Show HN: Pyversity – Fast Result Diversification for Retrieval and RAG
github.comHey HN! I’ve recently open-sourced Pyversity, a lightweight library for diversifying retrieval results. Most retrieval systems optimize only for relevance, which can lead to top-k results that look almost identical. Pyversity efficiently re-ranks results to balance relevance and diversity, surfacing items that remain relevant but are less redundant. This helps with improving retrieval, recommendation, and RAG pipelines without adding latency or complexity.
Main features:
- Unified API: one function (diversify) supporting several well-known strategies: MMR, MSD, DPP, and COVER (with more to come)
- Lightweight: the only dependency is NumPy, keeping the package small and easy to install
- Fast: efficient implementations for all supported strategies; diversify results in milliseconds
Re-ranking with cross-encoders is very popular right now, but also very expensive. From my experience, you can usually improve retrieval results with simpler and faster methods, such as the ones implemented in this package. This helps retrieval, recommendation, and RAG systems present richer, more informative results by ensuring each new item adds new information.
Code and docs: github.com/pringled/pyversity
Let me know if you have any feedback, or suggestions for other diversification strategies to support!
The biggest problem with retrieval is actually semantic relevance. I think most embedding models don't really capture sentence-level semantic content and instead act more like bag-of-words models averaging local word-level information.
Consider this simple test I’ve been running:
Anchor: “A background service listens to a task queue and processes incoming data payloads using a custom rules engine before persisting output to a local SQLite database.”
Option A (Lexical Match): “A background service listens to a message queue and processes outgoing authentication tokens using a custom hash function before transmitting output to a local SQLite database.”
Option B (Semantic Match): “An asynchronous worker fetches jobs from a scheduling channel, transforms each record according to a user-defined logic system, and saves the results to an embedded relational data store on disk.”
Any decent LLM (e.g., Gemini 2.5 Pro, GPT-4/5) immediately knows that the Anchor and Option B describe the same concept just with different words. But when I test embedding models like gemini-embedding-001 (currently top of MTEB), they consistently rate Option A as more similar measured by cosine similarity. They’re getting tricked by surface-level word overlap.
I put together a small GitHub repo that uses ChatGPT to generate and test these “semantic triplets:
https://github.com/semvec/embedstresstest
gemini-embedding-001 (current #1 on MTEB leaderboard ) scored close to 0% on these adversarial examples.
The repo is unpolished at the moment but it gets the idea across and everything is reproducible.
Anyway, did anyone else notice this problem?
Producing a diverse list of results may still help in a couple of ways here.
* If there are a lot of lexical matches, real semantic matches may still be in the list but far down the list. A diverse set of, say, 20 results may have a better chance of including a semantic match than the top 20 results by some score.
* There might be a lot of semantic matches, but a vast majority of the semantic matches follow a particular viewpoint. A diverse set of results has a better chance of including the viewpoint that solves the problem.
Yes, semantic matching is important, but this is solving an orthogonal and complementary problem. Both are important.
I’m not sure what the “biggest” problem is, but I do think diversity is vastly underappreciated compared to relevance.
You can have maximally relevant search results that are horrible. Because most users (and LLMs) want to understand the range of options, not just one type of relevant option.
Search for “shoes” and only see athletic shoes is a bad experience. You’ll sell more shoes, and keep the user engaged, if you show a diverse range of shoes.
This seems like a good template to generate synthetic data, with positive/negative examples, allowing an embedding model to be aligned more semantically to underlying concepts.
Anyways, I'd hope reranking models do better, have you tried those?
Might also be useful for dataset curation, or even just prompt engineering. For example when training a classification task and picking a diverse set of examples for training or evaluation.
True, I think that's also a great usecase! Though these algorithms likely won't scale to very large datasets (e.g. millions of samples), but for smaller datasets, like fine-tuning sets, I think this would work very well. I've worked on something similar in the past that works for larger datasets (semantic deduplication: https://github.com/MinishLab/semhash).
Fascinating!