© 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. 6923
    Prev
    Next
    INDEX
    Explanations

    The neuron strongly activates on tokens that begin a new chunk—especially words starting with a capital letter or immediately following punctuation (e.g. the first subtoken of a sentence, clause, or after symbols like “/”).

    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
     एप्लिकेशन
    0.78
     μην
    0.69
     Sym
    0.67
    Đi
    0.67
    सीमा
    0.67
    ل
    0.65
     இ
    0.64
    ار
    0.63
     ”
    0.63
    тивы
    0.62
    POSITIVE LOGITS
     materialize
    0.80
     hectares
    0.76
    דים
    0.70
     télécharger
    0.70
     categorize
    0.68
     monetize
    0.68
    ખ્ય
    0.68
     énorm
    0.67
     Visualize
    0.67
     sorprendente
    0.67
    Activations Density 0.055%

    No Known Activations