© 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
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 130866
    Prev
    Next
    INDEX
    Explanations

    The neuron strongly activates on words that launch math‐problem instructions—imperative task verbs like “Solve,” “Sort,” “Calculate,” or “What is…” at the start of a question.

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

    Solve equations or calculations

    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
     heo
    -1.27
    pso
    -1.27
    쐈
    -1.25
    nother
    -1.23
     sê
    -1.23
    opies
    -1.17
     cliquer
    -1.16
    ଝ
    -1.16
    teten
    -1.16
    %%%%%%%%
    -1.15
    POSITIVE LOGITS
     only
    1.44
     M
    1.22
     when
    1.14
     J
    1.13
     D
    1.13
     even
    1.12
     K
    1.09
     N
    1.08
     F
    1.07
     G
    1.07
    Activations Density 0.045%

    No Known Activations