© 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. 104157
    Prev
    Next
    INDEX
    Explanations

    start of abstracts

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

    The neuron fires on structured, metadata‐style tokens—especially section headings like “Abstract” or “Summary,” punctuation used as separators (e.g. “—” or “--”), and embedded numerical or statistical values—i.e. the elements that mark formal document structure or data presentation.

    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
     then
    -2.13
     of
    -1.91
     now
    -1.80
     being
    -1.77
     You
    -1.71
     your
    -1.63
    _
    -1.59
    "
    -1.58
     \
    -1.55
     To
    -1.55
    POSITIVE LOGITS
    瘆
    2.03
    鼴
    1.84
     plafon
    1.77
     barriga
    1.77
     craindre
    1.76
     skak
    1.75
     egens
    1.71
     vatten
    1.69
     egent
    1.67
    鳜
    1.67
    Activations Density 0.089%

    No Known Activations