c0g

Scatter gradient implementations

1

Not sure if this is an issue or I'm just holding it wrong :-) I've got an alternative approach that seems to work with gradients by building up my output...

[FEATURE REQUEST] mx.grad doesn't alias argnums and argnames

3

Not a bug, a nice-to-have-or-at-least-better-error message-please kind of thing. If you differentiate a function, you can't pass an argument in as a kwarg, even if the fn is differentiated w.r.t...

Linear in Float32 is inaccurate

3

I'm seeing some pretty large divergences in my model when I port to transformer engine. While tracking these down (checking if I'm crazy or TE is crazy), I'm pretty sure...

Best practices for fast rendering with MJX

3

Hi, I'm a engineer and I'm trying to use MuJoCo for fun. I'm looking for some help with rendering simulations in MJX. My model is extremely simple so this might...

RuntimeWarning: os.fork() was called. os.fork() is incompatible with multithreaded code, and JAX is multithreaded, so this will likely lead to a deadlock.

1

## Describe the bug Running jax and clearml in the same process, I get /home/tom/miniforge3/envs/tomrl/lib/python3.10/site-packages/clearml/utilities/process/mp.py:629: RuntimeWarning: os.fork() was called. os.fork() is incompatible with multithreaded code, and JAX is multithreaded, so...