Jina NOW is a cloud-based AI tool that enables users to leverage the power of multimodal neural search with just one command. This innovative tool is designed to be fast, easy-to-use, and provides a range of features, including the ability to customize and support various modalities. The tool's simple installation and user-friendly interface make it accessible to individuals with less technical experience, using a no-code system to make neural search more accessible.
Jina NOW is compatible with Linux and macOS and works with various data types such as text, image, and audio. Additionally, it provides a Swagger UI interface for front-end integration and a "playground" for experimenting with search use cases.
Jina NOW operates with a freemium model, offering basic features for free with certain limitations and paid plans with more advanced features and capabilities.
TLDR
Jina NOW is an accessible and user-friendly AI tool that facilitates multimodal neural search with a single command. The tool comes with many features, including customization, modality support, deployment and integration, and compatibility with Linux and macOS operating systems. Jina NOW provides a "playground" for experimentation with search use cases and a Swagger UI interface for front-end integration.
The company operates using a freemium model, offering basic features for free and paid plans with more advanced features and capabilities. Overall, Jina NOW is an attractive option for individuals with minimal technical experience seeking accessibility and fast, accurate search results.
Company Overview
Jina NOW is a cloud-based AI tool that provides access to multimodal neural search with just one command. Designed for fast and easy use, users can set up their search use case in minutes with minimal effort. Jina NOW offers a quality service that fine-tunes models that are labelled by users.
Additionally, non-technical people can deploy with ease, thanks to their nocode system.
Jina NOW supports various formats for uploading a dataset to a search application, including demo datasets hosted by Jina NOW, custom datasets, DocumentArray, local folder, or S3 bucket. Users can choose a demo dataset to get started quickly, which is very useful for building a search application. A wide variety of datasets is available, including images, text, and audio.
Users can also upload their own custom data as DocumentArray by providing the DocumentArray ID or name. Alternatively, they can choose a local folder and provide the path to the folder containing their data. When data is stored in an S3 bucket, users can provide the URI to the S3 bucket, as well as the credentials and region.
Fields of data that users want to use for search and filter can also be selected in this step.
After processing, Jina NOW provides two links; the Swagger UI is useful for front-end integration, while the "playground" lets users run example queries and experiment with their search use cases. The AI tool supports various modalities such as text, image, music, video (for GIFs), and 3D Mesh (coming soon).
Jina NOW is readily available with an easy installation process. To install, users can just run ' pip install jina-now' in their command prompt terminal.
It is important to note that Jina NOW is only available on Linux and macOS.
Features
Easy Installation and User-Friendly Interface
Simple Installation Process
Users can easily install the Jina NOW AI tool by running the command 'pip install jina-now' in their command prompt terminal. The tool is readily available for use after installation. The installation process is quick and straightforward, making it user-friendly even for those with less technical experience.
Visual Interface with No-Code System
Jina NOW offers a quality service that fine-tunes models, and non-technical people can deploy it with ease thanks to its no-code system. Its visual interface makes it easy to use, and users can set up their search use case in minutes with minimal effort. The AI tool's quick and easy access and simple interface make it an attractive option for many users.
Demo Datasets with a Wide Variety of Data Modalities
Jina NOW offers a variety of datasets that are available on their platform, including demo datasets hosted by Jina NOW, custom datasets, DocumentArray, local folders, or S3 buckets. Users can easily choose a demo dataset to get started quickly, which is beneficial when building a search application. They offer a wide variety of dataset modalities, including text, images, and audio, making it a versatile AI search tool.
Customization and Modality Support
Support for Various Formats and Custom Data
Aside from its demo datasets, Jina NOW supports various formats for uploading a dataset to a search application, including custom data as DocumentArray by providing the DocumentArray ID or name, or by selecting a local folder and providing the path to the folder containing the data. When data is stored in an S3 bucket, users can provide the URI to the S3 bucket, as well as the credentials and region. This flexibility allows users to customize their search and tailor it to their specific needs and data.
Support for Different Modalities
Jina NOW supports various modalities such as text, image, audio, video (for GIFs), and 3D Mesh (coming soon). The seamless support for different modalities allows users to search for the most relevant and accurate results using different data types or modalities.
Fields Selection for Search and Filters
One essential feature of Jina NOW is the ability to select fields of data that users want to use for search and filter. This feature allows users to focus their search on specific fields that are relevant to their search query. It helps reduce the noise in the search results and provides users with more accurate and useful search results.
Deployment and Integration
Swagger UI Interface for Front-End Integration
After processing, Jina NOW provides two links; the Swagger UI interface is useful for front-end integration with other applications. Users can easily integrate this into their front-end applications, enabling seamless search capabilities for their users.
Playground for Experimentation with Search
Another useful feature that Jina NOW provides is its "playground." This feature lets users run example queries and experiment with their search use cases. Users can experiment with different parameters to get the best search results, and the playground feature allows them to refine and improve their searches.
Cloud-Based Solution with Quality Assurance
Jina NOW hosts, scales, and deploys common neural search tasks to production quickly, validating user needs with custom data, using a simple visual interface, and without writing a single line of code. Jina NOW constantly fine-tunes models that are labeled by users, providing a quality assurance system that helps users achieve more reliable and accurate search results with minimal effort.
Platform Compatibility
Availability on Linux and macOS
Jina NOW is readily available to Linux and macOS users. However, it is not available for Windows operating systems. The AI tool's compatibility with only two of the three major operating systems may be a limitation, but the software provides sufficient options for users who utilize Linux and macOS operating systems.
Compatibility with Mac M1
Jina NOW is also compatible with Mac M1, and the developers recommend using a Conda environment for M1. In a new Conda environment, users can run 'conda install grpcio tokenizers protobuf' to install the necessary components for Jina NOW. These updates make Jina NOW more accessible to a more significant number of users who are utilizing the latest operating systems.
Automatic Updates and Changelog Tracking
Jina NOW follows semantic versioning and generates automated release notes for every update. This feature lets users know what new features, bugs, refactorings, and breaking changes have been introduced. This feature helps users keep the AI tool running smoothly by keeping them informed of any new developments or changes.
Pricing
Jina NOW offers a range of AI tool solutions to empower your enterprise, from on-premise fine-tuning solutions to no-code solutions for creating neural search apps. The company operates on a freemium model, meaning customers can utilize the tools for free with certain limitations and then upgrade to a paid plan for more advanced features and capabilities.
The free plan includes access to the basic features of Jina NOW's AI tools, including automatic prompt engineering for ChatGPT, GPT3.5, DALL-E, and Stable Diffusion. This plan also includes the ability to create neural search and multi-modal AI applications in the cloud, as well as access to the data structure for multimodal data and CLIP, which allows users to embed images and sentences into fixed-length vectors. Additionally, Jina NOW offers a Human-in-the-Loop workflow for creating HD images from text and creating Disco Diffusion artworks in one line of code.
Jina NOW's paid plans start at $119/month for the Starter plan, which includes advanced features such as the ability to fine-tune embeddings on domain-specific data for better search quality, as well as sharing and discovering building blocks for multimodal AI applications. The Pro plan, priced at $499/month, includes all the features in the Starter plan, as well as the ability to create neural search apps without any coding necessary. The Enterprise plan, which has custom pricing and specialized features, is geared towards large organizations with more complex AI needs.
Overall, Jina NOW's pricing is reasonable and provides customers with a wide range of AI tools and capabilities to choose from. The freemium model is an excellent way for customers to test the waters before committing to a paid plan, and the paid plans offer a variety of features for different levels of AI sophistication.
FAQ
What is Jina NOW's neural search framework?
Jina NOW's neural search framework is an AI tool that leverages the power of deep neural networks to build search systems. This tool helps developers create intelligent search systems by integrating Jina's examples and tweaking them as needed.
By using Jina's framework, developers can build advanced neural search systems using intermediate-level Python and a working PC or Mac. Previous knowledge of ML and AI is a plus but isn't a requirement.
The implementation is as simple as possible for developers who want to create neural search systems.
What do I need to use Jina?
To use Jina, you'll need intermediate-level Python and a working PC or Mac. Previous knowledge of ML and AI is a plus, but it's not a requirement. Jina provides open-source repos for you to fork and tweak as needed.
How do I install Jina?
Jina can be installed in three ways:
- Using Docker
- Using Anaconda
- Using PIP
For more details about installing Jina, visit our documentation. We recommend installing WSL2 and running Jina from there.
Alternatively, you could try a VM like Virtualbox running a recent Linux distribution. However, you may have trouble installing or using Jina on your Mac if you have a newer model with an M1 chip.
Please follow the steps provided on our blog if you have issues.
What are the resources required to build a neural search system with Jina?
The resources required to build a neural search system with Jina depend on the business requirements, such as data volume, stability requirements, required response time, etc. The CPU can cope with a lesser data volume (up to one million) for a single data type.
For applications with larger data volume, such as retrieving hundreds of millions of videos and requiring millisecond-level feedback, it is necessary to use GPU. Some customers use Jina to build a search system for a company's internal resources search, including PDF search, through text directly searching the relevant semantic content, or through the text to match the images in the PDF.
Can I integrate a traditional search system with Jina?
Yes, you can fully integrate a traditional search system with Jina. As in the DocQA example, the first step is pulling candidates and finding out similar passages by vector indexing. The second step is to use a more computationally intensive deep learning model to find the needed answers from the passages.
When doing the recall in the first step, you can use the method based on vector index or TF-IDF or BM25, so it is entirely possible to use the traditional inverted index to recall Jina.
How can I reduce the size of my indexed data?
You can reduce the size of your indexed data by projecting the embedding to a smaller dimensionality, using pre-trained ResNet results in features represented as 2048d (if you're using a fully connected layer as an embedding layer). You can further encode it into another dimensionality, such as 512. You can achieve this with Finetuner by adopting a multi-layer perceptron on top of your embedding model.
For instance, in this tutorial, we attached a SimpleMLP on top of the embedding model, and the final embedding has been encoded into 1024d, two times smaller than the pre-trained embedding. You can do it in 128/256/512 or any compact representation. This should significantly reduce your embedding size.
Alternatives
If you're looking for alternatives to Jina, here are a few options:
Elasticsearch
Elasticsearch is an open-source search engine that can be used to index and search text, images, and other data. It has a robust set of features, including full-text search, real-time updates, and support for multiple languages.
Algolia
Algolia is a cloud-based search and discovery platform that allows you to build fast, relevant search experiences for your users. It has features such as typo-tolerance, synonyms, and advanced filtering capabilities for better user engagement and experience.
Amazon CloudSearch
Amazon CloudSearch is a fully managed service in the AWS Cloud that makes it easy to set up, manage, and scale a search solution for your website or application. It provides features such as automatic scaling, query autocompletion, and facet exploration for better search relevance and analytics.
Apache Lucene/Solr
Apache Lucene is a powerful open-source information retrieval library, while Apache Solr is a search platform built on top of Lucene. Both can be used independently or together to build search applications with features such as full-text search, faceting, and hit highlighting.
