© 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
    2. Gemma-2-27B
    3. 22-GEMMASCOPE-RES-131K
    4. 72325
    Prev
    Next
    INDEX
    Explanations

    frequently occurs; often stated

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

    The neuron strongly fires on out-of-ordinary, technical “keyword” tokens—things like acronyms, code identifiers, and other domain-specific jargon—rather than common words.

    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
     â
    -1.16
     can
    -1.11
     as
    -1.11
     on
    -1.09
     in
    -1.05
     so
    -1.04
     will
    -1.04
     AND
    -0.98
     and
    -0.97
    connectionString
    -0.97
    POSITIVE LOGITS
    significantly
    1.23
     souvent
    1.15
    larg
    1.13
    dras
    1.09
    commonly
    1.08
     normalmente
    1.08
     vários
    1.08
     rembourse
    1.05
    frequently
    1.05
     nombreux
    1.05
    Activations Density 0.089%

    No Known Activations