© 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. 34-GEMMASCOPE-RES-131K
    7. 90977
    Prev
    Next
    INDEX
    Explanations

    CNN anchor or correspondent reports

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

    This neuron detects uppercase bylines and speaker/network labels in news transcripts (e.g. reporter names like “Eleanor Beardsley,” networks like “NPR,” “CNN,” and other all‐caps identifiers).

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_34/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
     Нужно
    -0.84
     ਬ
    -0.81
    """.
    -0.80
     ork
    -0.78
    TRAIN
    -0.78
    ${\
    -0.76
    "+"
    -0.76
     ရ
    -0.75
    %+
    -0.73
    advice
    -0.73
    POSITIVE LOGITS
     joining
    0.84
     фаса
    0.77
     closer
    0.77
     reports
    0.77
     correspondent
    0.75
    ounty
    0.75
    ortunadamente
    0.75
    ၄
    0.75
     anchor
    0.74
    phur
    0.74
    Activations Density 0.017%

    No Known Activations