HF RL Explorer

Task: Diagnose and Fix Adam Optimizer Training Instability

Task: Diagnose and Fix Adam Optimizer Training Instability: a task in Terminal-Lego-15k (Harbor dataset). You are given a TensorFlow training script that implements a multi-task regression network. The network performs four different regression tasks simultaneously:

The task

You are given a TensorFlow training script that implements a multi-task regression network. The network performs four different regression tasks simultaneously:

Part of PrimeIntellect/Terminal-Lego-15k.