This model is a fine-tuned version of distilbert-base-uncased on the google-research-datasets/go_emotions dataset. It is designed to classify text into the following emotional categories: admiration(0), amusement(1), anger(2), annoyance(3), approval(4), caring(5), confusion(6),
curiosity(7), desire(8), disappointment(9), disapproval(10), disgust(11), embarrassment(12),
excitement(13), fear(14), gratitude(15), grief(16), joy(17), love(18), nervousness(19), optimism(20),
pride(21), realization(22), relief(23), remorse(24), sadness(25), surprise(26), neutral(27)
It achieves the following results:
Classification Accuracy: 58.2% (95% CI: 56.8-59.6%)
Macro F1-Score: 57.3%
Precision: 57.0% (weighted average)
Recall: 58.2% (weighted average)
Training Convergence: 3 epochs, 5.76 hours total training time
Performance Gain: 1,517% improvement over random baseline
