© 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 EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    2. Gemma-2-27B
    3. 22-GEMMASCOPE-RES-131K
    4. 128463
    Prev
    Next
    INDEX
    Explanations

    The neuron fires strongly on single-word, domain‐specific nouns—especially technical or scientific terms (e.g. WEBSITE, TV, Internet, images, insects, limbs, granulocytes).

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

    specific nouns and categories

    np_acts-logits-general · gemini-2.5-flash-lite
    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
     to
    -1.23
     that
    -1.16
     each
    -1.09
     in
    -1.07
    に見える
    -1.06
     of
    -1.05
     ซึ่ง
    -0.98
     potes
    -0.98
     kock
    -0.97
     which
    -0.96
    POSITIVE LOGITS
     etc
    1.73
     or
    1.56
     and
    1.41
    ↵↵
    1.12
    などは
    1.10
    etc
    1.09
    Ẓ
    1.01
     occurs
    0.98
     flera
    0.96
     but
    0.95
    Activations Density 0.071%

    No Known Activations