Train an agent to find the best emoji choise for a sentence by using multi-class NLP model and Bert model.
Emotions dataset for NLP in Kaggle
Environment:
Google colab (baseline metrics)
Kaggle notebook with GPU (main model)
Packages version:
Python 3.6.6
numpy 1.18.2
pandas 0.25.3
transformers 3.0.2
Usage:
batch size = 32
#epochs = 3
Hyperparameters:
optimization algorithm: Adam learning rate optimization algorithm (type of Stochastic Gradient Descent with momentum; Momentum is described as the moving average of the gradient instead of gradient itself.)
Experiment results: Accuracy
