© 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
    3. Gemma-3-12B
    5. 31-GEMMASCOPE-2-RES-16K
    7. 5755
    Prev
    Next
    INDEX
    Explanations

    several translations

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

    The neuron fires most strongly on the first meaningful word of a sentence, effectively marking sentence beginnings.

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

    the phrase "there are" or similar existential constructions.

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

    phrases indicating existence or quantity, often used to introduce lists, counts, or general statements in expository text.

    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_31_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
     Hälfte
    0.61
     այդ
    0.58
    これで
    0.56
     misalnya
    0.52
    足以
    0.52
    icama
    0.52
    ៕
    0.51
    ̮
    0.50
    饶
    0.50
     siquiera
    0.50
    POSITIVE LOGITS
     several
    2.88
    several
    2.69
     нескольких
    2.50
     няколко
    2.47
     plusieurs
    2.39
    几种
    2.39
    いくつかの
    2.39
    Several
    2.38
    いくつか
    2.38
     عدة
    2.34
    Activations Density 0.754%

    No Known Activations