Towards a Mechanistic Understanding of Language Models in Syntactic Code Completion
Year: 2025
Type: dissertation
Abstract: Recently, language models (LMs) have shown impressive proficiency in code generation tasks, especially when fine-tuned on code-specific datasets, commonly known as Code LMs. However, our understanding of the internal decision-making processes of Code LMs, such as how they use their (syntactic or semantic) knowledge, remains limited, which could lead to unintended harm as they are increasingly used in real life. This motivates us to conduct one of the first Mechanistic Interpretability works to understand how Code LMs perform a syntactic completion task, specifically the closing parenthesis task, on the CodeLlama-7b model [2]. Our findings reveal that the model requires middle-later layers until it can confidently predict the correct label for the closing parenthesis task. Additionally, we identify that while both multi-head attention (MHA) and feed-forward (FF) sub-layers play essential roles, MHA is particularly crucial. We also discover attention heads that keep track of the number of already closed parentheses precisely but may or may not promote a correct number of closing parentheses that are still missing, leading to a positive or negative impact on the model’s performance. Following, we discover a circuit for the Two Closing Parentheses sub-task and use it to identify the information flows required to for the correct token prediction.
DOI: 10.13021/mars/14914
Author: Miller, Samuel
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: green
OpenAlex ID: W7077149841
Cites:
Cited by:
Related to: