deepfabric-generator

To generate a fine-tuning dataset from a topic, provide a descriptive sentence to the generator. It interactively configures parameters like knowledge graph depth and degree, then outputs a structured dataset.jsonl file and a _train.py script.

Yes, you can use Gemini or Anthropic as the LLM provider to create synthetic training data. The dataset generation process interactively prompts you to select your preferred LLM provider for building the knowledge graph and dataset.

The best way to automate dataset generation for LLM fine-tuning is using a framework that builds a knowledge graph from your topic. This approach automatically transforms a single sentence into a structured dataset.jsonl file and accompanying training script.

Yes, the generated dataset supports tool-use scenarios with mock files. During the interactive configuration, you can integrate optional mock files to simulate tool-use cases within the output dataset.jsonl file.

You configure knowledge graph depth and degree through an interactive configuration process. After providing your topic, the generator guides you through defining these specific dataset parameters to control the synthetic data generation.

After generating a fine-tuning dataset, you get a structured dataset.jsonl file containing the synthetic data and a _train.py script. These outputs are immediately ready for use in your model fine-tuning pipeline.