© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 10-GEMMASCOPE-RES-131K
    7. 99884
    Prev
    Next
    INDEX
    Explanations

    modal verbs of possibility and condition

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

    The neuron fires on conditional/hypothetical modal constructions (e.g. “would,” “could,” “desirable,” “easily,” “follow”) indicating possibility or expectation.

    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
     två
    -1.68
     valiente
    -1.40
    -\\
    -1.39
     colère
    -1.38
     delincuentes
    -1.35
     inocente
    -1.35
    zó
    -1.35
    тно
    -1.30
     mycket
    -1.29
    籥
    -1.29
    POSITIVE LOGITS
    7
    1.61
    4
    1.41
    3
    1.41
     Hmm
    1.41
     Especially
    1.38
     but
    1.38
    6
    1.34
     Should
    1.34
    as
    1.31
    Something
    1.30
    Activations Density 0.047%

    No Known Activations