Overview
An organization’s data is often messy or incomplete, synthetic data can help fill in the gaps and ensure data quality meets model training standards. Generating accurate, high-quality synthetic data at scale is a core function of the Gretel platform. By leveraging Gretel, you can create domain-specific data tailored to fine-tune your models.
Gretel Navigator
Gretel Navigator is the world’s first compound AI system designed to create high-quality tabular datasets using natural language or code. With Navigator, you can iteratively create, edit, and augment tabular datasets with just a simple prompt.
Resources
Gretel Tabular Fine-Tuning
Gretel Tabular Fine-Tuning (formerly known is Navigator Fine Tuning) is an AI system that combines a Large-Language Model pre-trained on tabular datasets with schema-based rules. It can train on datasets of various sizes and generate synthetic datasets with unlimited records. Navigator Fine Tuning excels at matching correlations (both within a single record and across multiple records) and distributions in its training data across various tabular modalities, including numeric, categorical, free text, JSON, and time series values.
