© 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. 24-GEMMASCOPE-2-RES-16K
    7. 590
    Prev
    Next
    INDEX
    Explanations

    that followed by clause

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

    The neuron fires on high-frequency function words and discourse markers—things like pronouns (“everyone,” “they’re,” “their”), subordinating conjunctions (“because,” “whether”), and similar tokens that often introduce or link 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_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
    2
    0.33
    ="
    0.32
    =
    0.32
    '
    0.32
    '.
    0.31
    1
    0.31
    _
    0.31
    4
    0.31
    3
    0.30
    ).
    0.30
    POSITIVE LOGITS
     ஒவ்வொரு
    0.40
     niemand
    0.40
    there
    0.40
     unscrupulous
    0.38
     there
    0.37
     wszyscy
    0.37
    すべての
    0.36
     هناك
    0.36
     människor
    0.35
     شما
    0.34
    Activations Density 0.957%

    No Known Activations