© 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. 100482
    Prev
    Next
    INDEX
    Explanations

    past tense verbs

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

    The neuron is detecting past-tense verb forms (especially auxiliary or helping verbs like was, did, have) that signal the past tense or perfect aspect.

    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
     then
    -1.36
     moderni
    -1.27
    他就
    -1.21
    ble
    -1.16
    顔を
    -1.14
    veyard
    -1.13
     ultimi
    -1.12
    了一口
    -1.10
     ...........
    -1.10
     .........
    -1.10
    POSITIVE LOGITS
     on
    1.45
     as
    1.35
     from
    1.27
     after
    1.26
     new
    1.25
     intensely
    1.16
     evidently
    1.15
     by
    1.11
     UIFont
    1.09
    湃
    1.04
    Activations Density 0.178%

    No Known Activations