© 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. 11410
    Prev
    Next
    INDEX
    Explanations

    mistakes and errors

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

    The neuron flags words that mark key problem or event statements—like “error,” “lucky,” “hour,” or “else”—i.e. tokens signaling critical incidents or pivots in a report or narrative.

    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
    越來越
    0.74
     étab
    0.72
    Experiences
    0.68
     //-->
    0.67
    ہے
    0.66
    Satisfaction
    0.66
     établ
    0.66
     стран
    0.65
    我們要
    0.65
     असेल
    0.65
    POSITIVE LOGITS
     inadvert
    1.41
     mistakenly
    1.41
     inadvertently
    1.33
     실수
    1.33
     mistake
    1.31
     mistakes
    1.27
     unknowingly
    1.25
     unintentionally
    1.23
     unwittingly
    1.23
    พลาด
    1.20
    Activations Density 0.586%

    No Known Activations