Researcher, New York University
3 papers at NeurIPS 2025
We generalize CLIP training to worldwide web-scale, with +0.8% better than English only counterpart on zero-shot ImageNet classification (no compromise), SoTA on zero-shot multilingual: 57.4% on CVQA and 50.2% on Babel-ImageNet.
We introduce NaturalReasoning, a 2.8M-question dataset spanning diverse domains, enabling effective knowledge distillation and unsupervised self-training to enhance LLM reasoning capabilities.
