Release Notes: What’s new on the NeuralSpace Platform
We at NeuralSpace have been constantly developing to elevate the NeuralSpace Platform and give our users more and more language AI capabilities. The wait is finally over! We are excited to share the latest updates on our Platform.
Whenever you create a new project on the NeuralSpace Platform, you can now do so with a simple 4-step process: Select your language (or choose multilingual/code-mixed), upload or import your dataset, train your model with AutoNLP, and deploy it with AutoMLOps. With new Apps being added to NeuralSpace constantly, we are also working on an onboarding process where multiple Apps, for example Speech-to-Text and Language Understanding are combined in one project.
On the Language Understanding App, you can now upload CSV, JSON, and Rasa YML datasets. We have also made a couple of datasets available for you to import and you can directly navigate to the Import tab to have them in your project. You can now choose from a collection of 74 datasets covering domains like banking, hate speech detection, sentiment analysis and emotion detection in more than 50 languages. It also includes the recently launched Amazon Science’s MASSIVE dataset (link to our benchmark).
Annotated Translation & Annotated Augmentation
If you have chatbot datasets with entities tagged and are struggling with translating the dataset into other languages, you can now do that now with our annotated machine translations. Whenever you have an entity tagged in, let’s say, English and you want to translate the example sentence into Tamil, the same word will be automatically tagged as the given entity in the Tamil translation.
We have also incorporated this feature in our data augmentation offering. You simply tag an entity in your example sentence, generate up to ten additional example sentences with the same intent, and your entities will be automatically tagged in the additional examples, too.
AutoNLP and AutoMLOps
Using AutoMLOps you can deploy your models with a single click! Additionally we now have a “Scale” option where you can deploy multiple replicas (or copies) of your models according to your needs. The more replicas you deploy, the more examples per second can be processed.
Our language support has been expanded with seven new Arabic dialects.
What’s coming next?
We will double down on our STT offering and go live with 15 languages in the mid of June 2022. We will focus on code-mixed languages like Hinglish and Arabizi. Further, we will have the complete no-code interface for Entity Recognition and Transliteration live where our automatic one-click training and deployment (with AutoNLP and AutoMLOps) are built in.
If you haven’t yet, sign-up on the NeuralSpace Platform to try and test it out by yourself! Early sign-ups get $500 worth of credits — what are you waiting for?
Join the NeuralSpace Slack Community to connect with us. Also, receive updates and discuss topics in NLP for low-resource languages with fellow developers and researchers.
Check out our Documentation to read more about the NeuralSpace Platform and its different Apps.