© 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. 112548
    Prev
    Next
    INDEX
    Explanations

    past tense verbs

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

    The neuron fires on passive‐voice constructions—especially the auxiliary “be” (and its past forms like “was,” “been”) together with past participles or similar markers of passive phrasing.

    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
     besos
    -1.75
    urement
    -1.59
    zied
    -1.58
     banyo
    -1.45
     foton
    -1.42
     magazin
    -1.38
     otel
    -1.36
     eko
    -1.35
     mavi
    -1.35
    alogical
    -1.33
    POSITIVE LOGITS
     quedará
    1.32
     deliver
    1.30
     want
    1.27
    čnega
    1.27
    都被
    1.21
    ↵↵↵
    1.19
     each
    1.18
     habrá
    1.16
     have
    1.15
     necesitar
    1.15
    Activations Density 0.158%

    No Known Activations