© 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 EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 128693
    Prev
    Next
    INDEX
    Explanations

    The neuron lights up on longer, multi-syllable content words—especially technical or abstract terms (e.g. “forming,” “combination,” “industrial,” “refueling,” “inequalities,” “discrimination,” “communication”).

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

    different nouns and properties

    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
     simply
    -0.97
     inmediato
    -0.85
    ことが多い
    -0.84
    ardı
    -0.84
     never
    -0.83
    が多い
    -0.81
     barely
    -0.81
     only
    -0.80
    ὲν
    -0.79
     simplesmente
    -0.79
    POSITIVE LOGITS
     разные
    1.15
    ちゃんと
    1.10
     разных
    1.09
     distintas
    1.06
     diferite
    1.04
     dokonce
    1.04
     diferentes
    1.02
    Different
    1.02
     different
    1.02
    different
    0.99
    Activations Density 0.098%

    No Known Activations