What is Dioptra?
Dioptra is an open source data curation and management platform designed for computer vision, natural language processing (NLP), and large language models (LLMs). It helps users curate valuable unlabeled data, register metadata, diagnose model failure modes, and integrate with labeling and retraining stacks.
How to use Dioptra?
1. Curate the most valuable unlabeled data to improve domain coverage and model performance. 2. Register your metadata to Dioptra to ensure your data remains with you. 3. Diagnose root cause model failure modes and regressions using Dioptra's data centric toolkit. 4. Use active learning miners to sample the most valuable unlabeled data. 5. Integrate with your labeling and retraining stack using Dioptra's APIs.