© 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. 71202
    Prev
    Next
    INDEX
    Explanations

    role and roles

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

    The neuron fires on any token containing the substring “role,” i.e. it detects occurrences of “role” in code, JSON, or text.

    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
    祧
    -2.66
    嵛
    -2.58
    觏
    -2.45
     gewöhnlich
    -2.42
     häufigsten
    -2.42
     decorados
    -2.39
    稣
    -2.38
    虯
    -2.27
     hervorragend
    -2.27
     命
    -2.27
    POSITIVE LOGITS
    .”
    2.72
    鲼
    2.31
     vap
    2.28
     THEM
    2.28
    
    2.27
    $.
    2.27
    en
    2.25
     所以
    2.25
    cdot
    2.23
    ⡑
    2.23
    Activations Density 0.027%

    No Known Activations