About

Fine-tuning AI is a powerful way to improve its performance on translation and linguistic tasks. However, collecting training data can be a time-consuming process for linguists, often requiring numerous edits before AI starts adapting to the right translation patterns.

The Synthetic Data Generator app streamlines this process. When a linguist identifies a repetitive mistake in AI translations, they can easily describe the issue within the app. The app then generates a synthetic dataset for training, helping to quickly produce a variety of examples that demonstrate the correct translation approach.

Key Features:

• Generate Synthetic Data: Describe a translation issue and automatically generate training data. Edit and Review: Review and fine-tune the generated synthetic data before adding it to your Translation Memory (TM).
• Translation Memory Integration: Store synthetic data in a dedicated TM for future use in AI model fine-tuning.

Note: Please configure the app in the project Tools section before linguists can start using it directly in the Crowdin Editor.

Screenshots