© 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. 15273
    Prev
    Next
    INDEX
    Explanations

    dialogue, quotes

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

    The neuron fires on tokens that mark who’s speaking—that is, speaker‐attribution labels or handles (e.g. character names like “Fuego,” IRC/chat tags like “<live>,” “asked,” “said,” etc.).

    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
     namin
    -0.99
    GER
    -0.96
     bensin
    -0.91
    ош
    -0.90
     standart
    -0.86
     and
    -0.85
     kemenangan
    -0.84
     pertandingan
    -0.84
     naruto
    -0.83
    جمة
    -0.81
    POSITIVE LOGITS
    ANIUM
    0.88
     okam
    0.85
     object
    0.85
    лік
    0.85
    
    0.84
     suddenly
    0.84
    Reak
    0.83
     tuvo
    0.82
    otriva
    0.80
    してしまう
    0.79
    Activations Density 0.005%

    No Known Activations