© 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. 22-GEMMASCOPE-RES-131K
    4. 70617
    Prev
    Next
    INDEX
    Explanations

    would be [adjective] to

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

    The neuron spotlights conditional modal constructions—instances of “would” (or its ’d contraction) plus a verb (be, have been, miss, etc.) indicating hypothetical or counterfactual statements.

    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
    ļas
    -1.16
    ユニクロ
    -1.15
     verkar
    -1.11
    AIzaSy
    -1.09
     Hae
    -1.09
    ينو
    -1.04
     Faça
    -1.02
    怎么回事
    -1.02
    Acerca
    -1.00
    Zw
    -0.99
    POSITIVE LOGITS
     would
    1.30
     anyone
    1.23
     gdyby
    1.16
     if
    1.11
     cualquiera
    1.10
     any
    1.09
     to
    1.05
     cualquier
    1.04
    ieť
    1.04
     qualsiasi
    1.02
    Activations Density 0.010%

    No Known Activations