Christoph Weniger
Christoph Weniger
Hi Enrico, I just looked at the code, and in that case, you would have to write your own custom one. Best, Christoph
PS: We have another development version of swyft, which is based on pytorch-lightning, and meant to be much more flexible than the current version. If you want DM me and...
Performance depends on the details. For very simple data, and very simple networks, the stores are the bottleneck. If you have a way to improve that, I would be very...
Even after the above merge there are issues when running "infer" with lightning 2.0.0, because of some incomplete limitations of the CombinedLoader in 2.0.0. This probably will solve itself over...
Recent discussions in pl here: - https://github.com/Lightning-AI/lightning/issues/16830 - https://github.com/Lightning-AI/lightning/pull/17163 But it still looks like it will not solve all problems (predict loop is not touched).
Solution strategy: Write custom CombinedLoader within swyft. Might also be useful for #133
- Is parallel training of estimators in the same network enough? - Could this be achieved with MC dropout?
## Suggested Plan ### Functions - `plot_1d`: Plots single 1-dim posteriors - `plot_2d`: Plots single 2-dim posteriors - `plot_corner`: Plots corner plots with multiple 1-dim and 2-dim posteriors; optionally the...
@NoemiAM Good suggestion regarding `plot_pair`. How would you signal that distinction to the routine? One option would be to have [par1, par2, par3, ...] generate a corner-type plot, and [[par1,...
v0.3 not further developed