© 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. 14333
    Prev
    Next
    INDEX
    Explanations

    tutorial hell, negative events, undesirable traits

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

    The neuron strongly activates on the very first token of a new instructional step or sentence (i.e. the start‐of‐step/step‐number header).

    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
    р
    0.78
    цкі
    0.76
     लोगो
    0.75
     cargas
    0.75
    м
    0.74
    あった
    0.71
    то
    0.71
    точные
    0.71
     человека
    0.70
     Побе
    0.69
    POSITIVE LOGITS
     અમા
    0.85
     assort
    0.83
     empê
    0.81
     sele
    0.80
     sul
    0.79
    频道
    0.78
     selezione
    0.77
    集團
    0.77
     kami
    0.75
    專
    0.75
    Activations Density 0.000%

    No Known Activations