7 papers
Prediction-Only Distillation in Linear and Logistic Regression
Hien Dang, Pratik Patil, Alessandro Rinaldo
Self-distillation (SD) is typically studied when the student is retrained on the teacher's original training inputs. In many practical deployments, however, the labeled training da…
Optimal Self-Distillation for Rectified Flow via Linear Probing
Saptarshi Roy, Debepsita Mukherjee, Pratik Patil
Modern generative models are increasingly trained using model-generated signals, creating both opportunities for self-improvement and risks of collapse. We study optimal self-disti…
Evaluating Stochasticity in Deep Research Agents
Haotian Zhai, Elias Stengel-Eskin, Pratik Patil +1
Deep Research Agents (DRAs) are promising agentic systems that gather and synthesize information to support research across domains such as financial decision-making, medical analy…
Optimal Unconstrained Self-Distillation in Ridge Regression: Strict Improvements, Precise Asymptotics, and One-Shot Tuning
Hien Dang, Pratik Patil, Alessandro Rinaldo
Self-distillation (SD) is the process of retraining a student on a mixture of ground-truth labels and the teacher's own predictions using the same architecture and training data. A…
Precise Model Benchmarking with Only a Few Observations
Riccardo Fogliato, Pratik Patil, Nil-Jana Akpinar +1
How can we precisely estimate a large language model's (LLM) accuracy on questions belonging to a specific topic within a larger question-answering dataset? The standard direct est…
A Framework for Efficient Model Evaluation through Stratification, Sampling, and Estimation
Riccardo Fogliato, Pratik Patil, Mathew Monfort +1
Model performance evaluation is a critical and expensive task in machine learning and computer vision. Without clear guidelines, practitioners often estimate model accuracy using a…
Asymptotically free sketched ridge ensembles: Risks, cross-validation, and tuning
Pratik Patil, Daniel LeJeune
We employ random matrix theory to establish consistency of generalized cross validation (GCV) for estimating prediction risks of sketched ridge regression ensembles, enabling effic…