3 ms·AMD GPUs handle all inference for Llama3 at Meta btw.by ipsum2 2y agoAMD GPUs handle all inference for Llama3 at Meta btw.