Stijn De Weirdt

@yaoyasong @codingl2k1 my apologies, but how was this fixed? i just ran into this with version 0.12 i see that on https://github.com/xorbitsai/inference/blob/main/xinference/model/rerank/core.py#L211 you convert the `arg` to an int, there...

@codingl2k1 the bge gemma v2 one. but i do not always run into the issue

@aresnow1 @codingl2k1 is it ok/safe to cast the `arg` to an int?

it is a different cause, i opened https://github.com/xorbitsai/inference/issues/1775

@wing328 i was not suggesting to drop all older python versions. i am well aware that i spartically impossible. the impact depends on how you use the enums in code:...

@johrstrom can you share how to enable this in 3.1? i would like to have a look at it, and esp see what we can/need to adapt to our setup.

@johrstrom how odd. i already did that (i found that bit in the code), and then the projects link in the navbar showed up; but didn't do anything. but now...

@johrstrom permit only allows scalars (see PERMITTED_SCALAR_TYPES), so i made set of ugly patches for 3.1 to squash the array to a string (and have the setter unpack it again)....

@johrstrom thanks for considering it for 4.1. i'll take any release over maintaining patches and hacks

@johrstrom i found the issue with the cacheable i think. the current cacheable behaviour doesn't like reusing the same attribute name with very different values or even widgets. eg in...