© 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
    3. Gemma-3-12B
    5. 24-GEMMASCOPE-2-RES-16K
    7. 9013
    Prev
    Next
    INDEX
    Explanations

    people who might

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

    The neuron consistently lights up on long, domain-specific noun phrases or technical terms (e.g. “export experience,” “narcissistic personality disorder,” “b2b transactions,” “mild cognitive impairment,” etc.).

    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
     заключается
    0.80
     culminated
    0.77
     CONCLUSIONS
    0.77
     entailed
    0.75
     получилось
    0.72
     necessitates
    0.71
     resulted
    0.70
     hinges
    0.70
     compels
    0.69
     corresponds
    0.69
    POSITIVE LOGITS
     who
    1.10
     quienes
    0.96
    who
    0.91
     whom
    0.86
     or
    0.85
     quien
    0.84
     الذين
    0.80
    하거나
    0.80
     quién
    0.79
     живут
    0.79
    Activations Density 0.142%

    No Known Activations