© 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. 93369
    Prev
    Next
    INDEX
    Explanations

    questions, Q:

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

    The neuron is highly responsive to numeric tokens (i.e. standalone numbers and numeric strings) in the 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
    🐠
    -0.98
    rahi
    -0.96
    夯
    -0.91
    恭喜
    -0.88
    дено
    -0.88
    钛
    -0.88
    reuse
    -0.85
    卧槽
    -0.84
    うさぎ
    -0.83
    ᒪ
    -0.82
    POSITIVE LOGITS
     then
    0.95
     would
    0.93
     so
    0.93
     even
    0.85
    猾
    0.84
     verdadera
    0.84
     still
    0.82
     nonetheless
    0.81
     should
    0.79
     techniques
    0.79
    Activations Density 0.003%

    No Known Activations