© 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. 31-GEMMASCOPE-2-RES-16K
    4. 10990
    Prev
    Next
    INDEX
    Explanations

    bu, buj, buddy

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

    The neuron selectively fires on relatively long or uncommon content words—especially proper names, brand names, and technical/Latin‐style terms—whereas it stays off on common short 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_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
    raines
    0.63
    तया
    0.60
    никами
    0.59
    cedores
    0.59
     Carrington
    0.55
    కుండా
    0.54
    ➋
    0.54
     Tayler
    0.53
    瞽
    0.53
    Secondary
    0.52
    POSITIVE LOGITS
    apest
    0.61
    ILT
    0.61
    ongiorno
    0.59
    ổi
    0.58
     yüzden
    0.58
     motivo
    0.57
    propion
    0.57
    Bu
    0.57
    ழுகு
    0.55
     hafta
    0.54
    Activations Density 0.339%

    No Known Activations