Benchmarking training throughput of contemporary diffusion models on AMD's GPU hardware. - View it on GitHub