how customizable is the orchestration pipeline for fine-tuning large language models?

Question

In Azure AI Foundry, how customizable is the orchestration pipeline for fine-tuning large language models? Can you integrate custom data preprocessing and evaluation steps, or are you limited to predefined components within the Foundry framework?

Answer 1

Hi there preeth preeth

Yes, Azure AI Foundry allows customization in orchestration pipelines for fine-tuning large language models. You can integrate custom data preprocessing steps and define your own evaluation metrics using code or prompt-based evaluators. The platform supports fine-tuning of models like Meta Llama and Mistral with adjustable training parameters. Its flexibility, especially through Prompt Flow, makes it well-suited for tailoring workflows to specific project needs.

If this helps kindly acccept the answer thanks