You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
We have the original draft of the SKL LabelEncoder code uploaded in this branch. All credit here goes to @scnakandala, the original author and brains behind this.
It contains an un-edited implementation of LabelEncoder here.
There is a test file here that needs to be cleaned up and passing (and also tests added for errors).
Notes:
We had to treat strings a bit differently because of the way they are represented in tensors
We should add more expansive tests
In the numeric label encoder, there is a tricky spot here related to running on GPU vs CPU.
There may be some bugfixes/improvements that are available now. Our CI/CD pipeline is not currently GPU-enabled, but it would be helpful to be able to run this on a GPU-enabled machine when developing.
The text was updated successfully, but these errors were encountered:
Btw, many people that are using LabelEncoder should actually be using OneHotEncoder. LabelEncoder was meant for labels i.e. targets, not features, and only supports a single column, while OneHotEncoder supports multiple columns.
We have the original draft of the SKL LabelEncoder code uploaded in this branch. All credit here goes to @scnakandala, the original author and brains behind this.
It contains an un-edited implementation of LabelEncoder here.
There is a test file here that needs to be cleaned up and passing (and also tests added for errors).
Notes:
The text was updated successfully, but these errors were encountered: