© 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. 82369
    Prev
    Next
    INDEX
    Explanations

    hostage taking and hijackings

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

    The neuron is highly responsive to numeric tokens—especially multi-digit numbers such as years and similar numeric data—in the text.

    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
     сбо
    -0.73
    ्यू
    -0.71
    رژی
    -0.70
     apesar
    -0.69
     accusing
    -0.69
    好吃的
    -0.69
     sourcing
    -0.68
     ハン
    -0.68
    pakah
    -0.67
    ありがとうございます
    -0.67
    POSITIVE LOGITS
     hostage
    2.20
     hostages
    2.03
     siege
    1.33
     seizure
    1.21
     Host
    1.20
     seized
    1.17
     seize
    1.07
     rescue
    1.07
     hijack
    1.05
     зах
    1.01
    Activations Density 0.026%

    No Known Activations