© 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
    2. Gemma-3-12B
    3. 24-GEMMASCOPE-2-RES-16K
    4. 11875
    Prev
    Next
    INDEX
    Explanations

    actions, explanations, or descriptions

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

    The neuron responds to uncommon, content-heavy tokens (i.e. low-frequency or specialized words) rather than frequent function words.

    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
     extraordinaire
    0.57
     auspices
    0.54
    Ј
    0.54
    
    0.54
    http
    0.54
     extravagance
    0.54
    J
    0.52
     גע
    0.51
     здесь
    0.50
     venerable
    0.50
    POSITIVE LOGITS
     onboarding
    0.87
     tiktok
    0.73
    डिंग
    0.72
     iniziale
    0.71
    ছোট
    0.69
     TikTok
    0.68
     Messaging
    0.68
     Technically
    0.68
     Figma
    0.67
     Trainings
    0.66
    Activations Density 0.031%

    No Known Activations