-
### 🚀 The feature, motivation and pitch
In PyTorch, we have `torch.nn.Linear` and `torch.nn.functional.linear(x, w, b)`.
Do we have similar counterpart? For example, `torch_geometric.nn.GCNConv` and…
-
### Description
A sufficiently-large (7x7) `nnx.Conv` with `float32` parameter dtype, when sharded across multiple devices, generates `nan`, seemingly due to overflow.
The `nan` is avoided by maki…
-
Hi -- first off, this package is incredible. Thank you.
Second, have you considered adding further random sampling functionality (for other distributions on spaces of rotations). In the title, I re…
-
Dear team,
I'm having an issue restoring pure dict from nnx.state,
Orbax will change the key has int type into string type.
Is there any way to fix it?
I put a sample code that can reproduce the issu…
-
### Report
Hello,
When running the lda visualisation plot as part of the mixscape workflow, I am not getting consistent results between runs, the plot looks slightly different (although very simil…
-
I was wondering if the equinox community has good advice / best practices for creating dataloaders that work well with jax? I've done some stuff from scratch but I tend to find that the training (on G…
-
Feature Request bessel function in scipy ( J0 J1 J2 Y0 Y1 Y2) which have applied in autograd.
:
File "/user_work_path/miniconda3/envs/lwm/lib/python3.10/runpy.py", line 196, in _run_module_as_main
return _run_code(code, main_global…
-
I'm currently making my own implementation of a gaussian process, and I'd like to cache certain results like the correlation matrix (`K`) and the inverse of the correlation matrix multiplied by the tr…
-
A to-do list of things that still need to be done on the statespace models, to organize future PRs. Anyone is free to add to (or subtract from!) the list.
- [ ] **High priority**: Allow post-estima…