3 ms·
Also GPT-3.5 was another extreme if I remember correctly. They first trained only on code then they trained on other text. I can't seem to find the source thoug
by YetAnotherNick 2y ago
Also GPT-3.5 was another extreme if I remember correctly. They first trained only on code then they trained on other text. I can't seem to find the source though.