Hierarchical Pooling: Compression for Late-Interaction Retrieval Models
antoine_chaffin · x · 2026-07-07
Background: The author previously proposed "hierarchical pooling" to compress the storage of late-interaction retrieval models (like ColBERT variants) by performing hierarchical clustering on document embeddings and iteratively merging the closest clusters. This method halves storage losslessly and can be generalized across multiple modalities. Blog and paper links are included.
More from Research
- Lab monkeys at $30-80K each: a 4-year breeding cycle is the real preclinical drug bottleneck — zebird0 · 2026-10-27
- OpenAI Foundation funds UNC tumor antigen project, hiring senior experimental scientist — iskander · 2026-10-27
- Steering attention heads toward honesty: up moral patienthood, down psychopathy — edelwax · 2026-10-27
- Are transcoders overkill? New paper argues plain neurons suffice for circuit analysis — burny_tech · 2026-10-27
- EMNLP paper GENIE: a fine-grained metric locating where LLM stories are novel — gregd_nlp · 2026-10-27
- Hashed n-gram value embeddings push GPT to 3.2115 loss, beating 30M track record — RobertTLange · 2026-10-27