© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-3-12B
    5. 31-GEMMASCOPE-2-RES-16K
    7. 3350
    Prev
    Next
    INDEX
    Explanations

    The neuron activates on words from non-English text segments (i.e. foreign-language tokens).

    oai_token-act-pair · o4-mini
    Triggered 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
    Negative Logits
     also
    1.30
    also
    1.26
     actually
    1.05
    也在
    1.00
     always
    0.99
    actually
    0.98
     aslında
    0.98
     također
    0.98
     också
    0.95
    也
    0.93
    POSITIVE LOGITS
     Aplic
    0.68
     Aprove
    0.63
     Chứng
    0.61
     fizemos
    0.61
     ป
    0.60
     aprovech
    0.60
     Tiene
    0.58
     보
    0.57
     Anticip
    0.57
     acompan
    0.56
    Activations Density 0.135%

    No Known Activations