For ML maybe? You can get the AMD builds of TF and PyTorch and then build a cluster out. Realistically it'd likely to be more trouble than its worth. The AMD

For ML maybe? You can get the AMD builds of TF and PyTorch and then build a cluster out. Realistically it'd likely to be more trouble than its worth. The AMD versions are always behind the core versions.