Decoding GPT-2: Mechanistic Insights into Sentiment Processing

This ArXiv paper provides valuable insights into how GPT-2 processes sentiment through mechanistic interpretability. Analyzing the lexical and contextual layers offers a deeper understanding of the model's decision-making process.

• •Applies mechanistic interpretability techniques to understand GPT-2's internal workings.
• •Specifically examines how lexical and contextual information influences sentiment classification.
• •Contributes to the broader field of explainable AI (XAI).