© 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. 9107
    Prev
    Next
    INDEX
    Explanations

    plot devices

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

    The neuron activates on words in explicit definitional contexts—i.e. sentences of the form “X is a Y” (or “X are Y”) that introduce or define a concept.

    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
     AKA
    1.09
     refers
    1.08
     aka
    1.07
     주로
    0.93
    也就是
    0.93
    被称为
    0.91
     involves
    0.91
     नामक
    0.90
    指的是
    0.90
    叫做
    0.89
    POSITIVE LOGITS
     hätte
    0.72
     そして
    0.71
     sábado
    0.70
     sobot
    0.68
     SmackDown
    0.68
    ábado
    0.68
     Haunted
    0.67
     FIGHT
    0.67
     дочь
    0.67
     localidades
    0.67
    Activations Density 0.050%

    No Known Activations