© 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. 34-GEMMASCOPE-RES-131K
    7. 64253
    Prev
    Next
    INDEX
    Explanations

    code file suffixes and headings

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

    The neuron detects metadata tokens marking file names and section headings—e.g. “.py” extensions and “sec” labels.

    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
    ウンド
    -0.76
    only
    -0.74
     Cro
    -0.73
     "..\..\..\
    -0.72
    ynki
    -0.71
    ทั่ว
    -0.71
     Newtonsoft
    -0.71
    cro
    -0.71
     प्रदान
    -0.70
    しな
    -0.69
    POSITIVE LOGITS
    #!/
    1.00
    SetTitle
    0.99
    fini
    0.86
     tưởng
    0.83
     conseillé
    0.79
    dampak
    0.79
     heureux
    0.79
     script
    0.78
     unison
    0.78
     resell
    0.77
    Activations Density 0.001%

    No Known Activations