© 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. 130877
    Prev
    Next
    INDEX
    Explanations

    The neuron strongly activates on number tokens (e.g. years, counts) and other digit-containing items, i.e. it’s a “number detector.”

    oai_token-act-pair · o4-miniTriggered by @jyhe0408

    past time references

    np_acts-logits-general · gemini-2.5-flash-lite
    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
    /
    -0.93
    了一眼
    -0.93
     my
    -0.93
    -
    -0.92
    ky
    -0.91
    p
    -0.90
     hogyan
    -0.90
    _
    -0.87
    pp
    -0.87
     fish
    -0.85
    POSITIVE LOGITS
     earlier
    3.19
     last
    2.86
     previously
    2.83
     previous
    2.70
    previously
    2.27
     recently
    2.17
    Earlier
    2.13
    earlier
    2.08
     ранее
    2.03
     предыду
    2.03
    Activations Density 0.119%

    No Known Activations