© Neuronpedia 2026
Privacy & Terms
Blog
GitHub
Slack
Twitter
Contact
Neuronpedia
Jacobian Lens
NEW
Natural Language
Autoencoders
NEW
Assistant Axis
NEW
Circuit Tracer
UPDATE
Releases
Jump To
Search
Steer
SAE Evals
Exports
interp-engine
NEW
Guides
API
Community
Blog
Privacy & Terms
Contact
Sign In
Home
Lubana, Rager, Hindupur, et al. · Temporal Feature Analysis
Gemma-2-2B
Residual - 9k
12-TEMPORAL-RES
gemma-2-2b · 12-temporal-res
Source from
temporal-feature-analysis
·
Residual - 9k
· Layer 12
Jump to Source/SAE
12-temporal-res
Source/SAE
Go
Jump to Feature
12-temporal-res
Source/SAE
INDEX
Go
Similarity Matrix
?
Mechanistic interpretability (often abbreviated as mech interp, mechinterp, or MI) is a subfield of research within explainable artificial intelligence that aims to understand the internal workings of neural networks by analyzing the mechanisms present in their computations. The approach seeks to analyze neural networks in a manner similar to how binary computer programs can be reverse-engineered to understand their functions.
GENERATE
Run Demo
👯♀️ Alice & Maya
🧠 Mech Interp
👔 Audrey's Job
🍦 Ice Cream
📖 Huck Finn
🐱 Dr. Seuss
Configuration
canrager/temporalSAEs/gemma-2-2B/layer_12/temporal
How To Load
Features
9,216
Data Type
bfloat16
Hook Name
blocks.12.hook_resid_post
Architecture
temporal
Dataset
monology/pile-uncopyrighted
Show All
Search via Inference
?
Residual - 9k
Layer 12
SEARCH
Run Example Search
Random
🌮 Food
📰 News
📖 Literary
👯 Personal
🧑💻 Programming
🧑🔬 Technical
🧑🏫 Academic
💼 Business
🧑⚖️ Legal
🧑🏫 Educational
🗼 Cultural
Search TopK by Token
MODEL
Residual - 9k
LAYER
RANDOM
SEARCH
Density Threshold:
0.75%
Reset
0%
1%
10%
100%
Show BOS
Hide BOS
Sort by Frequency
Sort by Max Act
Sort by Density
Search Explanations
All
By Release
By Model
By Sources
MODEL
Residual - 9k
All Layers
Show Dashboards
Hide Dashboards
Browse
MODEL
Residual - 9k
LAYER
Features in
GEMMA-2-2B
@
12-temporal-res
Hover over a feature on the left to preview its details.
Click a feature to lock it and interact with it.