Hypothesis Testing the Circuit Hypothesis in LLMs
Fuente:
arXiv
Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866910654004199424 |
|---|---|
| author | Shi, Claudia Beltran-Velez, Nicolas Nazaret, Achille Zheng, Carolina Garriga-Alonso, Adrià Jesson, Andrew Makar, Maggie Blei, David M. |
| author_facet | Shi, Claudia Beltran-Velez, Nicolas Nazaret, Achille Zheng, Carolina Garriga-Alonso, Adrià Jesson, Andrew Makar, Maggie Blei, David M. |
| contents | Large language models (LLMs) demonstrate surprising capabilities, but we do not understand how they are implemented. One hypothesis suggests that these capabilities are primarily executed by small subnetworks within the LLM, known as circuits. But how can we evaluate this hypothesis? In this paper, we formalize a set of criteria that a circuit is hypothesized to meet and develop a suite of hypothesis tests to evaluate how well circuits satisfy them. The criteria focus on the extent to which the LLM's behavior is preserved, the degree of localization of this behavior, and whether the circuit is minimal. We apply these tests to six circuits described in the research literature. We find that synthetic circuits -- circuits that are hard-coded in the model -- align with the idealized properties. Circuits discovered in Transformer models satisfy the criteria to varying degrees. To facilitate future empirical studies of circuits, we created the \textit{circuitry} package, a wrapper around the \textit{TransformerLens} library, which abstracts away lower-level manipulations of hooks and activations. The software is available at \url{https://github.com/blei-lab/circuitry}. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2410_13032 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Hypothesis Testing the Circuit Hypothesis in LLMs Shi, Claudia Beltran-Velez, Nicolas Nazaret, Achille Zheng, Carolina Garriga-Alonso, Adrià Jesson, Andrew Makar, Maggie Blei, David M. Artificial Intelligence Machine Learning Large language models (LLMs) demonstrate surprising capabilities, but we do not understand how they are implemented. One hypothesis suggests that these capabilities are primarily executed by small subnetworks within the LLM, known as circuits. But how can we evaluate this hypothesis? In this paper, we formalize a set of criteria that a circuit is hypothesized to meet and develop a suite of hypothesis tests to evaluate how well circuits satisfy them. The criteria focus on the extent to which the LLM's behavior is preserved, the degree of localization of this behavior, and whether the circuit is minimal. We apply these tests to six circuits described in the research literature. We find that synthetic circuits -- circuits that are hard-coded in the model -- align with the idealized properties. Circuits discovered in Transformer models satisfy the criteria to varying degrees. To facilitate future empirical studies of circuits, we created the \textit{circuitry} package, a wrapper around the \textit{TransformerLens} library, which abstracts away lower-level manipulations of hooks and activations. The software is available at \url{https://github.com/blei-lab/circuitry}. |
| title | Hypothesis Testing the Circuit Hypothesis in LLMs |
| topic | Artificial Intelligence Machine Learning |
| url | https://arxiv.org/abs/2410.13032 |