© 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. 22-GEMMASCOPE-RES-131K
    7. 28360
    Prev
    Next
    INDEX
    Explanations

    transgene integration sites

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

    closing parentheses, braces, or other code syntax elements in programming code or technical documentation.

    oai_token-act-pair · claude-3-7-sonnet-20250219Triggered by @neilrathi

    The neuron primarily activates on numeric literals (numbers and numerical values) in the 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_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
     fidé
    -0.88
     découver
    -0.87
    реги
    -0.84
     tenues
    -0.84
     répand
    -0.83
     pè
    -0.83
     conva
    -0.83
     chaussures
    -0.82
     finalised
    -0.81
    Lauren
    -0.81
    POSITIVE LOGITS
    これが
    0.94
     serve
    0.88
     became
    0.88
     dijadikan
    0.87
     becomes
    0.85
     it
    0.84
     become
    0.81
     later
    0.81
     velocità
    0.79
     replaces
    0.78
    Activations Density 0.025%

    No Known Activations