TY  - RPRT
TI  - Mechanistic Interpretability of GPT-2: Lexical and Contextual Layers in Sentiment Analysis
AU  - Amartya Hatua
PY  - 2025
UR  - https://arxiv.org/abs/2512.06681
ID  - 2512.06681
ER  -