© 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. 31-GEMMASCOPE-2-RES-16K
    7. 156
    Prev
    Next
    INDEX
    Explanations

    The AI neural network

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

    The neuron fires strongly on the very first token of a new paragraph or section—i.e. the leading word at the start of each block.

    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
    。
    0.44
    ”。
    0.41
     和
    0.39
     餐
    0.39
     初始化
    0.38
     的
    0.38
    、
    0.37
     の
    0.37
    Img
    0.36
     인
    0.36
    POSITIVE LOGITS
     onus
    0.65
    oretically
    0.56
     majority
    0.55
     impetus
    0.53
     jury
    0.52
     situation
    0.52
     outpouring
    0.51
     repercussions
    0.50
     uproar
    0.50
     outcry
    0.50
    Activations Density 1.182%

    No Known Activations