© 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. 26410
    Prev
    Next
    INDEX
    Explanations

    degrees, law, cultures, ID, root

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

    The neuron activates on tokens containing digits—that is, it flags numeric values and number-like strings.

    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
     føl
    -1.08
     tidligere
    -1.03
     eneste
    -1.02
    гу
    -1.00
    लेख
    -0.96
     здесь
    -0.94
     who
    -0.93
    不是
    -0.93
    工作
    -0.92
    或
    -0.91
    POSITIVE LOGITS
     studia
    1.13
     contento
    0.99
     divino
    0.96
     pantaloni
    0.96
    至於
    0.94
     fortific
    0.93
     salvare
    0.92
    使える
    0.91
     continuare
    0.91
    じゃん
    0.90
    Activations Density 0.120%

    No Known Activations