Hacker News
new
|
past
|
comments
|
ask
|
show
|
jobs
|
submit
login
WithinReason
4 days ago
|
parent
|
context
|
favorite
| on:
Faster convergence for diffusion models
I meant sequential generation, I didn't mean using an RNN.
Diffusion doesn't work on pixels directly either, it works on a latent representation.
kleiba
4 days ago
[–]
All NNs work on latent representations.
reply
barrkel
4 days ago
|
parent
[–]
The contrast here is real: there are pixel space diffusion models and latent space diffusion models. Pixel space diffusion is slower because there's more redundant information.
reply
Guidelines
|
FAQ
|
Lists
|
API
|
Security
|
Legal
|
Apply to YC
|
Contact
Search:
Diffusion doesn't work on pixels directly either, it works on a latent representation.