© 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. 31-GEMMASCOPE-2-RES-16K
    7. 6105
    Prev
    Next
    INDEX
    Explanations

    The neuron strongly activates on words and phrases that denote mistakes, errors, inaccuracies, or the need for corrections.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408

    disclaimers about potential inaccuracies or errors in content or data.

    oai_token-act-pair · claude-4-5-sonnetTriggered by @jyhe0408

    mentions of errors, inaccuracies, or mistakes, especially in contexts of identifying and correcting them.

    oai_token-act-pair · gpt-5Triggered 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
    ార్క్
    0.62
    橇
    0.57
    aktan
    0.56
    వంత
    0.55
     திட
    0.54
    ෑම
    0.52
     सूजन
    0.52
     சக்த
    0.52
     ആവശ
    0.50
    elh
    0.50
    POSITIVE LOGITS
     mistakes
    3.63
     errors
    3.55
     Errors
    3.36
     errores
    3.27
     erros
    3.23
    errors
    3.16
     Mistakes
    3.16
    Errors
    3.08
     Fehler
    3.05
     erreurs
    3.05
    Activations Density 0.629%

    No Known Activations