically constructed using an anchor image and a sample drawn from the entire dataset, excluding the anchor. However, t…
cs.CL2025
Discriminative Finetuning of Generative Large Language Models without Reward Models and Human Preference Data
Siqi Guo, Ilgee Hong, Vicente Balmaseda +6
Supervised fine-tuning (SFT) has become a crucial step for aligning pretrained large language models (LLMs) using supervised datasets of input-output pairs. However, despite being…