perone

10 ms·

1. 61.▲
   by perone 8y ago

   There is a lot going on in this field right now, take a look at the ONNX ecosystem ( https://onnx.ai/ ).
2. 62.▲
   by perone 8y ago

   We'll be integrating other indexing in near future (such as faiss), Annoy is just one option for indexing that was implemented. Each indexing method will have their pros/cons, so you'll be able to select the search engine bac
3. 63.▲
   by perone 8y ago

   If you don't know ML and you want to do similarity search, you can just use the models that come together with EuclidesDB and just make calls to add items and query for similarity with less than 10 lines of Python code. So it will real
4. 64.▲
   by perone 8y ago

   Thanks for the feedback !
5. 65.▲
   by perone 8y ago

   Just an extra note, that is the key point why EuclidesDB support multiple models, so you can have for instance a ResNet trained on ImageNet for some images and another ResNet (same architecture) fine-tuned on your data (domain adapted) for
6. 66.▲
   by perone 8y ago

   Hi, you're correct, LevelDB is the lower level building block for EuclidesDB, it's the underlying storage for item's features/predictions/metadata. EuclidesDB uses gRPC (reasons for that design decision are describe
7. 67.▲
   by perone 8y ago

   Thanks for posting it here drewvolpe !
8. 68.▲
   by perone 8y ago

   Hi, thanks for the feedback, I'm the creator of the framework. Answering your question: I chose LevelDB because it offers a good transactional capability for the key-value store, it also provides a very easy-to-use embedded library tha
9. 69.▲
   EuclidesDB, a multi-model machine learning feature db integrated with libtorch(euclidesdb.readthedocs.io)

   1 points by perone 8y ago|0 comments
10. 70.▲
    Time Maps: visualizing discrete Twitter events from 2018 Brazilian elections(blog.christianperone.com)

    2 points by perone 8y ago|0 comments
11. 71.▲
    PyTorch 1.0 JIT and LibTorch C++ API to Integrate PyTorch into Node.js(blog.christianperone.com)

    24 points by perone 8y ago|0 comments
12. 72.▲
    PyTorch 1.0 Tracing JIT and LibTorch C++ API to Integrate PyTorch into Node.js(blog.christianperone.com)

    2 points by perone 8y ago|1 comments
13. 73.▲
    by perone 8y ago

    I wrote about it in the past if someone is interested in a tutorial: http://blog.christianperone.com/2015/08/googles-s2-geometry-...
14. 74.▲
    Evaluation of Sentence Embeddings: ELMo, FastText, GloVe, Google USE, InferSent(arxiv.org)

    4 points by perone 8y ago|0 comments
15. 75.▲
    Evaluation of sentence embeddings in downstream and linguistic probing tasks(arxiv.org)

    3 points by perone 8y ago|0 comments
16. 76.▲
    NLP word representations and the Wittgenstein philosophy of language(blog.christianperone.com)

    2 points by perone 8y ago|0 comments
17. 77.▲
    by perone 9y ago

    haha nice, in the end, we squeeze what we know. Didn't know about OmniGraffle, will take a look.
18. 78.▲
    by perone 9y ago

    Clever trick.
19. 79.▲
    by perone 9y ago

    I used Sketch ( https://www.sketchapp.com/ ), the graphs of the objects are actually from a built-in template for iOS notifications lol, but it worked well for that purpose.
20. 80.▲
    PyTorch – Internal Architecture Tour(blog.christianperone.com)

    193 points by perone 9y ago|22 comments
21. 81.▲
    by perone 9y ago

    I trained a word2vec model in the past using the EVA transcription if someone is interested: http://blog.christianperone.com/2016/01/voynich-manuscript-w...
22. 82.▲
    by perone 9y ago

    Thanks ! You can also have PSI with OT-based protocol, actually, I think that the OT-based protocol is one of the fastest for PSI. ENCRYPTO has a very nice library for it: https://github.com/encryptogroup/PSI if you ar
23. 83.▲
    Privacy-preserving sentence semantic similarity(blog.christianperone.com)

    68 points by perone 9y ago|5 comments
24. 84.▲
    Privacy-preserving sentence similarity with InferSent and secure 2-party comput(blog.christianperone.com)

    1 points by perone 9y ago|0 comments
25. 85.▲
    Google's S2, geometry on the sphere, cells and Hilbert curve(blog.christianperone.com)

    2 points by perone 9y ago|0 comments
26. 86.▲
    New prime on the block(blog.christianperone.com)

    1 points by perone 9y ago|0 comments
27. 87.▲
    The Effective Receptive Field (ERF) on Convolutional Neural Networks(blog.christianperone.com)

    1 points by perone 9y ago|0 comments
28. 88.▲
    by perone 9y ago

    Pytorch is way ahead of tensorflow in terms of cuDNN API usage.
29. 89.▲
    The same old historicism, now on AI(blog.christianperone.com)

    3 points by perone 9y ago|0 comments
30. 90.▲
    Nanopipe: a C++ framework to connect message systems(nanopipe.readthedocs.io)

    4 points by perone 10y ago|0 comments