© 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
    3. Gemma-3-27B-IT
    5. 40-GEMMASCOPE-2-RES-65K
    7. 53424
    Prev
    Next
    INDEX
    Explanations

    problems causing harm

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2-27b-it/resid_post/layer_40_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
     inducing
    0.54
     collided
    0.54
     provoque
    0.52
     saturated
    0.48
     clashed
    0.46
     OnTrigger
    0.45
    遭遇
    0.45
     inundated
    0.45
     provoking
    0.45
     inflicted
    0.45
    POSITIVE LOGITS
     dangerous
    0.57
     Dangerous
    0.56
    Dangerous
    0.55
    dangerous
    0.50
     harmful
    0.49
    Harm
    0.48
     опас
    0.46
     कोणत्याही
    0.46
     خطر
    0.46
    Critical
    0.44
    Activations Density 0.114%

    No Known Activations