-
I'm trying to work with the `Q` matrix from a qr-factorization within Zygote. In an incomplete QR factorization for m>=n,
the `QRCompactWYQ`Matrix `Q` has `size=(m,m)` but only the first `n` columns …
rkube updated
3 years ago
-
TL;DR The calculated gradient of the max-operation is not always in the subdifferential.
**Idea**
Let **x** _= (x_1,...,x_N)_ be a real vector and consider the function
_f(x_1,...,x_N) = max(x_1,...…
-
# `.coordinates`
- [x] cartesian_to_poincare_polar (https://github.com/GalacticDynamics/coordinax/pull/136)
- [ ] make_greatcircle_cls (will be done in coordinax)
- [ ] pole_from_endpoints (will…
-
Hi.
As mentioned in A.4 in your paper, the center coordinates (x, y) and dimensions (w, h) of the box both will be refined iteratively, and the initial box is set with b_w=0.1 and b_h = 0.1.
https:…
-
Currently any code that reduces GPU arrays to a single scalar value (like `sum`) does device-to-host copy of a single element at the end to return `::Number` instead of `::GPUArray`.
But each such tr…
-
I have just read the very recent paper `[GaLore: Memory-Efficient LLM Training by Gradient Low-Rank Projection]`([arXiv 2403:03507](https://arxiv.org/abs/2403.03507)) that allows Llama 7B training to …
-
Hi, I have tried to run the code according to Usage in this repo:
`args = parse_args()
num_gpus = int(os.environ["WORLD_SIZE"]) if "WORLD_SIZE" in os.environ else 1
args.num_gpus = num_gpus
args.d…
-
Can I know why "From the geometric perspective, ...normal vector x"?
Also, I can't quite understand your figure 4 too. Why $\frac{\partial L}{\partial \bar{x}}$ points in the south east directi…
-
I went through and trialed the code and found an error:
```
// For the Poisson equation the divergence of the guidance field is necessary.
cv::Mat vxx, vyy;
cv::Mat kernelx = (cv…
-
Another potential calculation: velocity potential