© 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
    2. Gemma-3-12B
    3. 24-GEMMASCOPE-2-RES-16K
    4. 12208
    Prev
    Next
    INDEX
    Explanations

    what I can do

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

    This neuron isn’t looking for any particular words or meanings but rather for where you are in a long document—it tracks token position (e.g. activating strongly at characteristic offsets in the text).

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

    the indefinite article "an" before nouns or adjectives.

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

    uses of the indefinite article introducing singular noun phrases, especially in “such a …” constructions.

    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
    jų
    0.66
    一種
    0.64
    l
    0.62
     み
    0.59
     శరీర
    0.59
    ทุกคน
    0.58
    șa
    0.56
     আত্মা
    0.54
    我自己
    0.54
    ج
    0.54
    POSITIVE LOGITS
     milestone
    0.74
     endeav
    0.72
     kudos
    0.71
     savvy
    0.71
     setup
    0.70
     modalidad
    0.70
     undertaken
    0.69
     lineup
    0.69
    针对
    0.68
     designee
    0.68
    Activations Density 0.115%

    No Known Activations