Vector AI — A platform for building vector based applications. Encode, query and analyse data using vectors.

Overview


Release Website Documentation Discord


Vector AI is a framework designed to make the process of building production grade vector based applications as quickly and easily as possible. Create, store, manipulate, search and analyse vectors alongside json documents to power applications such as neural search, semantic search, personalised recommendations recommendations etc.


Features

  • Multimedia Data Vectorisation: Image2Vec, Audio2Vec, etc (Any data can be turned into vectors through machine learning)
  • Document Orientated Store: Store your vectors alongside documents without having to do a db lookup for metadata about the vectors.
  • Vector Similarity Search: Enable searching of vectors and rich multimedia with vector similarity search. The backbone of many popular A.I use cases like reverse image search, recommendations, personalisation, etc.
  • Hybrid Search: There are scenarios where vector search is not as effective as traditional search, e.g. searching for skus. Vector AI lets you combine vector search with all the features of traditional search such as filtering, fuzzy search, keyword matching to create an even more powerful search.
  • Multi-Model Weighted Search: Our Vector search is highly customisable and you can peform searches with multiple vectors from multiple models and give them different weightings.
  • Vector Operations: Flexible search with out of the box operations on vectors. e.g. mean, median, sum, etc.
  • Aggregation: All the traditional aggregation you'd expect. e.g. group by mean, pivot tables, etc
  • Clustering: Interpret your vectors and data by allocating them to buckets and get statistics about these different buckets based on data you provide.
  • Vector Analytics: Get better understanding of your vectors by using out-of-the-box practical vector analytics, giving you better understanding of the quality of your vectors.

Quick Terminologies

  • Models/Encoders (aka. Embedders) ~ Turns data into vectors e.g. Word2Vec turns words into vector
  • Vector Similarity Search (aka. Nearest Neighbor Search, Distance Search)
  • Collection (aka. Index, Table) ~ a collection is made up of multiple documents
  • Documents (aka. Json, Item, Dictionary, Row) ~ a document can contain vectors, text and links to videos/images/audio.

QuickStart

Install via pip! Compatible with any OS.

pip install vectorai

If you require the nightly version due to on-going improvements, you can install the nightly version using:

pip install vectorai-nightly

Note: while the nightly version will still pass automated tests, it may not be stable.

Check out our quickstart notebook on how to make a text/image/audio search engine in 5 minutes: quickstart.ipynb

from vectorai import ViClient, request_api_key

api_key = request_api_key(username=<username>, email=<email>, description=<description>, referral_code="github_referred")

vi_client = ViClient(username=username, api_key=api_key)

from vectorai.models.deployed import ViText2Vec
text_encoder = ViText2Vec(username, api_key)

documents = [
    {
        '_id': 0,
        'color': 'red'
    },
    {
        '_id': 1,
        'color': 'blue'
    }
]

# Insert the data
vi_client.insert_documents('test-collection', documents, models={'color': text_encoder.encode})

# Search the data
vi_client.search('test-collection', text_encoder.encode('maroon'), 'color_vector_', page_size=2)

# Get Recommendations
vi_client.search_by_id('test-collection', '1', 'color_vector_', page_size=2)

Access Powerful Vector Analytics

Vector AI has powerful visualisations to allow you to analyse your vectors as easily as possible - in 1 line of code.

vi_client.plot_dimensionality_reduced_vectors(documents, 
    point_label='title', 
    dim_reduction_field='_dr_ivis', 
    cluster_field='centroid_title', cluster_label='centroid_title')

View Dimensionality-Reduced Vectors

vi_client.plot_2d_cosine_similarity(
    documents,
    documents[0:2],
    vector_fields=['use_vector_'],
    label='name',
    anchor_document=documents[0]
)

Compare vectors and their search performance on your documents easily! 1D plot cosine simlarity


Why Vector AI compared to other Nearest Neighbor implementations?

  • Production Ready: Our API is fully managed and can scale to power hundreds of millions of searches a day. Even at millions of searches it is blazing fast through edge caching, GPU utilisation and software optimisation so you never have to worry about scaling your infrastructure as your use case scales.
  • Simple to use. Quick to get started.: One of our core design principles is that we focus on how people can get started on using Vector AI as quickly as possible, whilst ensuring there is still a tonne of functionality and customisability options.
  • Richer understanding of your vectors and their properties: Our library is designed to allow people to do more than just obtain nearest neighbors, but to actually experiment, analyse, interpret and improve on them the moment the data added to the index.
  • Store vector data with ease: The document-orientated nature for Vector AI allows users to label, filter search and understand their vectors as much as possible.
  • Real time access to data: Vector AI data is accessible in real time, as soon as the data is inserted it is searchable straight away. No need to wait hours to build an index.
  • Framework agnostic: We are never going to force a specific framework on Vector AI. If you have a framework of choice, you can use it - as long as your documents are JSON-serializable!

Using VectorHub Models

VectorHub is Vector AI's main model repository. Models from VectorHub are built with scikit-learn interfaces and all have examples of Vector AI integration. If you are looking to experiment with new off-the-shelf models, we recommend giving VectorHub models a go - all of them have been tested on Colab and are able to be used in as little as 3 lines of code!

Schema Rules for documents (BYO Vectors and IDs)

Ensure that any vector fields contain a '_vector_' in its name and that any ID fields have the name '_id'.

For example:

example_item = {
    '_id': 'James',
    'skills_vector_': [0.123, 0.456, 0.789, 0.987, 0.654, 0.321]
}

The following will not be recognised as ID columns or vector columns.

example_item = {
    'name_id': 'James',
    'skillsvector_': [0.123, 0.456, 0.789, 0.987, 0.654, 0.321]
}

How does this differ from the VectorAI API?

The Python SDK is designed to provide a way for Pythonistas to unlock the power of VectorAI in as few lines as code as possible. It exposes all the elements of an API through our open-sourced automation tool and is the main way our data scientists and engineers interact with the VectorAI engine for quick prototyping before developers utilise API requests.

Note: The VectorAI SDK is built on the development server which can sometimes cause errors. However, this is important to ensure that users are able to access the most cutting-edge features as required. If you run into such issues, we recommend creating a GitHub Issue if it is non-urgent, but feel free to ping the Discord channel for more urgent enquiries.


Building Products with Vector AI

Creating a multi-language AI fashion assistant: https://fashionfiesta.me | Blog

Demo

Do share with us any blogs or websites you create with Vector AI!

You might also like...
The end-to-end platform for building voice products at scale
The end-to-end platform for building voice products at scale

Picovoice Made in Vancouver, Canada by Picovoice Picovoice is the end-to-end platform for building voice products on your terms. Unlike Alexa and Goog

Python library containing BART query generation and BERT-based Siamese models for neural retrieval.
Python library containing BART query generation and BERT-based Siamese models for neural retrieval.

Neural Retrieval Embedding-based Zero-shot Retrieval through Query Generation leverages query synthesis over large corpuses of unlabeled text (such as

Repo for CVPR2021 paper
Repo for CVPR2021 paper "QPIC: Query-Based Pairwise Human-Object Interaction Detection with Image-Wide Contextual Information"

QPIC: Query-Based Pairwise Human-Object Interaction Detection with Image-Wide Contextual Information by Masato Tamura, Hiroki Ohashi, and Tomoaki Yosh

Deep Learning Datasets Maker is a QGIS plugin to make datasets creation easier for raster and vector data.
Deep Learning Datasets Maker is a QGIS plugin to make datasets creation easier for raster and vector data.

Deep Learning Dataset Maker Deep Learning Datasets Maker is a QGIS plugin to make datasets creation easier for raster and vector data. How to use Down

SEOVER: Sentence-level Emotion Orientation Vector based Conversation Emotion Recognition Model

SEOVER-Master This code is the implementation of paper: SEOVER: Sentence-level Emotion Orientation Vector based Conversation Emotion Recognition Model

QuakeLabeler is a Python package to create and manage your seismic training data, processes, and visualization in a single place — so you can focus on building the next big thing.
QuakeLabeler is a Python package to create and manage your seismic training data, processes, and visualization in a single place — so you can focus on building the next big thing.

QuakeLabeler Quake Labeler was born from the need for seismologists and developers who are not AI specialists to easily, quickly, and independently bu

The implementation of CVPR2021 paper Temporal Query Networks for Fine-grained Video Understanding, by Chuhan Zhang, Ankush Gupta and Andrew Zisserman.
The implementation of CVPR2021 paper Temporal Query Networks for Fine-grained Video Understanding, by Chuhan Zhang, Ankush Gupta and Andrew Zisserman.

Temporal Query Networks for Fine-grained Video Understanding 📋 This repository contains the implementation of CVPR2021 paper Temporal_Query_Networks

Generative Query Network (GQN) in PyTorch as described in
Generative Query Network (GQN) in PyTorch as described in "Neural Scene Representation and Rendering"

Update 2019/06/24: A model trained on 10% of the Shepard-Metzler dataset has been added, the following notebook explains the main features of this mod

Vision-Language Transformer and Query Generation for Referring Segmentation (ICCV 2021)

Vision-Language Transformer and Query Generation for Referring Segmentation Please consider citing our paper in your publications if the project helps

Comments
  • Accessing Discord

    Accessing Discord

    Hi Vector AI Team!

    I'm trying to access the Discord invite link mentioned in the readme: https://discord.gg/CbwUxyD But getting an "invalid invite link".

    I'm writing a new blog post covering the many neural search frameworks, in spirit of my blog post on Vector DBs: https://towardsdatascience.com/milvus-pinecone-vespa-weaviate-vald-gsi-what-unites-these-buzz-words-and-what-makes-each-9c65a3bd0696

    If that's okay, I'd like to ask a couple of questions on the inner workings of the framework and some of its features.

    Thanks,

    Dmitry

    opened by DmitryKey 0
  • Same search results for searching very different images.

    Same search results for searching very different images.

    Using the unsplash-images collection: https://playground.getvectorai.com/collections/?collection=unsplash-images

    result for: vi_client.search_image('unsplash-images', image_url, ['image_url_vector_']) with image_url as: https://www.rover.com/blog/wp-content/uploads/2020/06/siberian-husky-4735878_1920.jpg https://davidkerrphotography.co.nz/wp-content/uploads/2016/10/Slide01.jpg

    identical result for both:

    {'count': 17506,
     'results': [{'_clusters_': {},
                  '_id': 'tLUgvVaCQnY',
                  '_search_score': 0.6311334,
                  'dictionary_label_1': 'wineglasses',
                  'dictionary_label_2': 'delftware',
                  'image_url': 'https://images.unsplash.com/photo-1540735242080-bc0ad0cdcd1e?w=300&q=80',
                  'insert_date_': '2021-02-25T03:38:08.205446',
                  'likes': 150005},
                 {'_clusters_': {},
                  '_id': 'wVMuNOSt5KY',
                  '_search_score': 0.6278121000000001,
                  'dictionary_label_2': 'bootstrapping',
                  'image_url': 'https://images.unsplash.com/photo-1556912743-90a361c19b16?w=300&q=80',
                  'insert_date_': '2021-02-25T03:38:08.018132',
                  'likes': 173693},
                 {'_clusters_': {},
                  '_id': 'kkBXGVE9k-8',
                  '_search_score': 0.626989,
                  'dictionary_label_1': 'occupant',
                  'dictionary_label_2': 'catabolized',
                  'image_url': 'https://images.unsplash.com/photo-1526529516337-f40ddc5532e2?w=300&q=80',
                  'insert_date_': '2021-02-25T03:38:08.129598',
                  'likes': 627490},
                 {'_clusters_': {},
                  '_id': 'pLshzlb5yOA',
                  '_search_score': 0.6268415,
                  'dictionary_label_2': 'wood',
                  'image_url': 'https://images.unsplash.com/photo-1582459208380-f99d357adf33?w=300&q=80',
                  'insert_date_': '2021-02-25T03:38:08.096761',
                  'likes': 173756},
                 {'_clusters_': {},
                  '_id': 'sHmW616civc',
                  '_search_score': 0.6268100999999999,
                  'dictionary_label_2': 'trail',
                  'image_url': 'https://images.unsplash.com/photo-1556674524-65bf99573bef?w=300&q=80',
                  'insert_date_': '2021-02-25T03:38:08.000302',
                  'likes': 682592},
                 {'_clusters_': {},
                  '_id': 'VoTqMJLLSI8',
                  '_search_score': 0.6235797000000001,
                  'dictionary_label_1': 'trays',
                  'dictionary_label_2': 'dishware',
                  'image_url': 'https://images.unsplash.com/photo-1569272559969-2a9275513966?w=300&q=80',
                  'insert_date_': '2021-02-25T03:38:08.202763',
                  'likes': 172006},
                 {'_clusters_': {},
                  '_id': 'XcWKh-GF69M',
                  '_search_score': 0.6210401999999999,
                  'dictionary_label_2': 'obliging',
                  'image_url': 'https://images.unsplash.com/photo-1581280227715-56d3062138a9?w=300&q=80',
                  'insert_date_': '2021-02-25T03:38:20.517206',
                  'likes': 678324},
                 {'_clusters_': {},
                  '_id': 'b2_pVdk4lGI',
                  '_search_score': 0.6187004,
                  'dictionary_label_2': 'jukebox',
                  'image_url': 'https://images.unsplash.com/photo-1568967906094-1d0acfbf0676?w=300&q=80',
                  'insert_date_': '2021-02-25T03:38:20.509971',
                  'likes': 138088},
                 {'_clusters_': {},
                  '_id': '22HltbHJbPI',
                  '_search_score': 0.6182232000000001,
                  'dictionary_label_1': 'shoreline',
                  'dictionary_label_2': 'buckeens',
                  'image_url': 'https://images.unsplash.com/photo-1541514467948-60ec8a24e84f?w=300&q=80',
                  'insert_date_': '2021-02-25T09:44:25.156647',
                  'likes': 758805},
                 {'_clusters_': {},
                  '_id': 'uM3pEsEkPHA',
                  '_search_score': 0.6179558,
                  'dictionary_label_2': 'dewclaw',
                  'image_url': 'https://images.unsplash.com/photo-1572725364984-c2a074c6740c?w=300&q=80',
                  'insert_date_': '2021-02-25T03:38:08.111128',
                  'likes': 655907}]}
    
    opened by elliotsayes 4
  • Bulid type-safe assertive decorator

    Bulid type-safe assertive decorator

    With Python's type-safety is difficult but it can be implemented through smart use of Python decorators. An interesting example can be seen below:

    import itertools as it
    
    @parametrized
    def types(f, *types):
        def rep(*args):
            for a, t, n in zip(args, types, it.count()):
                if type(a) is not t:
                    raise TypeError('Value %d has not type %s. %s instead' %
                        (n, t, type(a))
                    )
            return f(*args)
        return rep
    
    @types(str, int)  # arg1 is str, arg2 is int
    def string_multiply(text, times):
        return text * times
    
    print(string_multiply('hello', 3))    # Prints hellohellohello
    print(string_multiply(3, 3))          # Fails miserably with TypeError
    
    # From: https://stackoverflow.com/questions/5929107/decorators-with-parameters
    
    enhancement 
    opened by boba-and-beer 0
Releases(v0.2.5)
Collision risk estimation using stochastic motion models

collision_risk_estimation Collision risk estimation using stochastic motion models. This is a new approach, based on stochastic models, to predict the

Unmesh 7 Jun 26, 2022
[LREC] MMChat: Multi-Modal Chat Dataset on Social Media

MMChat This repo contains the code and data for the LREC2022 paper MMChat: Multi-Modal Chat Dataset on Social Media. Dataset MMChat is a large-scale d

Silver 47 Jan 03, 2023
Codes for "Template-free Prompt Tuning for Few-shot NER".

EntLM The source codes for EntLM. Dependencies: Cuda 10.1, python 3.6.5 To install the required packages by following commands: $ pip3 install -r requ

77 Dec 27, 2022
《LightXML: Transformer with dynamic negative sampling for High-Performance Extreme Multi-label Text Classification》(AAAI 2021) GitHub:

LightXML: Transformer with dynamic negative sampling for High-Performance Extreme Multi-label Text Classification

76 Dec 05, 2022
Pytorch implementation of the paper: "SAPNet: Segmentation-Aware Progressive Network for Perceptual Contrastive Image Deraining"

SAPNet This repository contains the official Pytorch implementation of the paper: "SAPNet: Segmentation-Aware Progressive Network for Perceptual Contr

11 Oct 17, 2022
Qimera: Data-free Quantization with Synthetic Boundary Supporting Samples

Qimera: Data-free Quantization with Synthetic Boundary Supporting Samples This repository is the official implementation of paper [Qimera: Data-free Q

Kanghyun Choi 21 Nov 03, 2022
A simple program for training and testing vit

Vit This is a simple program for training and testing vit. Key requirements: torch, torchvision and timm. Dataset I put 5 categories of the cub classi

xiezhenyu 2 Oct 11, 2022
Official repository of the paper Learning to Regress 3D Face Shape and Expression from an Image without 3D Supervision

Official repository of the paper Learning to Regress 3D Face Shape and Expression from an Image without 3D Supervision

Soubhik Sanyal 689 Dec 25, 2022
Speeding-Up Back-Propagation in DNN: Approximate Outer Product with Memory

Approximate Outer Product Gradient Descent with Memory Code for the numerical experiment of the paper Speeding-Up Back-Propagation in DNN: Approximate

2 Mar 02, 2022
A Strong Baseline for Image Semantic Segmentation

A Strong Baseline for Image Semantic Segmentation Introduction This project is an open source semantic segmentation toolbox based on PyTorch. It is ba

Clark He 49 Sep 20, 2022
A different spin on dataclasses.

dataklasses Dataklasses is a library that allows you to quickly define data classes using Python type hints. Here's an example of how you use it: from

David Beazley 752 Nov 18, 2022
Official Implementation and Dataset of "PPR10K: A Large-Scale Portrait Photo Retouching Dataset with Human-Region Mask and Group-Level Consistency", CVPR 2021

Portrait Photo Retouching with PPR10K Paper | Supplementary Material PPR10K: A Large-Scale Portrait Photo Retouching Dataset with Human-Region Mask an

184 Dec 11, 2022
Spatial Action Maps for Mobile Manipulation (RSS 2020)

spatial-action-maps Update: Please see our new spatial-intention-maps repository, which extends this work to multi-agent settings. It contains many ne

Jimmy Wu 27 Nov 30, 2022
Efficiently Disentangle Causal Representations

Efficiently Disentangle Causal Representations Install dependency pip install -r requirements.txt Main experiments Causality direction prediction cd

4 Apr 01, 2022
Distance correlation and related E-statistics in Python

dcor dcor: distance correlation and related E-statistics in Python. E-statistics are functions of distances between statistical observations in metric

Carlos Ramos Carreño 108 Dec 27, 2022
Sequence to Sequence Models with PyTorch

Sequence to Sequence models with PyTorch This repository contains implementations of Sequence to Sequence (Seq2Seq) models in PyTorch At present it ha

Sandeep Subramanian 708 Dec 19, 2022
Baseline powergrid model for NY

Baseline-powergrid-model-for-NY Table of Contents About The Project Built With Usage License Contact Acknowledgements About The Project As the urgency

Anderson Energy Lab at Cornell 6 Nov 24, 2022
Film review classification

Film review classification Решение задачи классификации отзывов на фильмы на положительные и отрицательные с помощью рекуррентных нейронных сетей 1. З

Nikita Dukin 3 Jan 21, 2022
Solution to the Weather4cast 2021 challenge

This code was used for the entry by the team "antfugue" for the Weather4cast 2021 Challenge. Below, you can find the instructions for generating predi

Jussi Leinonen 13 Jan 03, 2023
Deep Crop Rotation

Deep Crop Rotation Paper (to come very soon!) We propose a deep learning approach to modelling both inter- and intra-annual patterns for parcel classi

Félix Quinton 5 Sep 23, 2022