© 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
    2. Gemma-2-27B
    3. 22-GEMMASCOPE-RES-131K
    4. 75382
    Prev
    Next
    INDEX
    Explanations

    sports player surnames

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

    The neuron most strongly activates on numeric tokens and attached punctuation (digits, years, scores, decimal numbers, and abbreviation dots/apostrophes).

    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
     FUCK
    -1.02
     简介
    -0.98
    ⊶
    -0.96
     Nedir
    -0.94
    鲥
    -0.93
     sekarang
    -0.92
     अब
    -0.91
     WWII
    -0.89
     növény
    -0.89
     介绍
    -0.88
    POSITIVE LOGITS
     return
    0.90
     can
    0.86
     again
    0.85
    su
    0.84
     asked
    0.83
    <eos>
    0.82
     (
    0.81
     returned
    0.81
     follow
    0.79
     added
    0.79
    Activations Density 0.002%

    No Known Activations