Notes for CS 229 Students
#1
by
louiecerv
- opened
This is an example of an app that can really benefit from using the Huggingface dataset. The actual dataset is just over 3 mb. If you load the csv file in the Huggingface space without first creating a Dataset, the performance is really poor compared to when it is first uploaded as a dataset then the App uses the huggingface hub code to access the dataset.
Examine the code also note the sequence of pre-processing tasks and the various commands to visualize the dataset.
The scikit-learn implementation of the MLP is much slower than the other ML models.
You can further explore ways to customize the MLP model to increase its performance.