© 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. 19419
    Prev
    Next
    INDEX
    Explanations

    Street Journal, Colombo, brown, wagon

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

    The neuron is detecting document‐level metadata tokens—things like dateline markers (dates, locations, sources in parentheses), section or headline names, footnote/citation symbols, event or media outlet titles—rather than ordinary running text.

    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
    sukienka
    -2.33
    台中
    -2.23
    เรา
    -2.19
    0
    -2.17
    𝕻
    -2.06
    leopard
    -2.05
     間取り
    -2.05
    kurtka
    -2.03
     provides
    -2.02
    Guinea
    -2.02
    POSITIVE LOGITS
    …….
    2.81
    他能
    2.70
    "
    2.66
    就得
    2.53
    ….
    2.53
    The
    2.50
    ……..
    2.47
     Similarly
    2.42
    ،
    2.41
    ным
    2.38
    Activations Density 0.001%

    No Known Activations