© 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. 1231
    Prev
    Next
    INDEX
    Explanations

    being subjected to

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

    The neuron strongly activates on main lexical verbs (especially those in multi‐word or “phrasal” constructions like “dwell in,” “come across,” “working with,” “managed to”)—i.e. it’s looking for content words that are the heads of action phrases.

    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
     caracteres
    0.74
     VERY
    0.66
    หรือ
    0.66
     muy
    0.65
     importantes
    0.64
     atrocious
    0.63
     undesirable
    0.63
     susceptibles
    0.61
     potrebbero
    0.61
     annoying
    0.61
    POSITIVE LOGITS
    看到
    0.75
    如此
    0.70
    可以看到
    0.69
    ឃើញ
    0.66
     선보
    0.65
    帶著
    0.65
    每個
    0.64
     reimag
    0.64
    seeing
    0.64
    छोटे
    0.64
    Activations Density 0.299%

    No Known Activations