© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-3-12B
    5. 31-GEMMASCOPE-2-RES-16K
    7. 9729
    Prev
    Next
    INDEX
    Explanations

    imperative verbs followed by specific endings

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

    The neuron is keyed to imperative or transactional base‐form verbs—especially calls-to-action like “take,” “buy,” “sell,” “spin,” “shoot,” “help,” or “follow.”

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2-12b-pt/resid_post/layer_31_width_16k_l0_medium
    Prompts (Dashboard)
    392,802 prompts, 256 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    种
    0.64
    可以看出
    0.61
    প্রাচ
    0.61
    ized
    0.59
     fontweight
    0.59
    ize
    0.59
     trapez
    0.58
    焕
    0.58
     int
    0.57
    ھ
    0.57
    POSITIVE LOGITS
    ings
    1.48
    able
    1.27
    INGS
    1.11
    backs
    1.03
    overs
    1.02
    ables
    1.00
    outs
    0.99
    away
    0.99
    arounds
    0.99
    ABLE
    0.97
    Activations Density 1.211%

    No Known Activations