Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Gemma-3-4B-IT
    3. 17-GEMMASCOPE-2-RES-65K
    4. 47195
    Prev
    Next
    INDEX
    Explanations

    truth accuracy honesty

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

    This neuron activates strongly for tokens related to truth, accuracy, and honesty, often contrasting them with deception or falsehoods.

    oai_token-act-pair · gemini-2.5-flashTriggered by @ananthanv94
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2-4b-it/resid_post/layer_17_width_65k_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
    an
    3.16
    د
    2.86
    et
    2.59
    o
    2.51
    دة
    2.50
    요
    2.34
    ுகிறது
    2.28
    aní
    2.24
    etary
    2.24
    ம்
    2.16
    POSITIVE LOGITS
    म
    2.18
    и
    2.10
    ৫
    2.06
    _{\
    2.04
     besieged
    1.98
    sning
    1.96
    PERCENT
    1.93
    ことが多い
    1.91
    가를
    1.91
     portrayal
    1.90
    Activations Density 0.072%

    No Known Activations