๐ฅ Refer to the setup video for a step-by-step visual guide:Documentation Index
Fetch the complete documentation index at: https://docs.neuraltalk.ai/llms.txt
Use this file to discover all available pages before exploring further.
Overview
The Corrections feature allows you to review, fix, and improve chatbot responses. It ensures your chatbot gives accurate, friendly, and high-quality answers by letting you approve, reject, or refine responses raised during conversations. This guide explains correction statuses, actions, and how to create and manage corrections.Accessing Corrections
Open Corrections
- Log in to your dashboard
- Navigate to Data Source from the main menu
- Click on Corrections in the left sidebar

Corrections Dashboard
The Corrections page is divided into multiple tabs for better management.Correction Tabs
- Pending: Corrections awaiting review
- Corrected: Approved and applied corrections
- Rejected: Corrections that were dismissed
Actions Menu
The Actions dropdown provides bulk and management options.- Create Correction: Manually add a correction to improve chatbot answers.
- Export Approved Corrections: Download all corrected entries.
- Export Pending Corrections: Export pending corrections for review.
- Mark Selected as Corrected: Approve the selected corrections.
- Mark Selected as Rejected: Reject the selected corrections.
- Import Corrections: Upload multiple corrections in bulk.
- Download Template: Download a template for bulk correction import.
Creating a Correction
Click Create Correction to manually add or fix a chatbot response.Correction Form Fields
- Question: Enter the user question that needs correction.
- Original Chatbot Answer: Displays the response originally given by the chatbot.
- Model Answer: Write the correct answer to the question raised.
Answer Enhancement Options
- More Friendly & Detailed: Improves tone and clarity
- Enhance with AI: Automatically refines your answer
Optional Fields
- Source URL: Add a reference link
- Image URL: Attach an image
- Extra Info: Additional instructions like tone or style
- Tags: Add tags for easy filtering
Best Practices
- Review pending corrections regularly
- Use friendly, clear, and accurate model answers
- Add sources for credibility
- Tag corrections for easier management
- Export data for audits or training purposes