© 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. 26796
    Prev
    Next
    INDEX
    Explanations

    linking words and phrases

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

    This neuron detects direct second-person address—words and phrases (like “you,” “are,” “here,” “And”) that directly speak to or guide the reader.

    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
     some
    -1.46
     most
    -1.40
     idiota
    -1.30
     many
    -1.25
     cortina
    -1.13
     discrepan
    -1.13
     bicicleta
    -1.11
     anivers
    -1.11
     fritas
    -1.11
    編曲
    -1.09
    POSITIVE LOGITS
     również
    1.41
    Caracteristici
    1.26
    brate
    1.15
    brates
    1.13
     essenziale
    1.13
     estremamente
    1.12
     forskjellige
    1.10
     tillegg
    1.10
    円で
    1.09
     esterno
    1.09
    Activations Density 0.000%

    No Known Activations