© 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. 5363
    Prev
    Next
    INDEX
    Explanations

    concept definition structure

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

    The neuron fires strongly on standalone content‐word tokens that tend to appear as list items or section headings (e.g. “list,” “comics,” “overall,” “according,” “cascading,” “effect,” “dates,” etc.).

    oai_token-act-pair · o4-miniTriggered by @jyhe0408

    periods at the end of sentences or list items in structured text formats.

    oai_token-act-pair · claude-4-5-sonnetTriggered by @jyhe0408

    language describing ordering and structured arrangement of items or actions, such as sequences, lists, organization, and procedural flow.

    oai_token-act-pair · gpt-5Triggered 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
    
    0.71
     inasmuch
    0.70
     ;
    0.60
     satisfactorily
    0.59
    ；
    0.58
     farther
    0.56
     子
    0.56
    ;
    0.55
    。
    0.55
    ――
    0.54
    POSITIVE LOGITS
     rebranded
    0.82
     geopolitical
    0.78
     impactful
    0.77
     disinformation
    0.75
     reimag
    0.75
     viewership
    0.75
     geopol
    0.74
     metaverse
    0.73
     livestream
    0.73
     chiamato
    0.71
    Activations Density 0.070%

    No Known Activations