Figure 2
The flowchart begins with a generative artificial intelligence system where a sample is inputted. This generates data which forms a dataset sample. The dataset sample is then translated from Creole to English, creating a Creole to English dataset. This dataset is used to train an untrained base BART model. The trained model is then fine-tuned, resulting in a fine-tuned BART model. If there are incorrect translations, manual tuning is applied to correct them, and the process loops back to the dataset sample stage.Proposed framework for retraining standard English LLM to Trinidad Creole