© 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. 67593
    Prev
    Next
    INDEX
    Explanations

    have a/an/the/some

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

    The neuron fires on three- or four-digit numeric tokens (e.g. “520,” “656,” “374,” etc.), i.e. it detects standalone large numbers like user-reputation scores.

    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
     οποία
    -2.17
    pergillus
    -1.89
     afspraak
    -1.86
     jego
    -1.85
     verkoop
    -1.82
     sommige
    -1.77
     verdieping
    -1.74
    unción
    -1.67
    corrhi
    -1.61
     czemu
    -1.59
    POSITIVE LOGITS
     a
    4.31
     an
    2.44
     to
    2.11
     some
    1.95
     another
    1.69
     such
    1.64
     something
    1.59
    了一个
    1.56
     more
    1.52
    かれています
    1.52
    Activations Density 0.186%

    No Known Activations