Baharan Mirzasoleiman

Understanding the Role of Training Data in Test-Time Scaling

Utilizing LLM Robustness for LVLM Safety via Reducing the Pretraining Modality Gap

Do We Need All the Synthetic Data? Targeted Image Augmentation via Diffusion Models

Data Selection for Fine-tuning Vision Language Models via Cross Modal Alignment Trajectories

DD-Ranking: Rethinking the Evaluation of Dataset Distillation

LoRA is All You Need for Safety Alignment of Reasoning LLMs

Dataset Distillation via Knowledge Distillation: Towards Efficient Self-Supervised Pre-training of Deep Networks

Mini-batch Coresets for Memory-efficient Language Model Training on Data Mixtures

Representations Shape Weak-to-Strong Generalization: Theoretical Insights and Empirical Predictions

Synthetic Text Generation for Training Large Language Models via Gradient Matching

Memory-Efficient Self-Supervised Contrastive Learning with a Supervised Loss