© 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. 92248
    Prev
    Next
    INDEX
    Explanations

    URLs and dates

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

    The neuron fires on tokens containing digits—e.g. years (“2015”), numeric literals, or digit‐rich identifiers—so it’s essentially spotting any token with numbers in it.

    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
    本体
    -0.84
     систему
    -0.82
    放送
    -0.82
     hej
    -0.82
     pani
    -0.82
     parfait
    -0.82
    ｂ
    -0.78
     Studi
    -0.78
    adaptiveStyles
    -0.77
    WSTR
    -0.75
    POSITIVE LOGITS
     crucible
    0.85
     ciertas
    0.85
     večer
    0.84
    ฒ
    0.80
     ži
    0.80
    יתה
    0.80
    כום
    0.78
    回來
    0.78
    ドキ
    0.77
     pokra
    0.76
    Activations Density 0.002%

    No Known Activations