© 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. 128470
    Prev
    Next
    INDEX
    Explanations

    The neuron fires strongly on ALL-CAPS warning tokens—especially “SPOILERS”—i.e. it detects uppercase spoiler-warning phrases.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408

    spoiling or spoilers

    np_acts-logits-general · gemini-2.5-flash-lite
    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
    afari
    -1.03
     validators
    -1.00
    继
    -0.98
    Merhaba
    -0.97
    tigheter
    -0.94
     लगा
    -0.93
     Hieronymus
    -0.92
    繼
    -0.92
     arī
    -0.92
    Tetrach
    -0.90
    POSITIVE LOGITS
     spoiling
    0.96
     spoiler
    0.94
     bowiem
    0.94
    zegovina
    0.93
    isait
    0.92
    ñata
    0.92
    珈琲
    0.91
     révé
    0.90
    belongs
    0.90
    dding
    0.90
    Activations Density 0.013%

    No Known Activations