© 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. 31-GEMMASCOPE-2-RES-16K
    4. 6279
    Prev
    Next
    INDEX
    Explanations

    suffix in Spanish, Russian, or Portuguese

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

    The neuron activates on section‐level metadata like repeated page or post titles and category/tag labels (e.g. “Esta entrada se ha publicado en…,” “Artículo,” “Efectos Secundarios,” repeated product names), i.e. it’s looking for document metadata headings rather than body text.

    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_31_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
     اجمع
    0.57
     rented
    0.55
     Rxa
    0.55
     Bekasi
    0.55
     الواحد
    0.55
     डाइजेस्टिव
    0.54
    जनबी
    0.54
     Saviour
    0.54
    jenigen
    0.54
     Kubo
    0.54
    POSITIVE LOGITS
     directe
    1.00
     directa
    0.97
     مستقیم
    0.89
     diretta
    0.88
     singkat
    0.86
     indirect
    0.85
     gradual
    0.84
     direta
    0.80
     rapide
    0.74
     directo
    0.74
    Activations Density 0.110%

    No Known Activations