© 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. 26286
    Prev
    Next
    INDEX
    Explanations

    how do I

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

    The neuron is keyed to question‐framing words and auxiliaries (e.g. how, is, know, common, list) that signal an interrogative context.

    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
    这也是
    -1.01
    を追加
    -0.92
     antidesliz
    -0.91
     dość
    -0.88
    会有
    -0.87
     przyczyn
    -0.86
     lived
    -0.86
    可能有
    -0.85
     myce
    -0.85
    いち
    -0.84
    POSITIVE LOGITS
     in
    1.05
     to
    0.96
    layın
    0.93
     when
    0.93
     on
    0.91
    čeno
    0.89
     國際
    0.81
     turist
    0.79
     (!
    0.79
    waard
    0.79
    Activations Density 0.124%

    No Known Activations