● Building a web app that recognizes emotions given a sentence using a transformer-based model converted to ONNX format.
● Optimizing the model with techniques such as quantization to improve response speed and reduce latency for greater user satisfaction.
● Using a BERT-based model for emotion detection, using hugging face APIs to train the model on the dataset.
● Installing necessary libraries and imports for model training and deployment, including transformers, ONNX, and accelerate.
● Utilizing ONNX and ONNXRuntime for converting and running the model from the frontend.
Author: Marcello Politi