-
We still don't have a gradient implementation for `Eig` (and a few other `aesara.linalg` `Op`s like https://github.com/aesara-devs/aesara/issues/836). [Here's](https://github.com/aesara-devs/aesara/p…
-
There's a nice duality between `custom_gradient` and `jax.jvp`/`tf.gradients` that is exactly what is needed for piping gradients back and forth, e.g.,
```python
# untested!
import jax
import tens…
-
From Slack-comment by @koehlerson; how to implement custom gradient calculation for a multi-argument function.
It is common to have such a case for autodiff, so would be good to have a clear way of …
-
This code produces a compilation error:
```
using Enzyme
using EnzymeCore
# Required presently
Enzyme.API.runtimeActivity!(true)
Enzyme.API.looseTypeAnalysis!(true)
Enzyme.API.maxtypeoffset!(20…
-
Hello, has TensorNet accelerated during training? Hope you can share the results of the training speed, thank you
WWWJL updated
4 years ago
-
-
### Description
Hey, folks!
If I don't provide consistent data for gradient calculation, and try to filter the mess at later stages (pre_filter=False), I get NaN gradient while pre-filtering works…
-
I implemented a simple halo exchange:
```julia
using MPI
using Enzyme
struct Heat
Tnext::Vector{Float64}
end
function halo(heat)
next = heat.Tnext
np = MPI.Comm_size(MPI.COMM_…
-
I'm porting my model from lasagne/theano .
I found there are some layers I need not exist in neon.
The first question I want to ask is " Is there any tutorials about creating custom layers ?"
…
-
I observed that I was getting surprisingly different LQR gains for a `MultibodyPlant` with time_step=0 vs time_step=small and spent some time investigating. I eventually winnowed it down to a minimal…