-
Currently the code runs on one device, which doesn't allow scaling to larger computational network such as TPU pods.
Parallelizing over local sampler should be relatively simple, since that does no…
-
I had assumed that by making use of an analytical solver there would be gradients flowing back through HybrIK. However, when I set
` pred_xyz_jts_24.register_hook(lambda grad: print(grad))
`…
-
In the simplest case, if the flow is a zero tensor, all source elements are copied to the same position in the target tensor. If flow changes slightly, the target tensor will also change, that means t…
-
I have a doubt related to the Photometric loss.
Am I correct in thinking that, the surface points that is calculated through bilinear interpolation isn't a function of network parameters. As such g…
-
The features of the photos I got using the TVL1 method are not obvious.Below is an optical flow feature that I got using casmeii's EP02-01f.How to change to get obvious optical photo?
![optical_flow_…
-
```julia
function max1(a, b)
r = 0
if a > b
r = a
else
r = b
end
return r
end
gradient(max1, 2.0, 3.0) # (nothing, 1.0)
```
```julia
function max2(a,…
-
Thanks for your amazing work! While the results of training looks decent, I noticed that the triangle walking algorithm is somehow not working in the implementation.
To be more specific, in `walki…
-
### Problem
Compilation can't be used with run-time control flow. This stops some code from taking advantage of tape compilation.
### Possible solution
Enable ReverseDiff's tape caching functiona…
-
#!/bin/bash
CUDA_VISIBLE_DEVICES=7 accelerate launch \
--mixed_precision bf16 \
--num_cpu_threads_per_process 1 \
flux_train_network.py \
--pretrained_model_name_or_path flux1-s…
-
### Issue type
Bug
### Have you reproduced the bug with TensorFlow Nightly?
No
### Source
source
### TensorFlow version
2.15.1
### Custom code
Yes
### OS platform and dis…