X Open-Sources Grok-Based Algorithm for ‘For You’ Feed
- Social media platform X has released its Grok-based transformer model, which ranks ‘For You’ feed posts by predicting user actions without hand-engineered features.
- The GitHub repository includes all code determining organic and advertising post visibility, built in Rust and Python for modular retrieval and scoring.
- Elon Musk’s pledge to update the algorithm every four weeks is accompanied by comprehensive developer notes.
- The algorithm retrieves content from in-network posts users follow and out-of-network posts discovered through ML-based retrieval, combining both through a scoring system that predicts engagement probabilities.
- Midhun Krishna M, co-founder of TknOps.io, stated that this open-source release could change industry standards by forcing other platforms to follow suit or explain their lack of transparency.
X’s release of the Grok-based transformer model marks a significant move towards transparency in social media algorithms, potentially setting new industry standards for openness in recommendation systems. This initiative allows developers to understand and improve upon these systems while providing users with insights into what makes posts go viral.
By exposing its machine learning architecture, X is encouraging innovation and accountability within the tech community, as developers can now access a blueprint previously kept secret by many platforms. Source
