© 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
    3. Gemma-3-12B
    5. 24-GEMMASCOPE-2-RES-16K
    7. 8100
    Prev
    Next
    INDEX
    Explanations

    fill in the blank verb completions

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

    The neuron responds strongly to the very first token of a sentence, i.e. sentence-initial (often capitalized) words.

    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
    。
    0.59
    とか
    0.57
     Important
    0.55
     everybody
    0.55
    これも
    0.55
    Important
    0.55
    い
    0.53
    という
    0.52
     Everybody
    0.52
     আগুন
    0.50
    POSITIVE LOGITS
     disapproved
    0.66
     Nxf
    0.59
     stroked
    0.59
    कालय
    0.57
     immunoblot
    0.56
     coerced
    0.56
     zuvor
    0.55
     няко
    0.55
    ulmonary
    0.55
     kontinuier
    0.55
    Activations Density 0.002%

    No Known Activations