© 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. 10-GEMMASCOPE-RES-131K
    4. 3687
    Prev
    Next
    INDEX
    Explanations

    domain names and codes

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

    The neuron strongly fires on isolated tokens made up primarily of uppercase letters, digits, punctuation, or other non-word characters (e.g. acronyms in parentheses, numeric strings, URLs, “.com,” etc.).

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_10/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
     there
    -1.95
     it
    -1.89
     but
    -1.70
     our
    -1.57
    </em>
    -1.56
     what
    -1.49
     both
    -1.46
     some
    -1.35
     do
    -1.29
     don
    -1.28
    POSITIVE LOGITS
     gangen
    1.41
    whelmed
    1.36
     arrêter
    1.34
    hauled
    1.30
     vertus
    1.30
     noix
    1.30
     kochen
    1.30
     ideen
    1.30
     deuil
    1.29
     eksemp
    1.28
    Activations Density 0.374%

    No Known Activations