© 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. 22-GEMMASCOPE-RES-131K
    7. 130530
    Prev
    Next
    INDEX
    Explanations

    The neuron consistently lights up on content‐bearing words (nouns, verbs, adjectives, etc.) and stays silent on common function words. In other words, it flags “informative” tokens and ignores high-frequency stopwords.

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

    Q: format questions

    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
    anthrene
    -0.82
    ("")){
    -0.80
     идея
    -0.80
    iels
    -0.80
    Bye
    -0.76
     искусство
    -0.75
    exactly
    -0.75
     ráp
    -0.75
     erd
    -0.75
    profen
    -0.75
    POSITIVE LOGITS
    什么样的
    1.00
    니다
    0.88
     tanam
    0.85
    úc
    0.82
    imenta
    0.81
     این
    0.81
    𝓬
    0.81
     Վ
    0.80
     памя
    0.79
    ぼ
    0.79
    Activations Density 0.014%

    No Known Activations