© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 93054
    Prev
    Next
    INDEX
    Explanations

    reporting statements by experts

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

    The neuron detects reporting-speech attribution markers—i.e. verbs and labels like “said,” “argue,” “told,” or speaker identifiers that introduce quotations.

    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
     gesti
    -1.15
     poète
    -1.11
     movi
    -1.09
     スカート
    -1.04
     Skyscanner
    -1.02
     VOLTAGE
    -1.00
     fotógrafo
    -1.00
     -----
    -0.98
    :{
    -0.98
     ciasc
    -0.97
    POSITIVE LOGITS
     expert
    1.43
     industry
    1.32
     former
    1.26
     seasoned
    1.23
     Mike
    1.23
     longtime
    1.18
     analysis
    1.18
     veteran
    1.16
     Michael
    1.13
     experts
    1.13
    Activations Density 0.022%

    No Known Activations