© 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
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 128538
    Prev
    Next
    INDEX
    Explanations

    The neuron strongly activates on numeric tokens—especially digits and multi‐digit numbers used as section, rule, or citation identifiers.

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

    rules and statements

    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
     chaste
    -1.09
     undetermined
    -1.08
    jectures
    -1.05
     '';
    -0.99
    fié
    -0.98
    மைய
    -0.98
     unul
    -0.98
    Opere
    -0.97
    背影
    -0.96
     some
    -0.95
    POSITIVE LOGITS
     says
    1.30
    が悪
    1.27
     requirement
    1.20
     requirements
    1.20
     is
    1.20
     section
    1.16
    曰
    1.14
    が悪い
    1.10
     provides
    1.09
    contains
    1.08
    Activations Density 0.033%

    No Known Activations