← Search

Preston Fu

2 accepted papers

2025

Compute-Optimal Scaling for Value-Based Deep RL

NeurIPS 2025poster

As models grow larger and training them becomes expensive, it becomes increasingly important to scale training recipes not just to larger models and more data, but to do so in a compute-optimal manner that extracts maximal performance per unit of compute. While such scaling has been well studied for…

Cited by 0SourcecodeScholar
2025

Value-Based Deep RL Scales Predictably

ICML 2025poster

Scaling data and compute is critical in modern machine learning. However, scaling also demands _predictability_: we want methods to not only perform well with more compute or data, but also have their performance be predictable from low compute or low data runs, without ever running the large-scale…

Cited by 0SourcePDFScholar