© 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. 10-GEMMASCOPE-RES-131K
    7. 12135
    Prev
    Next
    INDEX
    Explanations

    verbs and actions

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

    The neuron fires on verbs in non‐finite (infinitival or passive) constructions—i.e. it picks out instances of “to + verb” or “may/can be + past‐participle.”

    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
     deler
    -1.14
     completamente
    -1.09
     atmosfera
    -1.07
    を
    -1.06
     herrera
    -1.05
     totalmente
    -1.02
     lavabo
    -1.00
     kien
    -1.00
    这是一个
    -1.00
     truc
    -1.00
    POSITIVE LOGITS
     because
    1.56
     through
    1.54
     from
    1.52
     various
    1.31
    各式
    1.27
     two
    1.25
     six
    1.24
    مثلة
    1.23
     five
    1.19
     components
    1.19
    Activations Density 0.065%

    No Known Activations