Data collection for LLMs#

Argilla Feedback is purpose-built to support customized and multifaceted feedback in Large Language Model (LLM) projects. Serving as a critical tool for LLM fine-tuning and Reinforcement Learning from Human Feedback (RLHF), Argilla Feedback provides a flexible platform essential for the evaluation and fine-tuning stages of LLMs tailored to unique use cases. The figure below encapsulates the stages detailed in these guides:

LLM fine-tuning stages

For a practical, hands-on introduction, you can dive straight into our How-to Guides or Examples section. Alternatively, get started exploring one of the guides below:

Data collection for RLHF

Introduction to data collection for LLMs and RLHF.

Collecting demonstration data

Introduction to data collection for supervised and instruction fine-tuning.

Collecting comparison data

Introduction to data collection for reward modeling.