Uploading Your Own Datasets to Floydhub

Uploading Your Own Datasets to Floydhub


If you are using Floydhub to train your network you might run into the following issue:

Directory contains too many files to upload. If you have data files in the current directory, please upload them separately using "floyd data" command and remove them from here.

This occurs right before you are trying to upload files to Floydhub. It makes sense since you might have a large dataset to train on.


The solution is to upload your dataset to Floydhub.

  1. Open your command line program or terminal
  2. Navigate to your dataset directory:
    cd ~/Desktop/images-dataset
  3. Initialize the floyd dataset on the current directory:
    floyd data init images-dataset
    Your default browser should open the first time you init the dataset. Just retype the same command after you create the dataset.
    Floydhub admin
  4. The final step is you upload your dataset:
    floyd data upload
  5. Wait for your data to upload

Additional Resources:

Edgar Pino

About Edgar Pino

I’m a passionate software engineer working @ Pluralsight interested in distributed systems, machine learning, and the web. I am also a graduate student at the University of Illinois.