-
Right now, the FreshnessSamplingStrategy without resets and with a limit implements the limit either using a sampling uniformly at random, or by just taking the newest data. We should probably add a s…
-
In GitLab by @git.ligo:alec.gunny on Aug 16, 2023, 21:59
When sampling a prior with a `conversion_function` with `size=1`, any keys added inside the `conversion_function` end up in the sampled parame…
-
## Expected Behavior
- Would be nice to have both posterior and prior sampling options from a `BaseRV` object
- Is it even useful to have "uniform" vs "normal" etc variables if they just change afte…
-
Hi! I'm doing a deep dive into the implementation and have some questions.
1. I see that the key caching-related codes are in `DeepCache.sd.pipeline_stable_diffusion.py` under StableDiffusionPipel…
-
### ✨ Short description of the bug [tl;dr]
The current implementation of the random start of `PGDL2` does not sample uniformly in the L2 ball (as done by the [original paper](https://arxiv.org/abs/17…
-
We'd like to integrate several varieties of negative sampling into the NVT/Merlin dataloaders:
- [ ] Popularity biased negatives from in-batch examples
- [ ] Uniform negatives from all items
- [ …
-
Dear Daniel,
There is an interesting method for almost uniform sampling of SO(3) (i.e. the space
of all possible rotations).
This can be useful in some numerical simulations.
cf. "Super-Fibona…
-
Hello,
I have a question regarding the use of SALib for Sobol analysis and wasn't sure where to post it, so I'm posting it here.
Can SALib's sobol.analyze function calculate Sobol indices for a …
-
The paper proposes the non-uniform time step sampling to increase the probability of t falling in the early sampling stage but in the code of the forward method in LatentDiffusion, the time t is gener…
-
(1) Why use different latent factor matrix to represent the source vertex (P, Q) and the context vertex(PC, QC)?
(2) In the negative sampling, you uniformly draws a random negative sample from the wh…