© 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-3-12B
    5. 31-GEMMASCOPE-2-RES-16K
    7. 8003
    Prev
    Next
    INDEX
    Explanations

    ultra and ul prefixes

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

    This neuron strongly activates on tokens containing the substring “Ul” (as in “Ultra,” “Ultima,” “Ugland,” “UEFA,” etc.), effectively flagging proper names or brands that begin with “Ul.”

    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
    帮
    0.63
    apt
    0.55
    ިން
    0.55
    ⿹
    0.55
    ደበ
    0.53
    ᱐
    0.52
    isham
    0.51
    蔦
    0.51
     próp
    0.51
    owers
    0.51
    POSITIVE LOGITS
     Ult
    0.70
     Ul
    0.64
     лучше
    0.63
     Better
    0.63
     Unwrap
    0.60
    Ult
    0.59
     Ulm
    0.59
    raviolet
    0.58
     ultim
    0.58
    Better
    0.57
    Activations Density 0.183%

    No Known Activations