Guy Moss
Guy Moss
Once #908 is addressed there might be even better default args
Additional context: The [embedding network tutorial](https://sbi-dev.github.io/sbi/tutorial/05_embedding_net/) is currently titled titled "learning summary statistics with a neural net", which is not obviously about embedding networks. Suggestions for PR: - [x] Change...
Hi @danielmk! Is there an update from your end on the status of this PR, or any input from our end that could help at this stage?
Assigned myself here for now, happy for anyone else to jump on at any point
Thanks @tomMoral @janfb! DeepInverse and azula (cc @francois-rozet) are both super relevant! Would be grateful for input from the devs.
Can you add some details about your environment? The tests all pass locally for me.
> MAP > > MAP is using the score directly for doing gradient ascent on the posterior to find to MAP. this is currently not working accurately Is it the...
Update after talking to @manuelgloeckler, the easiest way to calculate the MAP here would be to use the score directly at a time t = epsilon, instead of calculating and...
@manuelgloeckler ping re: IID sampling
Related to #968