© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 10-GEMMASCOPE-RES-131K
    7. 9888
    Prev
    Next
    INDEX
    Explanations

    evade and invade

    np_acts-logits-general · gemini-2.5-flash-lite

    The neuron detects verbs (and related forms) meaning “to avoid or escape,” such as evade, elude, and evasion.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_10/width_131k
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    !"
    -2.86
    ?"
    -2.38
    N
    -2.33
     allgeme
    -2.31
    And
    -2.17
    J
    -2.14
    "
    -2.13
    出了一
    -2.11
     e
    -2.09
     $
    -2.09
    POSITIVE LOGITS
    ت
    2.80
    柃
    2.42
     twa
    2.38
    賅
    2.36
    
    2.31
     flink
    2.28
     and
    2.28
     stap
    2.27
     salat
    2.22
    嗫
    2.22
    Activations Density 0.004%

    No Known Activations