2 Commits (ca409366cb3fc0b5dbe85a8750c33ec88d4eebda)

Author	SHA1	Message	Date
Rene Cannao	cbf27eb607	Add vec0 KNN LIMIT constraint documentation for Posts embeddings ... Add documentation for common error "A LIMIT or 'k = ?' constraint is required on vec0 knn queries" with corrected query examples. Changes: 1. Add Example 3 showing correct LIMIT placement for "What is ProxySQL?" query 2. Include alternative syntax using 'k = ?' constraint 3. Explain key rules for vec0 KNN queries: - LIMIT or k = ? must be in same query level as MATCH - Cannot use both constraints together - When joining, put MATCH + LIMIT in subquery - Constraint tells sqlite-vec how many similar vectors to return	4 months ago
Rene Cannao	8e83635760	Add Posts embeddings setup documentation with optimized batch processing ... This documentation provides step-by-step instructions for setting up virtual tables for Posts embeddings using sqlite-rembed and sqlite-vec. Key features: - Virtual table creation for 768-dimensional embeddings - API client configuration for embedding generation - Optimized batch query using LEFT JOIN to process unembedded rows - Batch processing script with progress tracking - Similarity search examples - Performance considerations and troubleshooting The batch query uses a LEFT JOIN pattern to find unprocessed rows: INSERT ... SELECT ... FROM Posts LEFT JOIN Posts_embeddings WHERE Posts_embeddings.rowid IS NULL LIMIT 1000; This approach eliminates the need for tracking rowid ranges and can be run repeatedly until all 248,905 Posts have embeddings generated.	4 months ago