© 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. 10-GEMMASCOPE-RES-131K
    7. 99890
    Prev
    Next
    INDEX
    Explanations

    ignored truths and facts

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

    The neuron fires strongly on words that signal an analytic or epistemic stance—terms related to knowing, evaluating, and asserting facts or truths (e.g. “knowing,” “truth,” “numbers,” “considered,” “looking”).

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_10/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
     not
    -1.76
     with
    -1.53
     more
    -1.24
     for
    -1.14
    Not
    -1.11
     không
    -0.99
     no
    -0.98
    ฉัน
    -0.98
     on
    -0.97
     and
    -0.94
    POSITIVE LOGITS
     roligt
    1.30
     consists
    1.29
    ricanes
    1.22
    risen
    1.22
     Nedir
    1.20
     what
    1.16
     borbo
    1.16
     reaſon
    1.16
     étrang
    1.15
    ifico
    1.15
    Activations Density 0.155%

    No Known Activations