Large Language Model- Models worth considering to improve our project.

The MiniChefGPT project utilizes a GPT-style Transformer built with Python and PyTorch to generate cooking recipes from user-provided ingredients. Previously, the team developed TinyGPT, a small character or word-level Transformer trained from scratch. However, TinyGPT faced limitations including repetitive text, small vocabulary, and a short context window. To improve the project, the team is now evaluating various Large Language Models, including classic baseline recurrent networks such as LSTM and GRU for next-word prediction.