© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-3-12B
    5. 31-GEMMASCOPE-2-RES-16K
    7. 8742
    Prev
    Next
    INDEX
    Explanations

    race and horses

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

    The neuron fires on the substring “rac” in words like “race,” “racing,” or “racial,” i.e. it detects any mention of race (whether a competition or the concept of race).

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2-12b-pt/resid_post/layer_31_width_16k_l0_medium
    Prompts (Dashboard)
    392,802 prompts, 256 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    seamless
    0.63
    生日
    0.59
    masına
    0.59
    不定
    0.59
    邯
    0.59
     रीजन
    0.57
    Spin
    0.56
    Seamless
    0.56
    권
    0.55
    캘
    0.55
    POSITIVE LOGITS
    horse
    0.87
     Race
    0.82
    horses
    0.80
     races
    0.79
     Races
    0.78
     race
    0.77
     horse
    0.73
    course
    0.70
    Horse
    0.68
     horses
    0.65
    Activations Density 0.147%

    No Known Activations