© 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. 22-GEMMASCOPE-RES-131K
    7. 70707
    Prev
    Next
    INDEX
    Explanations

    when you get this

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

    The neuron fires on numeric and punctuation tokens that form timestamp‐style markers (e.g. “0:04”, “0:09”), i.e. time codes.

    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
     gangen
    -1.00
    ceğiniz
    -0.96
     indien
    -0.94
     frambo
    -0.94
    tungs
    -0.92
    getvalue
    -0.90
     कृपया
    -0.90
    🎲
    -0.90
    */
    -0.89
     hulle
    -0.88
    POSITIVE LOGITS
     how
    0.97
     why
    0.91
     who
    0.90
    定食
    0.89
     Who
    0.88
     suceder
    0.87
     очень
    0.86
     what
    0.85
     Abstract
    0.84
    ⽉
    0.82
    Activations Density 0.009%

    No Known Activations