Dan Saattrup Nielsen

It might not be what you want, but I have at least ported the (updated) source code for the quantile regression forests into my `doubt` package, https://github.com/saattrupdan/doubt, if that's of...

Implemented the above and more in https://github.com/scikit-garden/scikit-garden/pull/89

Given that this package does not seem to be maintained, I have forked the quantile regression forest bit of the code into my `doubt` package, https://github.com/saattrupdan/doubt, if that's of any...

We're faced with this lack of standardisation too. In our package we're working with metrics in a generic fashion, and not having a uniform output format of the metrics is...

Lower bounds would be great! It took almost 6 hours to install the package with unbounded constraints - hopefully lower bounds will bring that down significantly. Also, Poetry installed an...

> Actually, think this is not ray issue, but from cuda itself (that cuda's cache is not cleaned up). Can you try calling this before new initiailzation. > > ```...

> One more thing, can you try running it with tensor_parallel_size=1, and see if the cleanup is happening? In this case it's all good, the GPU memory is fully reset...

Bumping this as well, as we also need this to support Pydantic 2.x in our integrations.

> I have removed the `FSM` import that made this particular test fail. > > The thing is that the underlying implementation is the same as with the `Guide` interface...

> Indeed @saattrupdan I think it makes more sense to have the processors in vLLM so Outlines is not concerned with details of the implementation and can focus on structured...