© 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
    2. Gemma-2-27B
    3. 22-GEMMASCOPE-RES-131K
    4. 130633
    Prev
    Next
    INDEX
    Explanations

    The neuron flags very common function words—especially high‐frequency articles (e.g. “the,” “a”) and prepositions or short connector words.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408

    specific concepts across languages

    np_acts-logits-general · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_22/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
    
    -1.38
    が良い
    -1.34
     bienfaits
    -1.29
     brauch
    -1.28
    ebly
    -1.27
     médec
    -1.27
     sélectionner
    -1.26
    ഢ
    -1.24
     Suomessa
    -1.24
    atized
    -1.23
    POSITIVE LOGITS
     both
    1.35
    各種
    1.28
    Español
    1.15
     berbagai
    1.07
    Both
    1.07
     various
    1.07
    ทั้ง
    1.02
     چیه
    1.01
    各种
    1.01
     检查
    1.00
    Activations Density 0.037%

    No Known Activations