© 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. 33172
    Prev
    Next
    INDEX
    Explanations

    questions about tasks or specific topics

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

    This neuron fires on connective words and action markers—particularly prepositions (in, to, for, while) and gerund/participle forms (e.g. taking)—that often appear in procedural or question contexts.

    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
    signable
    -1.20
    pergillus
    -1.13
     Algunos
    -1.13
    戋
    -1.12
    unción
    -1.09
     opérateurs
    -1.07
    nių
    -1.07
    ignées
    -1.07
     حين
    -1.07
    corbic
    -1.07
    POSITIVE LOGITS
     it
    1.39
     of
    1.27
     these
    1.09
     from
    1.08
     unruly
    1.06
    咐
    1.05
     ALL
    1.05
    ?
    1.04
     antiquated
    1.04
     any
    1.03
    Activations Density 0.036%

    No Known Activations