© 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
    2. Gemma-3-27B-IT
    3. 38-GEMMASCOPE-2-TRANSCODER-262K
    4. 5518
    Prev
    Next
    INDEX
    Explanations

    "privilege" and "advantage". Therefore, explaining its behavior by focusing on these terms is appropriate.The explanation should combine the identified concept: "privilege" and "advantage".Let's check the word count. "privilege and advantage" is 3 words. This fits the 3-20 word requirement.It does not start with forbidden phrases.It does not mention tokens or patterns.It capitalizes no letters.It is specific.Therefore, the output should be "privilege and advantage".privilege and advantage

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2-27b-it/transcoder_all/layer_38_width_262k_l0_small_affine
    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
    功
    0.42
     실수
    0.36
    MoodAnalyser
    0.36
    ستان
    0.35
     কষ্ট
    0.34
     Significance
    0.34
    bahn
    0.34
     सिल्
    0.34
     Cong
    0.34
    蜻
    0.34
    POSITIVE LOGITS
     privileged
    0.56
     privilegi
    0.48
     приви
    0.47
    privileged
    0.45
     बाल
    0.42
     privile
    0.41
     privilege
    0.41
     sew
    0.41
     blast
    0.40
     privil
    0.39
    Activations Density 0.066%

    No Known Activations