© 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. 10-GEMMASCOPE-RES-131K
    7. 73622
    Prev
    Next
    INDEX
    Explanations

    Itanium

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

    The neuron strongly activates on very short, often single-character or two-character uppercase tokens (e.g. “H”, “J”, “R”, “If,” “It”), effectively detecting standalone initials or tiny capitalized fragments.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_10/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
    Ketua
    -2.70
    我们
    -2.61
    Kepala
    -2.59
     dekoratif
    -2.58
    鋮
    -2.56
    Sudah
    -2.53
    Untuk
    -2.53
    Bagi
    -2.48
    افز
    -2.48
    今まで
    -2.47
    POSITIVE LOGITS
     they
    2.55
     continúas
    2.53
    桫
    2.38
    “
    2.31
     isn
    2.20
     Then
    2.20
     is
    2.16
    有不少
    2.14
     you
    2.14
     it
    2.14
    Activations Density 0.004%

    No Known Activations