3 ms·

I see you already mention diffusion - iirc there was a result not too long ago that diffusion models keep improving with more epochs for longer than AR models d

• sdpmas 7mo ago
  diffusion is promising, but still an open question how much data efficient they are compared to AR. in practice, you can also train AR forever with high enough regularization, so let's see.
  • _0ffh 7mo ago
    Yes, it could go either way of course. Still, just for reference, here's the paper I remembered: https://arxiv.org/pdf/2507.15857 https://arxiv.org/pdf/2507.15857
    • sdpmas 7mo ago
      thanks, here's another one: https://arxiv.org/abs/2511.03276 https://arxiv.org/abs/2511.03276