© 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
    3. Gemma-2-27B
    5. 10-GEMMASCOPE-RES-131K
    7. 59300
    Prev
    Next
    INDEX
    Explanations

    Pittsfield, Pittsboro, Ruger, Harrington, angiotensin

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

    specific place names, especially cities or locations.

    oai_token-act-pair · claude-3-7-sonnet-20250219Triggered by @neilrathi

    The neuron fires most strongly on unusually long or low‐frequency word fragments or proper‐name substrings (i.e. rare, specialized tokens).

    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
    ’
    -3.00
    渽
    -2.91
    Going
    -2.69
    What
    -2.67
    Because
    -2.63
    Like
    -2.48
    Putting
    -2.44
    During
    -2.41
    Getting
    -2.38
    Surprisingly
    -2.38
    POSITIVE LOGITS
    {
    2.58
    えっ
    2.34
    ”,
    2.31
    鶿
    2.31
    瞠
    2.25
    on
    2.23
    “,
    2.20
    桕
    2.19
    恍
    2.19
    ldon
    2.19
    Activations Density 0.033%

    No Known Activations