-
I tested the `torchao.quantize_` and `torchao.autoquant` for the first time on a toy model and wanted to summarize my usability/readability comments, in case that is helpful.
# torchao.quantize_ fe…
vkuzo updated
3 weeks ago
-
Hello there!
At the moment the Debian image is released with kernel and u-boot binaries but no source code. This is a bit of a headache as it means you cannot build kernel modules or update the ker…
-
Fresh clone of this repo (tag v1.12.0), fresh install of Vagrant (2.2.7) and VirtualBox (on my host Mac - 6.1.6).
On `make sd-image` (after `vagrant up`), I get the following error (full output att…
-
Hi, I am trying to do some XAI evaluation experiments using classifiers and various Shap explainer algorithms.
I am using a breast_cancer binary classification toy data set in order to verify that t…
-
Was just testing some timings of your vectorized Mandelbrot version against a naive non-vectorized version that I had lying around, and my non-vectorized version was somehow faster. Example:
My non…
-
### Describe the bug
I've noticed a rather strange behavior when using **Scikit-Learn's SVC** implementation. Using the **built-in rbf kernel** with SVC is **slower** by magnitudes than passing a *…
-
Hello,
I was trying to setup a toy module here including a LSTM layer. Simply here it is:
>>> from keras import layers
>>> from keras import Model
>>> inp = layers.Input(shape = (120, 1, ))
>…
-
## 🚀 Feature
Intermediate tensors are currently re-computed instead of re-used in fragmented computation. This could cause regression comparing to eager, which reuses tensor and avoids wasted re-co…
-
```
I've compiled and installed latest mind-toys snapshot.
MinD-Toy v1.1 BETA
Revision: 103
This program is free software; you can redistribute it and/or modify
it under the terms of the GNU General…
-
### Validation
- [X] I've read the [FAQ](https://github.com/xenia-canary/xenia-canary/wiki/FAQ).
- [X] The Xenia build used is from a canary branch. (not MLBS/AlexVS/master/pull requests, etc)
- …