© 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. 65704
    Prev
    Next
    INDEX
    Explanations

    but instead rather

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

    The neuron flags “location” tokens—whether geographical place names (e.g. California, New York) or system/file‐path roots (e.g. root, usr, lib).

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

    file paths and directory locations, especially when referring to specific system or configuration directories.

    oai_token-act-pair · claude-4-5-sonnetTriggered by @jyhe0408

    phrases referring to locations or containers, especially filesystem directories/paths, project roots, code packages, and geographic places.

    oai_token-act-pair · gpt-5Triggered 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
     including
    -1.52
     especially
    -1.32
    尤其
    -1.26
     insbesondere
    -1.21
     consisting
    -1.05
     from
    -1.02
     may
    -1.00
    境的
    -1.00
     re
    -1.00
     particularly
    -1.00
    POSITIVE LOGITS
     rather
    2.56
     instead
    2.56
     but
    2.14
    instead
    2.13
    rather
    2.03
    而不是
    1.93
     whereas
    1.90
    Instead
    1.84
    ではなく
    1.78
     plutôt
    1.77
    Activations Density 0.095%

    No Known Activations