This week in Jina AI
No-code Image Search with Jina NOW, Simple cloud deployment, and DALL-E Mini
Not to mention: Movie recommender, product updates and more
Want to understand all the products Jina AI is building for the Neural Search Ecosystem? Starting from organizing your data to creating embeddings to fine-tuning to deploying to scaling and much more; we have a product for each problem to get you one step closer to shipping your Neural Search system in production. Here’s how:
New Releases
Jina NOW
Jina NOW is the first-ever no-code solution that lets you create and deploy multi-modal neural search in a matter of minutes!
With Jina NOW, you can create and deploy search solutions with pre-built functionalities. And you can do that with just one line of code. Here is how:
# Install jina-now first. Supports Python 3.7+
pip install jina-now# Start Jina NOW client
jina-now start
# It will ask you some questions such as dataset, models, etc. and you'll have everything ready for you
Check out the live demo, introductory blog and read the documentation or watch this video guide to dive deeper.
JCloud
JCloud is a CLI that simplifies deploying and managing Jina projects on Jina Cloud.
pip install jcloud
jc deploy ./my-project/
To use it, make sure to update your DocArray to the latest version — pip install -U docarray
Early free access — We offer to host your Jina project and provide computational and storage resources, for free! If you have already developed some Jina projects, try it out.
Also we have deployed a DALL-E Mini example on JCloud. Here’s the Python/curl code snippet you can copy paste into your terminal to create images from text.
More info: https://github.com/jina-ai/jcloud
DocArray 0.13
Apart from usual bug fixes and feature development, we did something many community members asked for.
We released 1 Million scale benchmarks on 6 different operations for DocArray backends. Check them out to understand the best Document stores for your use case and enjoy DocumentArray API and persistence at scale.
CLIP-as-service 0.3.0
New features:
- Now you can enjoy the biggest CLIP model
ViT-L/14@336px
released by OpenAI this week in PyTorch & ONNX - Rerank feature which allows one to rerank the cross-modal matches. Check out this code snippet to learn how and read this guide to dive deeper.
Make sure to update the server pip install -U clip_server
and update the client pip install -U clip_client
Read the full changelog here
Jina 3.3.10
We fixed many bugs and improved performance. We encourage you to upgrade to the latest version pip install -U jina
Find out the full list of changes since the release of 3.3(a few weeks ago) here
Jina Hub
We published a new Executor — DALL-E Mini. It can generate images from text prompts using DALL-E Mini model.
How to use it:
from jina import Flowf = Flow().add(uses='jinahub+sandbox://DalleMini')
Community Showcase
Movie Recommender built with Jina
Achintya showcased his Jina project — Movie Recommender which takes a text query as an input and gives you a list of movies according to the query.
Tech Stack → Jina, Rest API, Dart
Database→ IMDB Movies Dataset
Source Code → GitHub repo (Don’t forget to star it ⭐)
Tutorial → Achintya’s post on Medium
Blog
Get Jina Certified
Take our Learning Bootcamp quiz with beginner, intermediate and advanced tracks to etch your name in our Hall of Fame!
Contribute to our Good First Issues Repo
Our good first issues repo features some great issues for those looking to take their first steps into open source with Jina. Check them out here:
- Example: NFT/blockchain search engine
- Update Jina Hub Executors for Jina 3.0+
- React component for Jina
Join our Communities
- Slack: Our Slack community now boasts over 2,700 neural search enthusiasts. We would love to have you there!✨
- Meetup: Join our meetup group to stay updated with our latest events. Whether you’re AI engineers, Python Engineers, Data Scientists, CTO, Machine Learning Researchers, Open Source Evangelists, AI Product Managers, or just plain old interested, we’d love to meet up with you!
- Newsletter: Subscribe to our newsletter if you want to hear from us.