© 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. 31-GEMMASCOPE-2-RES-16K
    7. 10418
    Prev
    Next
    INDEX
    Explanations

    if

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

    The neuron strongly activates on subordinating conjunctions (e.g. “even if,” “when,” “not when”) that introduce dependent clauses.

    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_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
     Ако
    0.63
    oretically
    0.61
     ако
    0.60
     konnten
    0.58
    𝗴
    0.58
    oke
    0.58
     હતો
    0.58
    скольку
    0.57
    ᱫ
    0.57
     damals
    0.57
    POSITIVE LOGITS
    それを
    0.75
    それが
    0.73
     said
    0.72
    said
    0.71
    実際
    0.69
    Said
    0.68
    其余
    0.68
    むしろ
    0.66
    それは
    0.66
    所述
    0.65
    Activations Density 0.181%

    No Known Activations