© 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. 3049
    Prev
    Next
    INDEX
    Explanations

    is good

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

    The neuron fires on prominent content words—especially key action or outcome terms (often participial forms like “changing,” hyphenated metrics like “drop‐off,” or crucial guide words like “service” and “done”) that signal central processes or measures in instructional/business contexts.

    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
    很少
    0.64
    س
    0.61
     жить
    0.58
    en
    0.57
    only
    0.55
    น
    0.55
    ૪
    0.54
    少量
    0.54
     humanos
    0.53
    speople
    0.53
    POSITIVE LOGITS
     deems
    0.59
     needed
    0.59
    ING
    0.58
     happening
    0.57
     causation
    0.55
    မည်
    0.55
     potř
    0.55
     त्यांनी
    0.54
     malfunctioning
    0.54
     überhaupt
    0.54
    Activations Density 0.000%

    No Known Activations