© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    2. Gemma-3-12B-IT
    3. 12-GEMMASCOPE-2-RES-16K
    4. 4457
    Prev
    Next
    INDEX
    Explanations

    This neuron appears to detect or predict tokens related to programming, specifically Java and potentially other languages, often appearing in contexts describing data handling, authentication, or function definitions. The inclusion of foreign language fragments in the top logits suggests it might also have a broad sensitivity to linguistic components that are not standard English, possibly indicating its role in handling diverse text inputs or as part of a more general text processing mechanism.Let's refine

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2-12b-it/resid_post/layer_12_width_16k_l0_medium
    Prompts (Dashboard)
    238,145 prompts, 512 tokens each
    Dataset (Dashboard)
    lmsys + oasst1
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    았
    1.15
     hues
    1.09
    を集
    1.05
     showers
    1.05
    ्स
    1.04
     amines
    1.03
     endeavours
    1.03
    其他
    1.02
     operatives
    1.02
    はこの
    1.01
    POSITIVE LOGITS
    ί
    1.05
    pués
    1.04
    ęcie
    1.04
    annya
    1.02
    endereco
    1.00
    laut
    0.99
    énération
    0.98
    -}
    0.97
    ifference
    0.96
    poetry
    0.95
    Activations Density 0.000%

    No Known Activations