The author defends e-means as a superior, drop-in replacement for Lloyd's algorithm that uses scaling laws to automatically tune hyperparameters and escape local minima faster. Critics argue the method relies on known techniques like exponential moving averages and mini-batches without sufficient novelty or rigorous comparison against well-tuned baselines like scikit-learn's implementation.
81 readable replies · Supporting reply 1 · Supporting reply 2 · Supporting reply 3 · Supporting reply 4facebookresearch
faiss
C++2017–2026
facebookresearch / faiss
faiss
A library for efficient similarity search and clustering of dense vectors.
45Distinct people
53Mentions
1Explicit recommenders
41.6KGitHub stars
What to know before exploring
What it is
A library for efficient similarity search and clustering of dense vectors.
GitHub
Where it is used
No documented product use was verified in the last source check.
What people are saying
This tutorial is very complex. Here's how to get free semantic search with much less complexity: 1. Install sentence-transformers [1] 2. Initialize the MiniLM model - `model =…
vikp used a recommendation phrase. One indexed comment matches. This is a phrase match, not a verdict.
No classified warning is attached to this page. Read the discussions before treating silence as approval.
Also in these discussions
- spotify/annoy · linked in 4 of the same discussions
- searchivarius/nmslib · linked in 2 of the same discussions
Mentioned in the same Hacker News threads. That is not proof they are substitutes.
This week on Hacker News
Nov 7–Nov 14 UTC · Direct repository links and separately labeled related project stories. Indexed through 11/14/2026, 3:56:07 PM UTC.
Mentions over time
HN mentions by year53 dated mentions
Observed mentions: 2017-03-02 20:35:13 UTC to 2026-04-10 23:12:49 UTC.
* An incomplete calendar period is marked with an outlined bar. Its count is not directly comparable with a full period.
Counts come from the repository’s dated source evidence. Index cutoff: 2026-11-02 08:16:59 UTC.
View mention counts and date ranges
| Period | Start (UTC) | End (exclusive, UTC) | Mentions | Submissions | Comments |
|---|---|---|---|---|---|
| 2017 | 2017-01-01 00:00:00 UTC | 2018-01-01 00:00:00 UTC | 2 | 1 | 1 |
| 2018 | 2018-01-01 00:00:00 UTC | 2019-01-01 00:00:00 UTC | 3 | 0 | 3 |
| 2019 | 2019-01-01 00:00:00 UTC | 2020-01-01 00:00:00 UTC | 5 | 0 | 5 |
| 2020 | 2020-01-01 00:00:00 UTC | 2021-01-01 00:00:00 UTC | 1 | 0 | 1 |
| 2021 | 2021-01-01 00:00:00 UTC | 2022-01-01 00:00:00 UTC | 8 | 1 | 7 |
| 2022 | 2022-01-01 00:00:00 UTC | 2023-01-01 00:00:00 UTC | 3 | 0 | 3 |
| 2023 | 2023-01-01 00:00:00 UTC | 2024-01-01 00:00:00 UTC | 20 | 1 | 19 |
| 2024 | 2024-01-01 00:00:00 UTC | 2025-01-01 00:00:00 UTC | 8 | 1 | 7 |
| 2025 | 2025-01-01 00:00:00 UTC | 2026-01-01 00:00:00 UTC | 1 | 0 | 1 |
| 2026 (partial) | 2026-01-01 00:00:00 UTC | 2027-01-01 00:00:00 UTC | 2 | 2 | 0 |
What Hacker News says
53 sourcesWEwewewedxfgdf·Apr 10, 2026Submission
Faiss a library for efficient similarity search and clustering of dense vectors
2 points · Open discussionValori – Deterministic Substrate for AI (Code and ArXiv Paper)
I've published a paper on a fundamental problem in AI systems that I believe has significant implications for our work: "Valori: A Deterministic Memory Substrate for AI Systems." [https://arxiv.org/pdf/2512.22280]
THE PROBLEM :
AI systems have unreliable memory.
When you run the same AI agent twice with the same input, it can give you different answers not because the AI is "thinking differently," but because the underlying memory system is non-deterministic.
In simple terms, the math used in building current vector databases (FAISS, Qdrant, Milvus, Pinecone, etc.) is floating-point arithmetic, which behaves differently across CPUs (x86, ARM, Mac, Windows). This is a universal problem acknowledged by industry leaders, including the
FAISS team: https://github.com/facebookresearch/faiss/issues/4739
THE SOLUTION (What I Built) :
Valori: A d
Read the rest
eterministic memory substrate for AI systems.
In simple terms, I replaced floating-point math with fixed-point arithmetic (Q16.16) and built the memory engine using Rust (no_std).
Traditional systems store state (the answer). Valori stores events (how we got the answer).
This means you can:
Replay any decision exactly, Debug across different machines, Audit with mathematical certainty, Prove compliance to regulators
USE CASES (Where This Matters)
Simply put, wherever we need bit-level replay, Valori can be applied:
Compliance & Audit (Financial Services), Incident Investigation (SRE/DevOps), Multi-Agent Systems (Autonomous Operations), Quantitative Finance (Hedge Funds), AI Safety & Alignment (Research), Healthcare AI (Medical Decisions)
I'd be happy to discuss this further and explore potential applications. Open to answer any doubts, questions, Criticism and Feedback
why not use this? https://github.com/facebookresearch/faiss
In Vector database that can index 1B vectors in 48MIs My Approach to Vectorizing and Storing 1.5 Trillion Tokens Reasonable?
I'm planning to index and store 1.5 trillion tokens using Faiss and would love some feedback on my approach:
1. Partitioning: I'm thinking of using distributed k-means and inverted multi-index quantizers for efficient data partitioning.
2. On-Disk Storage: Due to the scale, I'm storing everything on disk using a Compressed Sparse Row format.
3. Distributed Search: I plan to implement a client-server model with multiple servers to handle search operations.
Does this approach sound feasible, or am I overlooking something crucial? Any advice or suggestions?
I'm mostly working off of this article: Indexing 1T Vectors (https://github.com/facebookresearch/faiss/wiki/Indexing-1T-vectors). I think the data is too big for AutoFaiss, but I can use that for experiments.
1 points · Open discussionhow different is this compared to Facebook's open-source tool Faiss[1]?
[1] https://github.com/facebookresearch/faiss/
In GraphRAG is now on GitHubBinary, quantize each dimension to +1 or -1
You can try out binary vectors, in comparison to quantize every pair of vectors to one of four values, and a lot more, by using a FAISS index on your data, and using Product Quantization (like PQ768x1 for binary features in this case) https://github.com/facebookresearch/faiss/wiki/The-index-fac...
In I’m writing a new vector search SQLite ExtensionThere are also FAISS binary indexes[0], so it'd be great to compare binary index vs binary index. Otherwise it seems a little misleading to say it is a FAISS vs not FAISS comparison, since really it would be a binary index vs not binary index comparison. I'm not too familiar with binary indexes, so if there's a significant difference between the types of binary index then it'd be great to explain what that is too.
[0] https://github.com/facebookresearch/faiss/wiki/Binary-indexe...
In Haystack DB – 10x faster than FAISS with binary embeddings by defaultQdrant lib project https://github.com/tyrchen/qdrant-lib, Qdrant SDK has also support for local mode, which means embeddable https://github.com/qdrant/qdrant-client
Or just use FAISS https://github.com/facebookresearch/faiss
In Show HN: Chromem-go – Embeddable vector database for GoSingle-GPU, optimal efficiency: unsloth + qlora + mistral-7b on runpod/vast/lambda
Blazing fast compared to out-of-the-box transformers, also make sure to use flash attention if you have A100s or better and context length >= 2k
Add FAISS (https://github.com/facebookresearch/faiss) if you need fast local RAG
In Ask HN: Most efficient way to fine-tune an LLM in 2024?https://github.com/facebookresearch/faiss/wiki/Vector-codecs has some good things available too.
In New embedding models and API updatesInteresting project. Would be nice to see how it compares to the built-in operations provided in Faiss.
https://github.com/facebookresearch/faiss/wiki/How-to-make-F...
In SimSIMD: Hardware-accelerated similarity metrics and distance functionsYou can try txtai (https://github.com/neuml/txtai) with a Faiss backend.
This Faiss wiki article might help (https://github.com/facebookresearch/faiss/wiki/Indexing-1G-v...).
For example, a partial Faiss configuration with 4-bit PQ quantization and only using 5% of the data to train an IVF index is shown below.
faiss={"components": "IVF,PQ384x4fs", "sample": 0.05}
In Investing in new vector database development vs enhancing existing databasesIf you want to experiment with vector stores, you can do that locally with something like faiss which has good multiplatform support and sufficient tutorials: https://github.com/facebookresearch/faiss
Doing full retrieval-augmented generation (RAG) and getting LLMs to interpret the results has more steps but you get a lot of flexibility, and despite what AI influencers say there's no standard best-practice. When you query a vector DB you get the most similar texts back (or an index integer in the case of faiss), you then feed those result to an LLM like a normal prompt, which can be optimized with prompt engineering.
The codifer for the RAG workflow is LangChain, but their demo is substantially more complex and harder-to-use than even a homegrown implementation: https://minimaxir.com/2023/07/langchain-problem/
In Approximate Nearest Neighbor Oh Yeah (Annoy)Indexes for vector databases in high dimensions are nowhere near are effective as the 2-d indexes used in GIS or the 1-d B-tree indexes that are commonly used in databases.
Back around 2005 I was interested in similarity search and read a lot of conference proceedings on the top and was basically depressed at the state of vector database indexes and felt that at least for the systems I was prototyping I was OK with a full scan and later in 2013 I had the assignment of getting a search engine for patents using vector embeddings in front of customers and we got performance we found acceptable with full scan.
My impression today is that the scene is not too different than it was in 2005 but I can't say I haven't missed anything. That is, you have tradeoffs between faster algorithms that miss some results and slower algorithms that are more
Read the rest
correct. Somebody might feel the glass is half full somebody else might feel it is half empty, I can really see it either way.
I think it's already a competitive business. You have Pinecone which had the good fortune of starting before the gold rush. Many established databases are adding vector extension. I know so many engineering managers who love postgresql and they're just going to load a vector extension and go. My RSS reader YOShInOn uses SBERT embeddings to cluster and classify text and certainly More Like This and semantic search are on the agenda, I'd expect it to take about an hour to get
https://github.com/facebookresearch/faiss
up and working, I could spend more time stuck on some "little" front end problem like getting something to look right in Bootstrap than it would take to get working.
I can totally believe somebody could make a better vector db than what's out there but will it be better enough? A startup going through YC now could spend 2-3 to get a really good product and find customers and that is forever in a world where everybody wants to build AI applications right now.
Are folks typically using HNSW for vector search these days? I thought maybe ScaNN has proven to be better? Especially since it's available in FAISS [2].
[1] https://ai.googleblog.com/2020/07/announcing-scann-efficient...
[2] https://github.com/facebookresearch/faiss/wiki/Fast-accumula...
In USearch: Smaller and faster single-file vector search engineShowing the 15 most recent of 53 sources. Enable JavaScript to search and browse every indexed mention.