Pythia was trained on only 300B tokens and is pretty dumb compared to LLaMA. Pyt... | Hacker News

Hacker Newsnew | past | comments | ask | show | jobs | submit

		MacsHeadroom on April 18, 2023 \| parent \| context \| favorite \| on: RedPajama: Reproduction of LLaMA with friendly lic... Pythia was trained on only 300B tokens and is pretty dumb compared to LLaMA. Pythia 13B is worse than LLaMA-7B and requires double the resources.

Tepix on April 18, 2023 [–]

Not all use cases need GPT-4 level performance. I'd argue that even LLaMA-7B is quite limited. Also, new and improved models are being released all the time.

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact