Comprehend-it is a next-generation text classification model designed for zero-shot, multi-label categorization with output scoring. Model performing well in zero-shot setting, processing classes it hasnโt seen during training.
Possible use cases of the model:
โZero-Shotโ means that without specific prior training on particular topics, Comprehend-it can analyze and classify any text across a wide range of subjects into thousands of distinct categories. It has notably surpassed 87% accuracy for our enterprise clients in industries like finance, retail, and healthcare. Additionally, it has been recognized as the best model for scientific literature review by the ISCB and FDA.
Each classification result comes with a confidence score, offering a reliable indicator of precision to guide your trust in automation and indicate when a manual review may be necessary.
Below, you can see the F1 score on several text classification datasets. All tested models were not fine-tuned on those datasets and were tested in a zero-shot setting.
API documentation - https://docs.knowledgator.com/docs/api-reference/comprehend-it-api
Give a try in AI Playground, without registration: https://playground.knowledgator.com/comprehend-it
Fine-tuning: https://docs.knowledgator.com/docs/support