4 ms·
Alpha zero used 5000 TPUs to generate games (inference only), and 16 to train the networks. The split definitely depends on what you're doing past developing/d
by gpm 4y ago
Alpha zero used 5000 TPUs to generate games (inference only), and 16 to train the networks.
The split definitely depends on what you're doing past developing/deploying.
(Source: https://kstatic.googleusercontent.com/files/2f51b2a749a284c2e2dfa13911da965f4855092a179469aedd15fbe4efe8f8cbf9c515ef83ac03a6515fa990e6f85fd827dcd477845e806f23a17845072dc7bd https://kstatic.googleusercontent.com/files/2f51b2a749a284c2...)
- varunkmohan 4y agoCompletely agreed. For some of these large language models, it would take a long time before inference spend dominates training spend.