© 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. 26534
    Prev
    Next
    INDEX
    Explanations

    a or an followed by a noun

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

    The neuron fires strongly on the very first content word of a news‐item segment—especially the leading article “A” or the opening token of a paragraph/headline.

    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
     -
    -1.05
    著者
    -1.04
     буквы
    -1.02
    躋
    -0.96
    Probably
    -0.94
     panahon
    -0.94
    色々
    -0.93
    rather
    -0.92
    Somehow
    -0.91
     Studied
    -0.91
    POSITIVE LOGITS
     said
    2.08
     previous
    1.38
     says
    1.27
     saying
    1.20
     spokesperson
    1.18
    ᾿
    1.17
    ările
    1.16
     recent
    1.16
     recently
    1.15
     staff
    1.14
    Activations Density 0.026%

    No Known Activations