-
if we have `vector_a + scalar_b + scalar_c` we have two vector additions. If we were to rewrite this to first sum all the scalars (`scalar_b + scalar_c + vector_a`) it would be twice as fast as we onl…
-
Hi!
I'm setting up a benchmark of different linear algebra libraries, and I was wondering how to produce a large matrix? Libraries often have a `fromList` method or similar.
-
**System information**
- OS Platform and Distribution (e.g., Linux Ubuntu 16.04): Mac
- TensorFlow installed from (source or binary): Source
- TensorFlow version (or github SHA if from source): 2.2…
-
" Thanks to our efficient kernels, AWQ achieves **1.45× and 2× speedup over GPTQ and GPTQ with reordering** on A100. It is also 1.85× faster than an FP16 cuBLAS implementation "
-
# 🚀 Feature request
Splitting the discussion that started here: https://github.com/huggingface/transformers/pull/10301#issuecomment-782917393 to add the potential future feature of transformers and…
-
Some MLFF models expose properties other than the forces and energies. For example, [CHGnet](https://www.nature.com/articles/s42256-023-00716-3) also predicts the magnitude of the magnetic moment. Is …
-
Why are the matrix and vector types separate depending on the value they contain? (i.e. `Matrix32[M, N]` and `Matrix64[M, N]` instead of `Matrix[M, N, T]`) ? Since the implementations of matrix and ve…
-
## 🚀 Feature
Conjugate gradient descent, and Linear operator as implemented in scipy needs to have a place in pytorch for faster gpu calculations.
## Motivation
Conjugate gradient Descent a…
-
# 🐛 Bug
Evaluating an acquisition function with `q=1` with `SaasFullyBayesianSingleTaskGP` requires an unnecessarily large amount of memory, due to an inefficient broadcasted `matmul` operation.
…
-
Hello,
I am trying to use EVA for a simple and encrypted MNIST model classifier.
The code for my ConvNet is the following
```
class ConvNet(torch.nn.Module):
def __init__(self, hidden…