© 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. 34-GEMMASCOPE-RES-131K
    4. 372
    Prev
    Next
    INDEX
    Explanations

    witch-hunt, liquors, required self-incrimination

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

    The neuron is triggered by tokens that represent document‐level numeric markers or metadata—e.g. line‐number or docketrdocket signs (“#”), case numbers, section numbers, or other standalone numeric identifiers.

    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
    ESM
    -0.71
     monastic
    -0.70
    jut
    -0.68
    朐
    -0.68
    ázs
    -0.68
     Usher
    -0.67
    lerc
    -0.66
    шайте
    -0.66
    tase
    -0.65
     PRT
    -0.65
    POSITIVE LOGITS
    csr
    0.75
    Typed
    0.67
    成本
    0.64
     GOOD
    0.61
    kasse
    0.60
    ífica
    0.60
    いない
    0.59
    平凡
    0.59
    链接
    0.58
     importa
    0.58
    Activations Density 0.095%

    No Known Activations