© 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. 22-GEMMASCOPE-RES-131K
    7. 25850
    Prev
    Next
    INDEX
    Explanations

    file extensions in URLs

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

    The neuron fires on long alphanumeric identifier tokens—e.g. multi‐digit numbers, hash strings, file names, DOIs or URLs—such as commit hashes, DOIs, PDF links, and similar IDs.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    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
     dessa
    -1.01
     of
    -0.96
     etc
    -0.95
     از
    -0.93
    ла
    -0.90
    wes
    -0.90
     IS
    -0.90
     клас
    -0.88
    /*!
    -0.87
    住宿
    -0.87
    POSITIVE LOGITS
     participado
    1.09
    ________,
    1.03
    ㌢
    1.03
    Tja
    1.03
    Svar
    1.02
     cek
    1.02
    Amical
    1.02
    Abon
    1.01
    Kako
    1.00
    Fluffy
    0.98
    Activations Density 0.037%

    No Known Activations