© 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. 12715
    Prev
    Next
    INDEX
    Explanations

    mathematical variables and country codes

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

    The neuron strongly activates on uppercase identifiers and acronyms (multi‐letter capitalized tokens) rather than ordinary lowercase 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_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
     comercio
    -0.91
     师
    -0.91
     probably
    -0.91
     resaltar
    -0.90
     comes
    -0.90
    ALLENGES
    -0.90
    mitsubishi
    -0.89
    renault
    -0.89
     negó
    -0.88
    مرة
    -0.88
    POSITIVE LOGITS
    some
    0.91
     “.
    0.90
     ')
    0.89
     parfums
    0.88
     potes
    0.86
     provar
    0.85
    [])
    0.85
     znajd
    0.85
     polie
    0.85
     こんな
    0.84
    Activations Density 0.170%

    No Known Activations