-
## 🚀 Feature
FSDP to offer the possibility to flatten parameters by group, for instance, to flatten all biases separately from the other weights.
## Motivation
Following issue https://github.…
-
#### Describe the bug
Errors:
You may have to run the code several times to see these errors. I think most of them is to do with the way the initial number of samples is selected; If the initial n…
-
Votez pour les séquences que vous souhaiteriez voir aborder avec 👍 !
-
"Make HemeLB Approach Adequacy Again"
As of today, there are 24 branches and 171 open issues. This is not cool. Let's drain the swamp.
Branch wrangling:
- [x] Rename master -> main (@rupertnash)
…
-
!cv1-1.png|width=200,height=183!
Would it be possible to implement Time Series K-Fold Cross-Validation?
For example:
With a Time Series Training/Validation Interval which goes from 2017-01-01 to 2…
-
Hi,
I've recently created a dataset using speech-to-text APIs on custom documents. The dataset consists of 1,000 audio samples, with 700 designated for training and 300 for testing. In total, this eq…
-
The difference between training accuracy and testing accuracy is large pointing at overfitting. The problem seems to be because of very basic pre-processing. We are just indexing each unique word and …
-
Hi, I am trying to perform CCA on my 2 data sets, both of which are of shape 6030 samples by 499 features. I am using
```
cca = rcca.CCA(kernelcca = False, reg = 0., numCC = nCCA)
cca.train([x1, x2…
-
There is less viscosity then there should be. Due to incorrectly ordered arguments in internally called functions.
-
Hi,
I'm attempting to run AROMA on resting state data that was preprocessed in SPM 12 (slice time corrected, functional data realigned to the mean volume, coregistered to structural, within normali…
hldot updated
6 years ago