© 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. 34-GEMMASCOPE-RES-131K
    7. 106898
    Prev
    Next
    INDEX
    Explanations

    hero or villain

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

    The neuron broadly activates on the vast majority of words in continuous descriptive or narrative text passages, effectively marking long-form prose rather than isolated keywords.

    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
    ksikon
    -0.99
     žaid
    -0.97
    細い
    -0.92
    ئ
    -0.91
    rophoto
    -0.90
    rikes
    -0.90
    шої
    -0.89
    riceps
    -0.87
    diameter
    -0.87
    ΡΙ
    -0.86
    POSITIVE LOGITS
     hero
    1.65
     protagonist
    1.55
     protagonists
    1.48
    tagonist
    1.38
     heroes
    1.33
     heroic
    1.32
     antagonists
    1.28
    Usually
    1.22
     heroine
    1.18
     villain
    1.16
    Activations Density 0.061%

    No Known Activations