generall

5 ms·

1.
Guidelines for Contributing with AI(github.com)
2 points by generall 7mo ago|0 comments
2.
by generall 2y ago
In this scenario posting list have to share all the same attributes as each element inside the list.
3.
by generall 2y ago
In IVF you can start checking conditions only in the final bucket. There are no guarantees if the bucket has any acceptable value, and there are no procedures to find the bucket which has acceptable values before scanning it.
4.
by generall 2y ago
IVF, unfortunately, is barely compatible with filtered search. It have to rely on post-filtering and retrieve more and more candidates if the result set is not big enough. If the query is in some form correlated with the filter, this approa
5.
by generall 3y ago
https://qdrant.tech/articles/dedicated-service/ - we have some arguments on this
6.
by generall 3y ago
This article contains a lot of inaccuracies. Based on your statements, like > Qdrant stores both the vectors and the metadata in a sqlite database. It looks like you have benchmarked local mode of qdrant. It doesn't even use vector
7.
by generall 3y ago
With a few optimization tricks, TL;DR: - ONNX inference in Rust - Embeddings cache & lookup - Parallel & Batch requests - hybrid search with full-text filtering + vector re-scoring
8.
Vector Search-as-You-Type in Rust(qdrant.tech)
5 points by generall 3y ago|1 comments
9.
Semantic Search-as-You-Type(qdrant.tech)
16 points by generall 3y ago|2 comments
10.
by generall 3y ago
I would say that text text and vector search are orthogonal. Some scenarios are better with one, others with combination. But fitting vector search into the interface designed for text is limiting vector search potential
11.
Vector Similarity Beyond Search(qdrant.tech)
39 points by generall 3y ago|11 comments
12.
Vector Search in Constant Time(qdrant.tech)
3 points by generall 3y ago|0 comments
13.
by generall 4y ago
Benchmark is open source - https://github.com/qdrant/vector-db-benchmark You are welcome to fork it and make your own measurements, if you suspect something. Benchmarks like https://big-ann-benchmarks.com&#x
14.
by generall 4y ago
> Please expand on how you are defining full text search distinctly from sparse search to continue the discussion In addition to the indexing algorithm, there is the tokenizer, which depends on the language, lemmatizer, synonyms, stop-wo
15.
by generall 4y ago
> There are a few arguments why adding sparse search doesn't require too much extra specialization Full-text search != sparse search, that's a naive oversimplification. Btw, sparse search is on Qdrant roadmap, so we should be a
16.
by generall 4y ago
Proposed architecture doesn't limit you to use self-hosted transformers only, you can use OpenAI just as easily. And you don't need a to install a "module" for that
17.
by generall 4y ago
The latency of the combination of parallel systems is equal to the slowest component. And obviously, specialized tools will be faster than a component of a multi-tool system cause while dedicated engines can invest in optimizing specific fu
18.
by generall 4y ago
There is more relevant benchmark of vector search engines end-to-end, not just algorithms: https://qdrant.tech/benchmarks/
19.
Which vector search engine is the fastest?(qdrant.tech)
13 points by generall 4y ago|2 comments
20.
by generall 4y ago
PML is a great collection of implementations, but not the best framework. Also you can use PML with Quaterion: https://github.com/qdrant/quaterion/blob/master/examples/tra...
21.
by generall 4y ago
Starting from 0.5.0 finetuner is no longer an open-source. > From 0.5.0, Finetuner computing is hosted on Jina Cloud. The last local version is 0.4.1, one can install it via pip or check out git tags/releases here. But there are som
22.
by generall 4y ago
It is definitely possible to do, if you have a proper training set. You would need to somehow give model a signal, that you are interesting in e.g. arguing style specifically, and not the topic of the text.
23.
by generall 4y ago
The idea is to fine-tune model, which can be used to produce embeddings required for Qdrant. Our design approach is to make things as modular as possible, so you can use framework and engine independently. But we are working on integrating
24.
Similarity Learning lacks a framework. So we built one(blog.qdrant.tech)
59 points by generall 4y ago|24 comments
25.
Show HN: Quaterion – x100 faster fine-tuning of similarity learning models(github.com)
19 points by generall 4y ago|1 comments
26.
by generall 4y ago
Check out https://github.com/qdrant/qdrant
27.
by generall 5y ago
I wounder if there is any alternative implementations of ScaNN in languages like Rust.
28.
by generall 6y ago
Is there a reason why this is not accompanied by Hans Zimmer?
29.
RP Flip: Ship Turned Research Platform Which Sinks for a Cause(ststworld.com)
1 points by generall 6y ago|0 comments
30.
by generall 6y ago
I built a very simple tool which generates exercises for any given set of option words. Source code can be found here https://github.com/generall/Exerciser

More ›