Lora Toolkit

Overview

Lora Toolkit is a unified suite of tools designed to streamline the management and processing of training data for LORA (Low-Rank Adaptation) models. It combines four essential utilities into a single, user-friendly interface with a dark mode, material design aesthetic.

Features

1. Site Builder

• Create a searchable website to visualize all your training images and associated tags.
• Generates an HTML gallery with individual pages for each image.
• Enhances data exploration and presentation.

2. Prompt Cleaner

• Remove specific words or phrases from all your training tags.
• Useful for cleaning up unwanted or repetitive text in your dataset.

3. Add Trigger Word

• Prepend any word or phrase to the front of your training tags.
• Ideal for adding consistent triggers or identifiers across your dataset.

4. Prompts to CSV

• Compile all your training tags into a single CSV file.
• Facilitates data analysis and sharing.

New Feature: Subfolder Support

• All tools now support processing of files in subfolders, allowing for more organized dataset structures.

Installation

Follow these steps to set up the application on your local machine.

Prerequisites

• Python 3.6 or higher installed on your system.
• pip (Python package installer).

Clone the Repository

git clone https://github.com/psdwizzard/Lora-Toolkit.git
cd lora-toolkit

Install Required Packages

Install the necessary Python libraries using pip:

pip install customtkinter pillow

• customtkinter: For an enhanced, custom-styled graphical user interface.
• pillow: For image processing in the Site Builder.

Usage

Launching the Application

Run the Python script to start the Lora Toolkit:

python lora_toolkit.py

Note: Replace lora_toolkit.py with the actual filename if it's different.

Navigating the Interface

• Upon launching, you'll see a dropdown menu at the top.
• Select the desired tool from the dropdown to switch between applications.
• The interface now uses CustomTkinter for an improved user experience.

Site Builder

Purpose: Create a searchable website to visualize all your training images and associated tags.

Steps:

1. Select Image and Text Files Folder:
   • Click the Browse button to choose the directory containing your images and corresponding .txt files.
   • The tool will now process files in the selected directory and all its subfolders.
   • Each image should have an accompanying .txt file with the same name (e.g., image1.png and image1.txt).
2. Enter Project Name:
   • Provide a name for your project. This will be used as the title of the website and the main HTML file (e.g., MyProject.html).
3. Select Output Folder:
   • Choose the directory where the generated website files will be saved.
4. Generate Website:
   • Click the Generate Website button.
   • The application will create an index HTML file with a gallery of images and a search functionality.
   • Individual HTML pages for each image will also be generated.

Prompt Cleaner

Purpose: Remove specific words or phrases from all your training tags.

Steps:

1. Select Directory:
   • Click Browse to choose the folder containing your .txt files (tags).
   • The tool will process all .txt files in the selected directory and its subfolders.
2. Enter Text to Remove:
   • Input the word or phrase you want to remove from all tags.
3. Execute:
   • Click Go Burr to start the cleaning process.
   • The specified text will be removed from all .txt files in the directory and its subfolders.

Add Trigger Word

Purpose: Add specific words or phrases to the beginning of your training tags.

Steps:

1. Select Directory:
   • Click Browse to choose the folder containing your .txt files (tags).
   • The tool will process all .txt files in the selected directory and its subfolders.
2. Enter Text to Add:
   • Input the word or phrase you want to prepend to all tags.
3. Execute:
   • Click Go Burr to start the process.
   • The specified text will be added to the front of each .txt file's content in the directory and its subfolders.

Prompts to CSV

Purpose: Compile all your training tags into a single CSV file.

Steps:

1. Select Directory:
   • Click Browse to choose the folder containing your .txt files (tags).
   • The tool will process all .txt files in the selected directory and its subfolders.
2. Enter CSV File Name:
   • Provide a name for the output CSV file (e.g., training_tags.csv).
3. Generate CSV:
   • Click Go Burr to create the CSV file.
   • The CSV will contain two columns: File Name and Content, including data from all processed subfolders.