Pringled

3 ms·

1. 1.▲
   Show HN: Hardware-Friendly Text Classification with Model2Vec(github.com)

   3 points by Pringled 2y ago|0 comments
2. 2.▲
   by Pringled 2y ago

   Thanks for the kind words! Unfortunately, it's hard to directly measure the precision/recall (or other metrics) since there are no real labels. This is one of the reasons we tried to design this in a way that's as explainable
3. 3.▲
   by Pringled 2y ago

   Thanks David, that's really nice to hear!
4. 4.▲
   by Pringled 2y ago

   Thank you, that's nice to hear!
5. 5.▲
   Show HN: SemHash – Fast Semantic Text Deduplication for Cleaner Datasets(github.com)

   19 points by Pringled 2y ago|6 comments
6. 6.▲
   by Pringled 2y ago

   Some backends/algorithms don't natively support dynamic inserts, and require you to rebuild your index when you want to add vectors to it (Annoy and Pynndescent are the only backends that don't support it). Hybrid search is a
7. 7.▲
   by Pringled 2y ago

   Thanks! This is actually something that we have been experimenting with a bit already (auto-tuning on a specific dataset basically). It turned out to be quite complicated given how many index and parameter combinations you get with a grid-s
8. 8.▲
   by Pringled 2y ago

   1: that could be something for the future, but at the moment this is just meant as a way to quickly try out and evaluate various algorithms and libraries without having to learn the syntax for them (we call those backends). 2: we adopted th
9. 9.▲
   Show HN: Vicinity – Fast, Lightweight Nearest Neighbors with Flexible Back Ends(github.com)

   57 points by Pringled 2y ago|8 comments
10. 10.▲
    by Pringled 2y ago

    I think a combination works quite well: first getting a small set of candidates from all the data using a lightweight model, and the using a heavy-duty model to rerank the results and get the final candidates.
11. 11.▲
    Show HN: Model2vec – Lightning-fast Static Embeddings for RAG/Semantic Search(github.com)

    28 points by Pringled 2y ago|4 comments