AIJan 9, 2018

Deep In-GPU Experience Replay

arXiv:1801.03138v12 citations
Originality Synthesis-oriented
AI Analysis

This work addresses training efficiency for reinforcement learning practitioners, but it is incremental as it adapts an existing method to a new hardware setup.

The authors tackled the problem of slow training in reinforcement learning by moving experience replay from RAM to GPU, eliminating CPU-GPU data transfers. They demonstrated a 2x speedup in training time when applying this in-GPU experience replay to train an agent for Super Smash Bros. Melee with a batch size of 128.

Experience replay allows a reinforcement learning agent to train on samples from a large amount of the most recent experiences. A simple in-RAM experience replay stores these most recent experiences in a list in RAM, and then copies sampled batches to the GPU for training. I moved this list to the GPU, thus creating an in-GPU experience replay, and a training step that no longer has inputs copied from the CPU. I trained an agent to play Super Smash Bros. Melee, using internal game memory values as inputs and outputting controller button presses. A single state in Melee contains 27 floats, so the full experience replay fits on a single GPU. For a batch size of 128, the in-GPU experience replay trained twice as fast as the in-RAM experience replay. As far as I know, this is the first in-GPU implementation of experience replay. Finally, I note a few ideas for fitting the experience replay inside the GPU when the environment state requires more memory.

Foundations

The foundational work for this paper's niche, ranked by how specifically the neighbourhood builds on it — not by global fame.

Your Notes