Entity trail

ColBERT

Source-backed findings, relationship evidence, citations, and briefing history from the public MindPattern archive.

Briefing refs

Findings

Edges

Sources

Corpus findings

2026-07-30 / arxiv-researcherDenseOn and LateOn: Fully Open 149M Retrievers Set Size-Class SOTA at 56.20 and 57.22 nDCG@10 on BEIRarXiv 2607.27178 (2026-07-29) releases an end-to-end open recipe against the closed-training-data reproducibility gap: 665M curated English contrastive pre-training pairs distilled from 1.4B across 34 public sources, plus 1.88M supervised fine-tuning pairs with mined hard negatives. The result is two 149M-parameter models — DenseOn (single-vector) and LateOn (ColBERT-style late interaction) — at 56.20 and 57.22 average nDCG@10 on BEIR, new state of the art for that size class, with models, datasets, and training code released. The multilingual finding is the useful one: after translating to eight languages (2.8B pairs) into 307M mmBERT-based variants, the dense model degrades outside translate-train support while late interaction generalizes to unseen languages and scripts.
2026-07-13 / skill-finderUse ColBERT-style late-interaction retrieval for identifier-heavy code search instead of dense + cross-encoder rerankFor queries dominated by function names, variable names, and other identifiers, late-interaction retrieval (ColBERT-style, storing per-token embeddings) often wins outright over the standard dense-retrieve-then-cross-encoder-rerank pipeline, delivering near cross-encoder accuracy at near bi-encoder speed. In 2026 retrieval — not generation — is the RAG bottleneck, so the retrieval primitive choice is where the accuracy is won. Actionable: benchmark a late-interaction index against your two-stage rerank pipeline specifically on symbol/identifier queries before defaulting to rerank.
2026-06-24 / skill-finderBolt ColBERT late-interaction reranking onto a cheap first-stage retriever for precision without the costInstead of paying for a heavy dense model across your whole corpus, keep a fast retriever (BM25 or a dense index) to pull 50–200 candidates, then rerank only those with ColBERT's late-interaction MaxSim — each query token matches its best document token, summed for the score — so latency stays lean while quality jumps. ColBERTv2's aggressive quantization shrinks per-token vectors from 256 bytes to ~36 bytes (2-bit) or 20 bytes (1-bit), making token-level storage practical on commodity hardware. This two-stage design is the production-ready middle ground between BM25 speed and cross-encoder accuracy.
2026-06-11 / skill-finderGet near cross-encoder accuracy at bi-encoder speed with ColBERT late interaction (MUVERA + RAGatouille)ColBERT stores a per-token embedding for every query and document token and scores relevance as the sum of each query token's max similarity (MaxSim) against the document — so phrases match the specific passage where they're relevant instead of competing in one pooled vector. The 2026 production recipe uses RAGatouille for indexing/search and a MUVERA+ approach: fast single-vector candidate generation followed by exact ColBERT MaxSim reranking over the top-K. The tradeoff is storage — multi-vector indexes are an order of magnitude larger — so apply it as a reranker, not the primary index.
2026-05-01 / agents-researcherTACHIOM: Token-Aware Multivector Retrieval Achieves 247x Faster Clustering and 9.8x Search Speedup (SIGIR 2026)Accepted at SIGIR 2026, TACHIOM introduces token-aware centroid allocation that accounts for token distribution rather than treating all tokens equally, combined with graph-based indexing over centroids and product quantization for final scoring. The system achieves up to 247x faster clustering than k-means and up to 9.8x retrieval speedup over state-of-the-art systems while maintaining comparable or superior effectiveness on MS-MARCO and LoTTE benchmarks. For builders running RAG systems with ColBERT-style multivector models, this addresses the primary scaling bottleneck.
2026-04-09 / rss-researcherDatabricks Co-Founder Matei Zaharia Wins $250K ACM Prize in Computing, Says 'AGI Is Here Already'Matei Zaharia, creator of Apache Spark and CTO of Databricks, won the 2026 ACM Prize in Computing ($250K, Infosys-endowed) for his contributions to distributed data systems enabling large-scale ML and AI. In his TechCrunch interview he declared 'AGI is here already — it's just not in a form that we appreciate,' arguing the field should stop applying human standards to AI models. Zaharia's work spans Spark, Delta Lake, MLflow, Dolly, and ColBERT.

Graph relationships

BENCHMARKED AGAINST
TACHIOM
TACHIOM achieves 247x faster clustering and 9.8x retrieval speedup.
Source finding
BENCHMARKED AGAINST
TACHIOM -> ColBERT
TACHIOM achieves 247x faster clustering and 9.8x retrieval speedup.
Source finding

Source trail

arXiv 2607.27178 source RecSys Substack — Late-Interaction Reranking ML Journey — ColBERT and Late Interaction Retrieval arXiv (SIGIR 2026)TechCrunch

Graph sources

entity graphfindings textkg edgeskg entitiesnewsletter issues

Fetching from the wire…