idonotknowwhy on Microsoft researchers developed a hyper-efficient AI model that can run on CPUs · Hackerly

Comment on Microsoft researchers developed a hyper-efficient AI model that can run on CPUs parent

idonotknowwhy 1y

Then i guess vocab is the IPC. 10k mistral tokens are about 8k llama3 tokens