Skip to content

Latest commit

 

History

History
78 lines (49 loc) · 5.25 KB

pundi-ai-data-platform-alpha-testnet-user-guide.md

File metadata and controls

78 lines (49 loc) · 5.25 KB

Pundi AI Data Platform (testnet) - User Guide

Introduction

This is a user guide to the Alpha Testnet version of the Pundi AI Data Platform. We encourage everyone to try out the Alpha Testnet, although do bear in mind that as this is not mainnet, some things are not 100% stable and might break.

Software requirements

Type Web application
Software requirement

Stable Internet connection

Recommended web browser: Google Chrome
Access link https://data.pundi.ai/testnet/

Blockchain & Wallet

Blockchain Pundi AIFX omnichain Testnet
Supported Wallets MetaMask
Cryptocurrencies involved

$USDT (testnet token)

$FX (testnet token)

How to tag data on the Pundi AI Data Platform

Step 1: Connect Your Wallet

  • Open the application and connect your wallet using metamask or wallet connect.(Add quick picture guide of connecting wallet)

  • Navigate to the Browser Tasks section to begin.

Step 2: Sort Available Tasks

  • You can sort the tasks based on:
    • Annotation Style (e.g., single label text classification)
    • Reward Amount
  • Choose the sorting method that best suits your preference.

Step 3: Choose Your Task Type

  • Decide whether you want to:
    • Label Data: Directly annotate the given dataset.
    • Review Data: Verify and rate previously labeled tasks to ensure data quality.

Step 4: Labeling Data

  1. Understand the Task: Notes- Improve picture quality.

    • Read the task description to get a clear idea of what is expected.
    • If you need more information, refer to the Annotation Guidelines provided.

  2. Annotate the Data

    • Select the option that best describes the data based on the Annotation Guidelines.
    • Click Submit to save your response.

  3. Complete the Task

    • Continue labeling until the entire task is complete.

  4. Review and Reward

    • Your labeled work will be reviewed by a team of reviewers.
      • Reviewers will give a rating on a scale of 1 to 5 (1 for Poor and 5 for Excellent) to measure the quality of the annotation (based on by metrics such as the accuracy, precision and completeness).
      • Each accepted labeled data row will be evaluated and rated by 3 Reviewers. Once the sufficient number of Reviewers provide their ratings, the data row will be removed from the review queue and go through the Inter-Reviewer Agreement process automatically.
    • After the accuracy and quality of your efforts have been verified, the reward will be released to your wallet.