x-algorithm
Algorithm powering the For You feed on X
Discover how visibility filtering in the X Algorithm acts as a deterministic gatekeeper, evaluating safety and context to control post visibility before ranking. Learn its crucial role.
How VFFilter Works After Candidate Selection in the Home-Mixer PipelineDiscover how VFFilter operates post candidate selection in the Home-Mixer Pipeline. Learn how it filters posts based on visibility actions and labels like Drop or Tombstone.
What Is PreviouslySeenPostsFilter? Timeline Deduplication in X-AlgorithmThe PreviouslySeenPostsFilter in X-Algorithm deduplicates timelines by removing already seen posts with `seen_ids` and Bloom filters. Discover how it keeps your feed fresh.
How the AgeFilter Works in Pre-Scoring: X-Algorithm Technical GuideLearn how the AgeFilter works in xai-org/x-algorithm pre-scoring. This technical guide explains JAX boolean masks and feature zeroing for user eligibility.
What Types of Filters Are Applied Before Scoring in x-algorithmDiscover x-algorithm pre-scoring filters like TaskFilter used in UPA, spam detection, and PTOS to ensure post eligibility before computational scoring.
What Is phoenix-rankall in X-Algorithm? Purpose and ArchitectureDiscover phoenix-rankall within X-Algorithm. Learn how this Rust micro-service generates and indexes ranking signals for the Phoenix recommendation system using Kafka events and time-windowed pipelines.
How the Candidate Pool Is Merged and Ranked in X-AlgorithmDiscover how X-Algorithm merges candidate pools by concatenating vectors and ranks them using a configurable Selector trait. Learn about score sorting, truncation, and diversity reranking.
What Is SimClusters in x-Algorithm Candidate Generation? A Technical Deep DiveExplore SimClusters in x-algorithm candidate generation. Learn how these sparse semantic embeddings map tweets and users to weighted cluster IDs for scalable retrieval from billions of items.
How Phoenix Retrieval Finds Candidate Posts: Inside the X Algorithm Corpus SystemDiscover how Phoenix retrieval finds candidate posts using a two-stage pipeline with RetrievalDataset and World.sample_candidate_posts for efficient post selection.
How Thunder Provides Candidate Posts for the Home-MixerLearn how Thunder provides candidate posts to the Home-Mixer by converting gRPC responses into PostCandidate objects for efficient downstream ranking.
What Are the Two Main Candidate Sources for the X For You Feed?Discover the two main candidate sources powering the X For You feed: the Follow-Based Timeline and the Recommendation Engine Feed. Learn how your personalized content is curated.
How Weights Are Scaled in the X-Algorithm Scoring SystemLearn how weights are scaled in the X-Algorithm scoring system. Understand how configurable weights and predicted probabilities create final ranking scores for actionable insights.
Have a question about this repo?
These articles cover the highlights, but your codebase questions are specific. Give your agent direct access to the source. Share this with your agent to get started:
curl -s "https://instagit.com/install.md" 
Maintain an open-source project? Get it listed too →