© 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 EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-3-12B
    5. 31-GEMMASCOPE-2-RES-16K
    7. 5624
    Prev
    Next
    INDEX
    Explanations

    suspect it is

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

    The neuron detects tokens commonly used in method‐style or procedural descriptions—things like “step,” “sub-step,” “degree,” “overlap,” “results,” and similar technical/quantitative terms.

    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
    ToDelete
    0.53
    শিক্ষা
    0.52
    AndDelete
    0.50
    罰
    0.48
    handleAdd
    0.48
    執行
    0.47
     শেষ
    0.47
    Consult
    0.46
     কৃষ্ণ
    0.46
    Сі
    0.46
    POSITIVE LOGITS
     sự
    0.86
     bị
    0.83
    มีการ
    0.75
     doped
    0.72
     被
    0.71
     acetylation
    0.69
     marinated
    0.68
    被
    0.67
     gilded
    0.66
     keter
    0.65
    Activations Density 1.902%

    No Known Activations