© 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. 66020
    Prev
    Next
    INDEX
    Explanations

    event titles and subject labels

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

    The neuron is highly sensitive to technical/metadata tokens—things like HTML tags and attributes, URLs, punctuation‐delimited code or menu labels, and similar structural markers.

    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
     our
    -1.23
     we
    -1.07
     aujourd
    -1.03
     your
    -1.01
     сегодня
    -0.99
    今天
    -0.96
     today
    -0.96
    我们
    -0.95
     oggi
    -0.95
    ५
    -0.95
    POSITIVE LOGITS
     offizi
    1.04
    遑
    1.02
    Aç
    1.01
     idéia
    0.98
     стъ
    0.96
     geprü
    0.95
    ayant
    0.94
    ipel
    0.94
     recuperação
    0.93
     didukung
    0.93
    Activations Density 0.009%

    No Known Activations