© 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. 31-GEMMASCOPE-2-RES-16K
    4. 215
    Prev
    Next
    INDEX
    Explanations

    He knows

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

    The neuron primarily fires on third-person plural pronouns (e.g. “they,” “they’re”) and similar generic references to others.

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

    the pronoun "they" when it refers to people or organizations.

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

    references to groups via the third-person plural pronoun (especially “they”).

    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
    <0x0D>
    0.38
     mediante
    0.36
     hingga
    0.35
       
    0.34
     כאשר
    0.34
     কর্তৃক
    0.33
     upang
    0.33
     тощо
    0.33
              
    0.32
     $\
    0.32
    POSITIVE LOGITS
    ってる
    0.57
    真的是
    0.55
     didn
    0.55
     weren
    0.55
     kinda
    0.54
     정말
    0.54
    してる
    0.53
     REALLY
    0.53
     진짜
    0.52
     definitely
    0.51
    Activations Density 0.702%

    No Known Activations