© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 34-GEMMASCOPE-RES-131K
    7. 59141
    Prev
    Next
    INDEX
    Explanations

    invalidates

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

    The neuron fires strongly on technical “action/process” words—i.e. verbs and noun‐forms carrying suffixes like –ing, –ation, or –ates.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_34/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
    ゴー
    -0.71
    ület
    -0.71
    řízení
    -0.69
     Temporary
    -0.68
    パパ
    -0.67
    awia
    -0.67
    どうしても
    -0.67
    ローマ
    -0.66
     Orleans
    -0.66
     Terry
    -0.65
    POSITIVE LOGITS
     invalid
    4.66
     invalidate
    4.44
    invalid
    3.73
    Invalid
    3.52
     Invalid
    3.50
    invalidate
    3.36
    Invalidate
    3.11
    INVALID
    3.06
     inval
    3.03
     INVALID
    2.94
    Activations Density 0.078%

    No Known Activations