SFT-data-builder: generate AI training data using free big model API, 0 cost big model training data generation

General Introduction

SFT-data-builder is an open-source project designed to generate high-quality SFT training data by utilizing the free Big Model API combined with users' private domain data. The tool supports a variety of AI model formats and provides one-click generation, batch generation, flexible editing and local storage functions to help users quickly generate training data suitable for pre-training, fine-tuning, function calling and other scenarios.

Function List

• Generate training data with one click: Numerous OpenAI format calls for local or cloud-based models are supported.
• Batch Generation: Generate training data from multiple different perspectives at once, support batch URL articles to generate data automatically.
• Flexible editing: All generated data can be edited and adjusted at any time.
• local storage: Automatically saves all data locally.
• Easy to export: Export standard format JSON files with one click.
• Multi-model support: Supports a wide range of mainstream AI models, which can be customized.
• Multi-format support: Support for PDF, Word, TXT and other file formats.

Using Help

Installation process

1. Installation of dependencies: Run in the project directory npm installThe
2. Initiation of projects: Run npm run start Launching the project.

Guidelines for use

1. Configuring the API::
   • Click on the "Open Configuration" button.
   • Set the API address and key.
   • Select or customize the AI model.
   • Sets the number of data entries generated at a time.
2. input::
   • Upload files (PDF, DOCX, TXT supported).
   • or enter the text content directly.
3. Generate data::
   • Click the "Generate AI Response" button.
   • Toggle through multiple generated results and edit the generated content as needed.
4. Management data::
   • Add to the data list.
   • Preview all generated data.
   • Delete unwanted data.
   • Export as a JSON file.

Functional operation flow

1. Generate training data with one click::
   • Select or customize the AI model.
   • Enter or upload text content.
   • Click the "Generate AI Response" button, the system will automatically generate training data.
2. Batch Generation::
   • Set the batch generation parameters on the configuration page.
   • Upload a file containing multiple URLs or enter multiple URLs.
   • Click the "Batch Generate" button, the system will automatically generate multiple training data.
3. Flexible editing::
   • On the Generated Results screen, click the data entry to be edited.
   • Modify the content in the editor to save the changes.
4. Local storage and export::
   • All generated data is automatically saved to local storage.
   • On the data management screen, select the data to be exported and click the "Export to JSON" button.