© 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. 68883
    Prev
    Next
    INDEX
    Explanations

    code libraries and historical context

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

    The neuron is detecting numeric tokens—sequences of digits like version numbers, years, indices, or other number literals.

    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
     vistos
    -0.88
    ่วย
    -0.86
     стандарт
    -0.84
    穫
    -0.84
     вашем
    -0.83
    kuma
    -0.83
     любых
    -0.82
    front
    -0.82
    nargs
    -0.82
    thag
    -0.80
    POSITIVE LOGITS
     вже
    0.96
    ۟
    0.94
     рішення
    0.91
    mbitos
    0.91
    engkapnya
    0.88
     not
    0.88
    droje
    0.87
    Maa
    0.86
    demort
    0.85
     验证
    0.85
    Activations Density 0.013%

    No Known Activations