© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-3-12B
    5. 31-GEMMASCOPE-2-RES-16K
    7. 9763
    Prev
    Next
    INDEX
    Explanations

    front of, terms of

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

    The neuron strongly activates on legal‐contract phrasing, especially on occurrences of “terms of” (and related stock legal words like “agreement,” “warranties,” “sever,” etc.).

    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
     मल्टीप्लिकेशन
    0.61
     hypoc
    0.59
     накопи
    0.58
     tetap
    0.55
    鬥
    0.55
     ligature
    0.55
     mengak
    0.55
    ிருக்க
    0.54
    ঠিক
    0.54
    ﺛ
    0.54
    POSITIVE LOGITS
    Term
    0.81
     Term
    0.71
     termini
    0.63
    Terms
    0.62
    terms
    0.61
     टर्म
    0.61
     TERMIN
    0.61
    میری
    0.59
     terms
    0.59
     term
    0.57
    Activations Density 0.118%

    No Known Activations