© 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. 130157
    Prev
    Next
    INDEX
    Explanations

    The neuron is primarily firing on isolated numeric tokens (e.g. years, measurement values) and brief uppercase-coded identifiers (e.g. “SH” in domain names), i.e. metadata‐like numbers and acronyms.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408

    SH3 domains or intent.putExtra

    np_acts-logits-general · gemini-2.5-flash-lite
    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
    先生が
    -1.12
    uer
    -0.94
     иногда
    -0.92
     соответству
    -0.88
     usually
    -0.88
    utf
    -0.86
    }:{
    -0.85
     tambi
    -0.85
    遵
    -0.84
     Usually
    -0.83
    POSITIVE LOGITS
     other
    1.24
    อื่น
    1.17
    other
    0.93
    autres
    0.93
    Paperback
    0.92
     autres
    0.89
     Sloveniji
    0.89
     poni
    0.88
    几种
    0.88
    Radiant
    0.88
    Activations Density 0.025%

    No Known Activations