"I worked on investigating Large Language Models' access to processing via the Syntax-Semantic Interface as part of a wider project throughout the computational linguistics world to understand how LLMs process language and if it is analogous to the human scheme thereof. During the summer, I produced a semantically and syntactically diverse data set of more than 30,000 sentences to be passed to six different models (Mistral, Pythia, ChatGPT, Olmo, Llama, and Qwen). Towards the end of my internship, I programmed the experiment to perform this processing by the models and return specific readings regarding the interpreted complexity of the sentences from each, then administered this experiment to the group's supercomputing cluster. I'll be continuing with this project for the foreseeable future."
