© 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
    2. Gemma-3-12B
    3. 24-GEMMASCOPE-2-RES-16K
    4. 854
    Prev
    Next
    INDEX
    Explanations

    The neuron strongly activates on common sentence-initial words (e.g. “This,” “You,” “In,” “Here,” “Download,” etc.), i.e. it’s detecting the first word of a new sentence or segment.

    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
    И
    0.43
    У
    0.40
    渗
    0.40
    -
    0.40
     via
    0.40
     directly
    0.39
     depending
    0.39
     alongside
    0.39
     μέχρι
    0.39
    ה
    0.39
    POSITIVE LOGITS
    jenigen
    0.55
    anwhile
    0.50
    clusion
    0.49
    ũng
    0.48
    ůže
    0.45
    ákat
    0.44
    umé
    0.43
    ítás
    0.42
    ala
    0.41
    Ꭳ
    0.41
    Activations Density 1.012%

    No Known Activations