© 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
    2. Gemma-2-27B
    3. 22-GEMMASCOPE-RES-131K
    4. 194
    Prev
    Next
    INDEX
    Explanations

    author documentation

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

    The neuron is firing specifically on numeric literals in code (tokens consisting entirely of digits).

    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
     توضی
    -1.24
    brida
    -1.08
    НИК
    -1.06
    dolo
    -1.02
    ням
    -1.02
    СКА
    -1.02
    [-\
    -1.01
    ''');
    -1.00
    copal
    -1.00
    声優
    -1.00
    POSITIVE LOGITS
     and
    1.23
     at
    1.16
    ٭
    1.05
     makes
    1.05
     has
    1.04
     really
    0.97
     terribly
    0.96
    Kary
    0.95
    Siga
    0.94
    Moth
    0.93
    Activations Density 0.012%

    No Known Activations