© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 34-GEMMASCOPE-RES-131K
    7. 59719
    Prev
    Next
    INDEX
    Explanations

    code generation or mocking

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

    The neuron fires on code‐like tokens—particularly identifiers, import‐path segments, XML tags, and all‐caps or CamelCase words—i.e. tokens characteristic of source‐code syntax rather than ordinary text.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_34/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
     Init
    -0.94
    systems
    -0.83
     GUST
    -0.81
     systems
    -0.80
     erreur
    -0.73
     JOSÉ
    -0.71
     KENN
    -0.71
     diterima
    -0.70
     aerobic
    -0.70
     GEORG
    -0.70
    POSITIVE LOGITS
     mock
    1.09
     calls
    0.93
     mocked
    0.93
    recorder
    0.90
     CTRL
    0.88
    omock
    0.88
     expectation
    0.88
     mocking
    0.86
     fake
    0.85
     expectations
    0.85
    Activations Density 0.001%

    No Known Activations