© 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
    2. Gemma-2-27B
    3. 10-GEMMASCOPE-RES-131K
    4. 66182
    Prev
    Next
    INDEX
    Explanations

    dateline punctuation and following location

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

    The neuron fires on the dateline-style location and date markers (e.g. “CITY, STATE –”, “CITY –(…DATE…)”), spotting the typical news‐wire header format.

    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
     all
    -1.55
     همه
    -1.55
    кос
    -1.49
    формы
    -1.45
     hierdie
    -1.45
    裳
    -1.39
    фом
    -1.37
     wildflowers
    -1.33
     iyong
    -1.32
    都有
    -1.31
    POSITIVE LOGITS
    </strong>
    2.36
    </b>
    2.17
    </em>
    1.71
    </sup>
    1.63
     about
    1.62
    But
    1.57
    法则
    1.57
     transferencia
    1.57
     especially
    1.55
     крайне
    1.48
    Activations Density 0.011%

    No Known Activations