The launch of our open-source Radicalbit AI monitoring platform last June marked a significant milestone for our team. This was just the beginning of our open source journey, and since then we’ve been relentlessly innovating.

After weeks of dedicated development, we’re excited to introduce the release of Radicalbit AI monitoring platform V1.0.0. Packed with enhanced features, this update is designed to further improve and streamline the way you can monitor Machine Learning and Large Language models.

Here are the new key features

• Residual Analysis for Regression Model Task: we added this statistical method to assess the reliability of regression models by determining the difference between actual and predicted values.
• PSI Algorithm for Drift Detection: the PSI metric has been added to our existing drift detection toolkit, which already includes the Kolmogorov-Smirnov and Chi-squared tests.
• Improved Drift detection: we enhanced the robust drift detection mechanism that automatically selects the most appropriate algorithm for each specific situation
• Log loss metric for binary classification: we added log loss as a metric to evaluate the performance of binary classification models.
• Users can remove features for which metrics are pointless: users are now able to upload data and selectively exclude features from analysis by directly flagging them, eliminating the need for prior data preparation.
• Users can change field type, as certain numerical features can be the encoding of categorical ones: users can specify numerical features that should be treated as categorical. The software will then perform appropriate categorical analysis on these specified features.

The release of this version comes with extended documentation to further assist users in deploying and utilizing the Radicalbit AI Monitoring platform.

What drove us to start the open source project is the possibility to directly collaborate with the AI community, collaboratively developing a platform to enhance AI-based projects and ensure trustworthy AI. With the release of the 1.0.0 version, we are enthusiastic to share such accomplishment that is testament to the dedication of our team and the AI community at large.

But what we are most excited about is to continue this journey with data enthusiasts, developers and whomever will decide to contribute to this ambitious project.

Visit our GitHub repository to learn more and experience the power of Radicalbit AI Monitoring V1.0.0!