© 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. 84329
    Prev
    Next
    INDEX
    Explanations

    would + action/recommendation

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

    The neuron fires strongly on modal‐auxiliary constructions (especially “would,” “could,” “should,” etc.) and associated verbs/adverbs expressing hypothetical, conditional, or volitional 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
    aternion
    -1.01
     postales
    -1.00
     beslu
    -0.97
     uska
    -0.96
     Inorg
    -0.95
     déterminer
    -0.95
     partij
    -0.95
     Limoges
    -0.95
    ことで
    -0.93
     oublié
    -0.93
    POSITIVE LOGITS
     if
    1.96
     would
    1.75
    would
    1.43
     any
    1.42
    Would
    1.38
     gdyby
    1.36
     myself
    1.30
     a
    1.30
    があれば
    1.24
     если
    1.20
    Activations Density 0.028%

    No Known Activations