© 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
    3. Gemma-2-27B
    5. 34-GEMMASCOPE-RES-131K
    7. 117053
    Prev
    Next
    INDEX
    Explanations

    thinking and believing

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

    The neuron strongly activates on Japanese grammatical morphemes—particularly auxiliary verb endings and particles (like の, させ, ようと, たら) that mark causative, speculation, and explanatory constructions.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_34/width_131k
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    Sie
    -0.76
     definitely
    -0.73
    ulum
    -0.69
    وب
    -0.69
     tenemos
    -0.68
    hängigkeit
    -0.67
    다는
    -0.66
    hena
    -0.66
    oplastic
    -0.66
    بة
    -0.66
    POSITIVE LOGITS
     think
    2.55
    以为
    2.47
    以為
    2.33
     believed
    2.19
     thought
    2.16
     believe
    2.09
     thinks
    1.99
    认为
    1.90
     believing
    1.89
    think
    1.87
    Activations Density 0.023%

    No Known Activations