© 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. 102860
    Prev
    Next
    INDEX
    Explanations

    technical terms and keywords

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

    compound or multi-word identifiers in code and technical documentation, especially those with underscores or camel case.

    oai_token-act-pair · claude-4-5-sonnetTriggered by @bfgcn7frf5

    The neuron fires strongly on technical or code identifiers and product/class names (e.g. software titles, unit‐test classes, struct/type names, exception names, macros).

    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
     just
    -2.44
     using
    -2.39
     all
    -2.39
     their
    -2.30
     and
    -2.28
     to
    -2.28
     during
    -2.25
     not
    -2.17
     any
    -2.17
     from
    -2.16
    POSITIVE LOGITS
     ſol
    1.95
    龇
    1.95
    惴
    1.91
     bross
    1.88
     attirer
    1.87
    泚
    1.86
     ventes
    1.86
     ſou
    1.85
     plafon
    1.84
    耷
    1.80
    Activations Density 0.077%

    No Known Activations