© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 10-GEMMASCOPE-RES-131K
    7. 13211
    Prev
    Next
    INDEX
    Explanations

    interleaved / deinterleaved

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

    The neuron detects specialized technical jargon—particularly domain‐specific terms (often hyphenated or compound words) from fields like graphics, imaging, printing, and signal processing.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-27b-pt-res/layer_10/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
    ↵↵
    -2.28
     However
    -2.03
     Such
    -1.89
     Undoubtedly
    -1.87
     Fortunately
    -1.87
     Unfortunately
    -1.86
     Ideally
    -1.86
     Therefore
    -1.80
     Then
    -1.73
     Certainly
    -1.73
    POSITIVE LOGITS
    
    2.00
    '
    1.92
    鉝
    1.77
    鱨
    1.75
    ه‌ی
    1.71
    潆
    1.71
    咵
    1.70
    神
    1.67
     hè
    1.66
     éclat
    1.63
    Activations Density 0.010%

    No Known Activations