© 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. 34-GEMMASCOPE-RES-131K
    4. 109354
    Prev
    Next
    INDEX
    Explanations

    heads and tails

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

    The neuron responds to digits and table‐style layout markers (e.g. runs of numbers, pipes, aligned columns), effectively detecting numeric characters in tabular or columnar text.

    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
    Arco
    -0.80
    Comb
    -0.79
    czegó
    -0.76
     Boug
    -0.75
    顶级
    -0.73
     Arco
    -0.72
     citar
    -0.71
     tvb
    -0.71
    accountId
    -0.71
    lastic
    -0.70
    POSITIVE LOGITS
     heads
    4.56
     Heads
    3.98
    Heads
    3.81
    heads
    3.69
     tails
    3.39
     Tails
    2.77
    tails
    2.55
    Tails
    2.55
     head
    2.50
     Head
    2.31
    Activations Density 0.043%

    No Known Activations