We got tired of labeling data in an external environment, training a model and then running inference and checking the confidences of the predictions and going through the data again for weeks.
We wanted a process were we only would have to spend a few hours to create a new model that was production ready. That's why we built Labelf. Active learning on your dataset for custom model creation.
We've built Labelf as a visual UI for a transformer encoders training process, turning training into a teaching session. You can now leave the training/labeling to the domain expert and use the deployed model directly with a few lines of code in your system.
You can create a POC of a system with a custom AI model implemented with your own code in less than an hour!