Home

Here we’ll walk you through how to use the Jaxon Platform with a toy dataset.

On this page you will:

  1. Download the Toy Dataset

  2. Create a Project

  3. Import the Toy Dataset

  4. Assign the Specification

  5. Split into Train and Test Sets

  6. Train a Neural Model

  7. Create an Ensemble

  8. Synthetically Label the Training Data

1. Download the Toy Dataset

  1. To follow along with this guide, download the associated dataset found below: Jaxon_IMDb_UI_Guide

<aside> <img src="https://s3-us-west-2.amazonaws.com/secure.notion-static.com/a8321f0f-f79a-4528-b424-f443c09df758/0_JAXON_Logo_Mark_2.jpg" alt="https://s3-us-west-2.amazonaws.com/secure.notion-static.com/a8321f0f-f79a-4528-b424-f443c09df758/0_JAXON_Logo_Mark_2.jpg" width="40px" /> We will not be covering labeling within the platform in this guide - for more information on how to label within the Jaxon Platform, go to the Labels tab of our Platform Reference Guide.

</aside>

This dataset is a version of the publicly-available IMDb dataset referenced in this paper. This particular version of the IMDB dataset has been significantly reduced for this guide.

Back to the top ↑

2. Create a Project

  1. On initial start, Jaxon will open on the Projects tab. Create a new project by clicking the + button

e.png

  1. Fill in the Name field and, optionally, the Description field in the box that pops up and click  Submit

j.png

  1. Click on the project you just created. Once the project has been successfully selected, you’ll see the Datasets and Specification tabs become available. We’ll upload a dataset next, so click on the Datasets tab.

101.png

Back to the top ↑

3. Import the Toy Dataset

  1. To import the toy dataset, click  +  from the Dataset Menu

k.png

<aside> <img src="https://s3-us-west-2.amazonaws.com/secure.notion-static.com/b1727c50-70e2-4a9b-9c05-8e8b63ce8049/0_JAXON_Logo_Mark_2.jpg" alt="https://s3-us-west-2.amazonaws.com/secure.notion-static.com/b1727c50-70e2-4a9b-9c05-8e8b63ce8049/0_JAXON_Logo_Mark_2.jpg" width="40px" /> If you haven’t downloaded the toy dataset yet, you can find it here.

</aside>

  1. Fill out the intake form and select Submit

103.png

  1. In the box to the left side of the dataset preview, check the box for the Header Row and click the Import button

r.png

  1. Once the dataset is imported, which columns are the features and which are the labels must be assigned. From the dropdown under the column titled “examples” select “Feature: Free-Form Text”. From the dropdown under the column titled “labels” select “Labels”.

s.png

<aside> <img src="https://s3-us-west-2.amazonaws.com/secure.notion-static.com/be355d7b-9f3e-40bb-b5b7-dacc12be57ad/0_JAXON_Logo_Mark_2.jpg" alt="https://s3-us-west-2.amazonaws.com/secure.notion-static.com/be355d7b-9f3e-40bb-b5b7-dacc12be57ad/0_JAXON_Logo_Mark_2.jpg" width="40px" /> It may take a few minutes (~5) for the dataset to be imported into Jaxon.

</aside>

Back to the top ↑

4. Assign the Specification

  1. Next, click on the Specification tab

106.png

  1. In the Import From Dataset dropdown, select the Jaxon_IMDb dataset

107.png

  1. After a few moments, you should see the two classes (neg and pos) appear on the screen

108.png

  1. Click the  Lock  button, and then OK on the popup that appears.

109.png

110.png

Back to the top ↑

5. Split into Train and Test Sets

  1. Back in the Datasets tab, select the Jaxon_IMDb dataset by clicking on it. When a dataset is successfully selected, the list of datasets disappears and only a preview of the selected dataset is shown.

t.png

  1. Select the Split Dataset icon

split dataset icon.png

u.png

<aside> <img src="https://s3-us-west-2.amazonaws.com/secure.notion-static.com/18bb22f2-ee84-46a9-8f2e-e74275b326e3/0_JAXON_Logo_Mark_2.jpg" alt="https://s3-us-west-2.amazonaws.com/secure.notion-static.com/18bb22f2-ee84-46a9-8f2e-e74275b326e3/0_JAXON_Logo_Mark_2.jpg" width="40px" /> Splitting the dataset may take several minutes.

</aside>

6. Train a Neural Model

Next we will train a model that we will then use to label unlabeled examples in the Train set.

For more details on these metrics, see this Example Neural Training Overview from the Platform Reference Guide.

Back to the top ↑