4 papers
No Free Lunch: Non-Asymptotic Analysis of Prediction-Powered Inference
Pranav Mani, Peng Xu, Zachary C. Lipton +1
Prediction-Powered Inference (PPI) is a popular strategy for combining gold-standard and possibly noisy pseudo-labels to perform statistical estimation. Prior work has shown an asy…
Valid Inference with Imperfect Synthetic Data
Yewon Byun, Shantanu Gupta, Zachary C. Lipton +2
Predictions and generations from large language models are increasingly being explored as an aid in limited data regimes, such as in computational social science and human subjects…
Expert Routing with Synthetic Data for Continual Learning
Yewon Byun, Sanket Vaibhav Mehta, Saurabh Garg +4
In many real-world settings, regulations and economic incentives permit the sharing of models but not data across institutional boundaries. In such scenarios, practitioners might h…
The Limited Impact of Medical Adaptation of Large Language and Vision-Language Models
Daniel P. Jeong, Pranav Mani, Saurabh Garg +2
Several recent works seek to adapt general-purpose large language models (LLMs) and vision-language models (VLMs) for medical applications through continued pretraining on publicly…