© 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. 6975
    Prev
    Next
    INDEX
    Explanations

    words ending in -al, -ary, -ant, or -ation

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

    The neuron consistently fires on five‐letter word‐prefix subtokens (e.g. “sagit,” “milit,” “sanit,” “digit,” etc.) that mark the beginnings of longer words.

    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
    JScrollPane
    -0.75
     Perio
    -0.75
    Perio
    -0.73
     Senna
    -0.73
    artner
    -0.72
    unpack
    -0.71
    Materia
    -0.70
     Azul
    -0.70
     jezik
    -0.70
    kmale
    -0.69
    POSITIVE LOGITS
    ary
    0.96
    ARY
    0.89
     tablo
    0.77
    哀
    0.77
     became
    0.74
     exchanged
    0.73
     Kate
    0.72
     веду
    0.70
     rígido
    0.70
     soda
    0.69
    Activations Density 0.036%

    No Known Activations