© 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
    2. Gemma-3-12B
    3. 24-GEMMASCOPE-2-RES-16K
    4. 8037
    Prev
    Next
    INDEX
    Explanations

    break down

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

    This neuron strongly activates on words that denote setting up, completing, or accomplishing tasks—especially the key “action” words or particles in common phrasal-verb and completion phrases (e.g. snuff in “up to snuff,” done in “job done,” up in “set up,” out in “scoped out,” transactions in “legal transactions,” etc.).

    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
     jamás
    0.93
     niemals
    0.86
    !!.
    0.86
    💖
    0.84
    !`
    0.84
    !.
    0.83
    CurrentByte
    0.83
     NEVER
    0.82
     никогда
    0.80
    !;
    0.80
    POSITIVE LOGITS
     prelimin
    1.05
    初步
    1.05
     initial
    1.04
     preliminary
    1.02
     initially
    0.98
     makeshift
    0.89
     negotiations
    0.89
     expedited
    0.85
     agreements
    0.82
     disagreed
    0.80
    Activations Density 0.149%

    No Known Activations