© 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. 128719
    Prev
    Next
    INDEX
    Explanations

    This neuron detects opening curly braces (“{”), i.e. the start of a code block.

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

    starting tokens for structured text

    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
    ṝ
    -1.45
    könig
    -1.20
     konflikt
    -1.20
     Tabelle
    -1.18
     Algunas
    -1.18
     瘦
    -1.15
     TextAppearance
    -1.15
    cestors
    -1.14
     château
    -1.13
     continuación
    -1.13
    POSITIVE LOGITS
     and
    1.42
    <eos>
    1.34
     same
    1.23
     stesse
    1.18
     children
    1.13
     lived
    1.07
     out
    1.05
    兩
    1.05
     although
    1.04
     is
    1.02
    Activations Density 0.000%

    No Known Activations