© 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-2-27B
    5. 10-GEMMASCOPE-RES-131K
    7. 100690
    Prev
    Next
    INDEX
    Explanations

    action completion or state

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

    The neuron activates on tokens that are clearly non-English—e.g. foreign words or snippets in Thai/Vietnamese scripts or with diacritics—i.e. it flags non-Latin or non-English tokens.

    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
    8
    -2.31
    6
    -2.31
     expresiones
    -2.30
    鸲
    -2.14
    拉克
    -2.03
     после
    -2.03
    媄
    -2.02
    3
    -1.98
    5
    -1.95
    𖥧
    -1.92
    POSITIVE LOGITS
     the
    2.08
     along
    2.05
     into
    2.03
     among
    1.91
     together
    1.80
     and
    1.80
     against
    1.73
    蹀
    1.64
     generalmente
    1.63
     ahead
    1.63
    Activations Density 0.001%

    No Known Activations