© 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. 128660
    Prev
    Next
    INDEX
    Explanations

    This neuron fires on content‐bearing, domain-specific tokens (technical terms, specialized nouns/verbs) rather than common function words.

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

    fired once

    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
     occurs
    -0.99
    を作成
    -0.88
     coû
    -0.88
    枚目
    -0.88
    悲伤
    -0.88
     anciennes
    -0.87
     juſ
    -0.85
     effectivement
    -0.85
     Occur
    -0.85
     exercer
    -0.85
    POSITIVE LOGITS
     will
    1.40
     now
    1.34
     subsequently
    1.23
     remain
    1.21
     still
    1.15
     keep
    1.14
     kept
    1.13
     remained
    1.11
     later
    1.10
     continue
    1.10
    Activations Density 0.097%

    No Known Activations