© 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
    2. Gemma-3-12B
    3. 31-GEMMASCOPE-2-RES-16K
    4. 1633
    Prev
    Next
    INDEX
    Explanations

    by category

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

    The neuron fires strongly on action words and short phrases that mark steps in a narrated sequence (e.g. “ready,” “go,” “kayaking,” “loaded,” “on the water,” the sentence-ending “.”), so it’s detecting procedural or event-progression cues in a story.

    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
     verdiği
    0.55
     अप्लाई
    0.54
    reb
    0.52
    uyendo
    0.52
    arez
    0.51
    사장
    0.51
    꿍
    0.50
     து
    0.50
    ಖ್ಯ
    0.50
    ጣ
    0.50
    POSITIVE LOGITS
     by
    2.67
     By
    2.31
    By
    2.23
     oleh
    2.19
     توسط
    1.88
    โดย
    1.87
    by
    1.84
     entro
    1.63
     soon
    1.62
    によって
    1.55
    Activations Density 0.731%

    No Known Activations