© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 32737
    Prev
    Next
    INDEX
    Explanations

    contrast and exceptions

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

    The neuron fires on discourse-structuring words—primarily adverbs and conjunctions (things like “just,” “now,” “because,” “when,” “instead,” etc.).

    oai_token-act-pair · o4-mini
    Triggered 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
    Negative Logits
    inducing
    -0.88
     mè
    -0.83
    ＇
    -0.82
    vint
    -0.81
     SOUND
    -0.81
    ⸮
    -0.79
    ૩
    -0.78
    quickly
    -0.78
     kis
    -0.78
     Lbs
    -0.77
    POSITIVE LOGITS
     however
    1.32
     only
    1.31
     even
    1.29
    ただし
    1.23
     modified
    1.20
     merely
    1.17
     However
    1.15
     pouze
    1.13
     allerdings
    1.13
     modify
    1.11
    Activations Density 0.057%

    No Known Activations