A sparse lexical embedding model using SPLADE with CoCondenser and ensemble distillation for efficient retrieval.