© 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. 67181
    Prev
    Next
    INDEX
    Explanations

    discharge, wave, energy, sized, yield, resistant, gate

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

    The neuron selectively fires on tokens that are negative numeric literals (i.e. a hyphen directly followed by digits).

    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
     --
    -1.30
    t
    -1.27
    a
    -1.20
    o
    -1.18
     [...]
    -1.17
    ‐
    -1.16
     is
    -1.16
     '
    -1.15
     mangled
    -1.14
     -----
    -1.13
    POSITIVE LOGITS
     invitaciones
    1.29
     examinar
    1.22
    いかがでしたか
    1.21
     robić
    1.21
     biela
    1.21
     któ
    1.21
     czemu
    1.19
    Ԫ
    1.19
    ｵ
    1.18
     bicicletas
    1.17
    Activations Density 0.159%

    No Known Activations