© 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
    3. Gemma-2-27B
    5. 34-GEMMASCOPE-RES-131K
    7. 84712
    Prev
    Next
    INDEX
    Explanations

    code structure and definition

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

    The neuron fires strongly on document‐structure metadata tokens (e.g. “begin document,” section markers, page-title/description fields, numeric IDs), i.e. the boilerplate header/footer markup rather than natural language.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_34/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
    借助
    -0.98
     sentimenti
    -0.89
    Պ
    -0.88
    lanatory
    -0.82
    ｺ
    -0.79
    带宽
    -0.78
     الدرا
    -0.76
    queles
    -0.75
     recogni
    -0.75
    一段时间
    -0.74
    POSITIVE LOGITS
     sentences
    0.97
    }}"
    0.89
    rceil
    0.88
    री
    0.88
    LSTM
    0.84
     fakty
    0.84
     añade
    0.83
     Glaube
    0.82
     token
    0.81
    número
    0.81
    Activations Density 0.005%

    No Known Activations