© 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. 83597
    Prev
    Next
    INDEX
    Explanations

    whichever or whoever

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

    This neuron fires on tokens that occur in software‐development metadata and legal boilerplate—e.g. package or language names, dependency/version keywords, license names, URLs, and related numbers.

    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
     ranging
    -1.05
    逆に
    -0.96
     ranged
    -0.91
     erity
    -0.87
     Какие
    -0.83
     OTO
    -0.83
    क्या
    -0.82
    especially
    -0.80
    美味しそう
    -0.79
    吃的
    -0.79
    POSITIVE LOGITS
     whichever
    5.03
    hichever
    3.44
     whoever
    1.90
    どちら
    1.66
    哪个
    1.63
     Whoever
    1.62
    Whoever
    1.57
     whome
    1.55
     whiche
    1.48
     whatever
    1.46
    Activations Density 0.045%

    No Known Activations