© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 34-GEMMASCOPE-RES-131K
    7. 56012
    Prev
    Next
    INDEX
    Explanations

    guardian, angel, protector

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

    The neuron activates on instances of the word “Guardian” (especially in legal headings and titles).

    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
    krieg
    -0.93
    🇿
    -0.92
     incendio
    -0.90
     manches
    -0.89
    decken
    -0.89
     expire
    -0.85
     Lazio
    -0.85
     laquelle
    -0.84
    ļa
    -0.84
    endorong
    -0.84
    POSITIVE LOGITS
     angel
    1.62
     angels
    1.52
     Angel
    1.45
     Angels
    1.26
    天使
    1.22
    angels
    1.15
    Angel
    1.09
     guardian
    1.05
     Guardian
    1.00
    angel
    0.99
    Activations Density 0.011%

    No Known Activations