François Rozet

Following @rdaysky, I think the number of indentation levels should be kept as small as possible and explosion should propagate outwards. Also I don't see why we should consider ```python...

Anything new about this? This issue is the only thing that prevents me (and most Python devs I know) to use and recommend black. Nested bracketed expressions just take too...

Hello @robsdedude, thanks for your PR, I hope it gets merged! > P.S. if you do a lot of `rebase`, avoid using issue identifiers in the commit message as it...

Hey, I'm not a good drawer but I can play with vector graphics and I've made a few logos in the past. I like very sober designs as they are...

Well there is nothing urgent, so we can wait for someone that can draw what you want. If you are tired of waiting, ping me and I'll send you the...

Instead of inferring the docstring style at once, it could be easier to infer the section styles independently. The rationale is that the style of a section is usually clear...

> This approach could actually simplify the way griffe parses docstrings as a lot of the parsing is common to all styles. It would also become easier to add parsing...

Hello 👋 I noticed you wanted to switch from `nflows` to `flowtorch` for your normalizing flows. I must warn you that `flowtorch` is still missing important features. Notably, there is...

@martinjankowiak I implemented the proposed changes (https://github.com/francois-rozet/pyro/commit/fa9d8e69ebb3c2f8c031d756be7c8130667faa6b) and, as I expected, using softplus (like `nflows`) instead of clamping completely solves the problem: the gradients don't explode anymore. Should I make...

@stefanwebb Should I file an issue for FlowTorch as well? Or should I provide evidences first? It's hard to "prove" but I can simply show the losses ^^