© 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. 24-GEMMASCOPE-2-RES-16K
    7. 11469
    Prev
    Next
    INDEX
    Explanations

    avoid jargon

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

    The neuron strongly activates on numeric tokens—especially multi‐digit numbers used in reference metadata (volumes, years, pages, issue numbers) in citations.

    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_24_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
     effectuer
    0.81
    Ré
    0.79
    SHOT
    0.76
     réelle
    0.76
     OFFICER
    0.74
    𝐑
    0.73
     движения
    0.73
     ALWAYS
    0.73
     biens
    0.73
    RENC
    0.72
    POSITIVE LOGITS
    root
    0.71
    向けの
    0.71
    ollowing
    0.69
     Manufacturing
    0.69
    芑
    0.68
    worst
    0.68
     kaam
    0.68
     bauen
    0.68
    klam
    0.67
     Plumbing
    0.67
    Activations Density 0.007%

    No Known Activations