© 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. 22-GEMMASCOPE-RES-131K
    7. 35277
    Prev
    Next
    INDEX
    Explanations

    safe, harmless, non-toxic

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

    The neuron detects words and phrases asserting safety or harmlessness (e.g. “safe,” “secure,” “low toxicity,” “biodegradable”).

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_22/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
    ulitis
    -1.13
     uitzicht
    -1.05
    inage
    -1.01
    tola
    -0.97
    nadequate
    -0.96
    zös
    -0.96
    ització
    -0.92
    ぐれ
    -0.91
     Futebol
    -0.90
    leyball
    -0.90
    POSITIVE LOGITS
     safe
    1.81
     harmless
    1.47
    safe
    1.40
     safety
    1.38
     felpa
    1.21
     keinerlei
    1.20
     children
    1.18
     safely
    1.16
     benign
    1.14
     problemlos
    1.14
    Activations Density 0.041%

    No Known Activations