© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 10-GEMMASCOPE-RES-131K
    7. 101396
    Prev
    Next
    INDEX
    Explanations

    external links and categories

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

    Looking at the activations, this neuron strongly activates for: - `:` (colons) with very high activation (58.5) - `AddField` (37.5) - `alive` (34) - `Category` (multiple instances around

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

    The neuron strongly responds to structural or meta‐text markers—tokens that start off headings or labels (e.g. section titles like “External,” “Category,” version numbers like “2.”, migration commands like “AddField,” etc.).

    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
    𝐍
    -1.80
     habile
    -1.80
     februari
    -1.72
     étan
    -1.68
    titud
    -1.65
    </h3>
    -1.63
     '&
    -1.63
     élo
    -1.62
    ם
    -1.62
    czę
    -1.61
    POSITIVE LOGITS
    については
    1.73
     is
    1.73
     новых
    1.64
     are
    1.63
    議論
    1.62
     discussed
    1.61
    on
    1.58
     too
    1.56
     also
    1.48
    则是
    1.46
    Activations Density 0.007%

    No Known Activations