© 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-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 13765
    Prev
    Next
    INDEX
    Explanations

    non-english words and numbers

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

    The neuron fires on mid‐ to high‐weight content words that carry evaluative or judgmental meaning (e.g. “erasing,” “respect,” “argue,” “best,” “value,” “power,” “improved”).

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    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
     might
    -1.23
    astanza
    -1.22
     didn
    -1.19
     Mostly
    -1.16
     occasionally
    -1.16
     Sometimes
    -1.16
     wasn
    -1.13
    อยาก
    -1.13
     hadn
    -1.13
    หลัง
    -1.12
    POSITIVE LOGITS
     trente
    1.17
     günstiger
    1.16
    niger
    1.09
     россия
    1.07
    xvii
    1.07
     lahat
    1.07
     Projektu
    1.06
     zwi
    1.05
    だけではなく
    1.05
     FÜR
    1.05
    Activations Density 0.123%

    No Known Activations