© 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. 34-GEMMASCOPE-RES-131K
    7. 75770
    Prev
    Next
    INDEX
    Explanations

    Holden, Holder, held

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

    The neuron fires on verbs that express retaining or grasping (especially various forms of “hold” such as hold, held, holding, and related synonyms like keep).

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_34/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
    penes
    -0.80
    ürtel
    -0.78
    gående
    -0.75
     Foley
    -0.74
    ത്ര
    -0.73
    Suomen
    -0.73
     Figures
    -0.72
     Rot
    -0.71
    Addressing
    -0.71
     Blak
    -0.70
    POSITIVE LOGITS
     aloof
    1.22
     fast
    1.07
     firmly
    1.00
     firm
    0.90
     aloft
    0.87
     alive
    0.84
    fast
    0.82
     erect
    0.82
     awake
    0.78
    住
    0.78
    Activations Density 0.029%

    No Known Activations