GLUScope: A Tool for Analyzing GLU Neurons in Transformer Language Models
Fuente:
arXiv
Saved in:
| Main Authors: | , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866908855591501824 |
|---|---|
| author | Gerstner, Sebastian Schütze, Hinrich |
| author_facet | Gerstner, Sebastian Schütze, Hinrich |
| contents | We present GLUScope, an open-source tool for analyzing neurons in Transformer-based language models, intended for interpretability researchers. We focus on more recent models than previous tools do; specifically we consider gated activation functions such as SwiGLU. This introduces a new challenge: understanding positive activations is not enough. Instead, both the gate and the in activation of a neuron can be positive or negative, leading to four different possible sign combinations that in some cases have quite different functionalities. Accordingly, for any neuron, our tool shows text examples for each of the four sign combinations, and indicates how often each combination occurs. We describe examples of how our tool can lead to novel insights. A demo is available at https: //sjgerstner.github.io/gluscope. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_23826 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | GLUScope: A Tool for Analyzing GLU Neurons in Transformer Language Models Gerstner, Sebastian Schütze, Hinrich Computation and Language Machine Learning We present GLUScope, an open-source tool for analyzing neurons in Transformer-based language models, intended for interpretability researchers. We focus on more recent models than previous tools do; specifically we consider gated activation functions such as SwiGLU. This introduces a new challenge: understanding positive activations is not enough. Instead, both the gate and the in activation of a neuron can be positive or negative, leading to four different possible sign combinations that in some cases have quite different functionalities. Accordingly, for any neuron, our tool shows text examples for each of the four sign combinations, and indicates how often each combination occurs. We describe examples of how our tool can lead to novel insights. A demo is available at https: //sjgerstner.github.io/gluscope. |
| title | GLUScope: A Tool for Analyzing GLU Neurons in Transformer Language Models |
| topic | Computation and Language Machine Learning |
| url | https://arxiv.org/abs/2602.23826 |