© 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. 22-GEMMASCOPE-RES-131K
    7. 12648
    Prev
    Next
    INDEX
    Explanations

    readread = 'true'

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

    The neuron fires on domain-specific jargon and acronyms—e.g. technical or medical terms, drug names, block names, programming keywords—wherever specialized vocabulary appears.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_22/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
     coute
    -1.06
     autonomie
    -1.05
     applau
    -0.99
     menac
    -0.99
     monstre
    -0.97
     épouse
    -0.96
    LEIA
    -0.94
    sapat
    -0.93
     vider
    -0.93
     capot
    -0.93
    POSITIVE LOGITS
     until
    0.97
     [...]
    0.95
     being
    0.90
     sammeln
    0.89
    Q
    0.88
    v
    0.85
     aiming
    0.84
    n
    0.83
    j
    0.82
     ....
    0.82
    Activations Density 0.192%

    No Known Activations