3 ms·
The issue here is GPT-3/4 were trained on bullshit. Surely bullshit is the desired and expected outcome when you train your model on bullshit Reddit comments.
by Stevvo 3y ago
The issue here is GPT-3/4 were trained on bullshit. Surely bullshit is the desired and expected outcome when you train your model on bullshit Reddit comments.