© 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. 64969
    Prev
    Next
    INDEX
    Explanations

    official actions and policies

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

    The neuron strongly activates on numerical and date expressions (including cardinal numbers and timestamps) common in news‐style reporting.

    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
    咂
    -1.05
    ори
    -1.03
    }}=$
    -1.02
     сахара
    -1.01
    Preface
    -1.01
    Ș
    -0.97
    ḻ
    -0.96
    ющая
    -0.95
    김
    -0.93
    𝐘
    -0.92
    POSITIVE LOGITS
     prevention
    1.60
     safety
    1.41
     similar
    1.36
     lesson
    1.29
     lessons
    1.27
     improved
    1.25
     preventive
    1.24
     future
    1.20
     subsequent
    1.20
     security
    1.18
    Activations Density 0.066%

    No Known Activations