Clarifai Community Launch
Clarifai Community puts state-of-the-art AI in the hands of developers everywhere. You can now share AI resources to accelerate global AI development. Clarifai Community is here to help the world’s AI communities to collaborate.
[Pre-Trained Models] Text Sentiment Classification in 6 Languages
You can now understand text sentiment on product reviews in six languages: English, Dutch, German, French, Spanish and Italian. Our new model predicts the sentiment of a text review as a star rating (between 1 and 5). Check it out in Clarifai Community.
[Pre-Trained Models] Text Style Transfer
Informal communication is extremely common these days. We tend to communicate more freely when using informal communication styles. But there are many channels where informal communication is not appropriate or helpful. The text style transfer model can help you automate changing text between formal and informal styles. Informal to formal. Formal to informal.
[Pre-Trained Models] Translations
Now you can add translation functionality to your products and services. Plug the translation model into your technology stack with the Clarifai API. Just visit Clarifai Community and filter by "translation" to see all languages supported from common languages such as Spanish, German, Chinese Mandarin, Arabic. Also, less common languages such as Welsh, Bulgarian, Czech, Danish, and Catalan are supported. We provide multi-source/multi-target translations as well as single-source target translation models. Check it out.
[Pre-Trained Models] Automatic Speech Recognition of 9 Languages
Take advantage of automatic speech recognition in 9 languages. Just visit the Community Model search bar and filter your input to “Audio.” Supported languages include English, Spanish, German, Polish, French, Portuguese, Italian, Dutch, and Arabic.
[Platform-Object] Model-Type: Input-Searcher
Users can now add a model-type to a workflow that can accept any input-type and search against any field in the input.data, make requests to any app_id (using either a key or pat) for authentication, limit results to the min_value and/or max_regions, and return output.data.regions[...] (each containing nested data alongside region_info.hit having the score and rank returned by the query).
