HF RL Explorer

Task: Implement an Optimal GPU Batch Size Calculator

Task: Implement an Optimal GPU Batch Size Calculator: a task in Terminal-Lego-15k (Harbor dataset). Create a Python utility that calculates the optimal (largest possible) batch size for a deep learning model that will fit into available GPU memory without causing an Out-Of-Memory (OOM) error.

The task

Create a Python utility that calculates the optimal (largest possible) batch size for a deep learning model that will fit into available GPU memory without causing an Out-Of-Memory (OOM) error.

Part of PrimeIntellect/Terminal-Lego-15k.