Expand description
Semantic Cluster Manager — online k-means-style document clustering over embeddings.
Manages semantic embedding clusters, assigns documents to clusters, updates centroids incrementally via exponential moving average, and detects cluster drift over time.
Structs§
- Batch
Cluster - A cluster produced by batch k-means.
- Batch
Cluster Config - Configuration for batch k-means clustering.
- Batch
Cluster Manager Stats - Aggregate statistics for
BatchSemanticClusterManager. - Batch
Semantic Cluster Manager - Batch k-means semantic cluster manager.
- Cluster
Assignment - Assignment of a document to a cluster.
- Cluster
Manager Config - Configuration for
SemanticClusterManager. - Cluster
Manager Stats - Aggregate statistics over all clusters managed by
SemanticClusterManager. - Semantic
Cluster - A single semantic cluster with a centroid maintained via incremental updates.
- Semantic
Cluster Manager - Online k-means-style semantic cluster manager.
Functions§
- euclidean_
distance - Compute Euclidean distance between two vectors of equal length.
- vec_
mean - Compute element-wise mean of a set of vectors.