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

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW
    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    2. Gemma-2-27B
    3. 22-GEMMASCOPE-RES-131K
    4. 129907
    Prev
    Next
    INDEX
    Explanations

    The neuron fires on short three‐letter character runs in the middle of longer identifiers or words (i.e. mid‐word trigrams).

    oai_token-act-pair · o4-miniTriggered by @jyhe0408

    legal, administrative, scientific, and technical terms

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

    programming and technical terms in computer science and software documentation.

    oai_token-act-pair · claude-3-7-sonnet-20250219Triggered by @neilrathi
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_22/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
    ():
    -1.16
     and
    -1.14
     giud
    -1.09
    .],
    -1.09
    !(
    -1.05
     meravigli
    -1.05
    
    -1.04
    吅
    -1.04
    >';
    -1.03
    Ogni
    -1.03
    POSITIVE LOGITS
     gaily
    1.12
     legales
    1.11
     sundry
    1.07
    总之
    1.05
     those
    1.02
     loath
    0.98
     administrativos
    0.97
    所以在
    0.94
    I
    0.94
     hairst
    0.93
    Activations Density 0.603%

    No Known Activations