© 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-3-12B
    5. 31-GEMMASCOPE-2-RES-16K
    7. 8002
    Prev
    Next
    INDEX
    Explanations

    problems and difficulty

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

    The neuron activates strongly on numeric tokens (e.g. multi‐digit answer or question numbers and codes), i.e. it detects numbers.

    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
    敀
    0.60
     টাকায়
    0.58
    硃
    0.57
    elius
    0.56
    恈
    0.55
    唝
    0.55
    显然
    0.54
    ＃
    0.54
     perceptions
    0.53
    ūk
    0.53
    POSITIVE LOGITS
    이너
    0.62
     எடுத்த
    0.58
    ๋า
    0.57
    worked
    0.56
    ทำการ
    0.56
    LoginID
    0.54
    ๋
    0.54
     পরীক্ষা
    0.53
     Baillargeon
    0.53
     sammeln
    0.53
    Activations Density 0.009%

    No Known Activations