Skip to content
NEURAX
Download

Add your dataset

A file or a whole folder, read where it is.

NEURAX reads the shape of your data — how many samples, which columns, how many classes, the size of one sample — and never shows or sends what is in it. The data stays where it is; a training run reads it there.

  1. 1Open Dataset in the toolbar and choose This machine (or your own Amazon S3, Google Cloud or Azure Blob storage).
  2. 2Give the file or the folder. On the desktop, Choose a folder picks a whole folder of images at once — one subfolder per class — never one image at a time. In a browser, type the full path.
  3. 3Press Read. The profile appears at the top of the menu.
1 — where the data is: this machine. 2 — its full path, a file or a folder. 3 — Read: NEURAX reads its shape; nothing is sent anywhere.
1 — what was read: 400 news titles, 4 classes (the topics), 64 tokens each. These figures drive the analysis, Decision and the run.
Your dataWhat NEURAX reads
A table.csv, .tsv, .parquet — one row per sample, the target in a column named label, target or class
Records.jsonl, .ndjson, or a .json array — one record per sample
ImagesA folder with one subfolder per class; train/ and val/ folders are kept as they are

To use another dataset, open Dataset again and read it: it replaces the first once it is read. One that cannot be read leaves the dataset in use as it was, and says why.

Page 5 of 33. This guide is in English for now. It is the same guide that is built into the NEURAX desktop app (the ? button, or F1). Found a mistake? Tell us.