© 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. 130548
    Prev
    Next
    INDEX
    Explanations

    The neuron reliably lights up on subword fragments that appear in named entities and technical terms (e.g. horse-race names, drug names like “mesylate,” F-1, plant species, etc.).

    oai_token-act-pair · o4-miniTriggered by @jyhe0408

    racetrack, spectrum disorders, mesylate

    np_acts-logits-general · gemini-2.5-flash-lite
    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
     from
    -1.24
     are
    -1.17
     to
    -1.17
     all
    -1.08
     any
    -1.06
    asta
    -1.03
     here
    -1.02
     within
    -1.02
     even
    -1.02
     at
    -1.01
    POSITIVE LOGITS
    parken
    1.09
     comprenant
    1.09
     لبنان
    1.06
    こともあります
    1.05
    familien
    1.05
    Dette
    1.03
     captiv
    1.03
    ίας
    1.02
    zahn
    1.02
    どく
    1.02
    Activations Density 0.100%

    No Known Activations