© 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. 10-GEMMASCOPE-RES-131K
    7. 100322
    Prev
    Next
    INDEX
    Explanations

    terms followed by descriptor

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

    This neuron selectively fires on capitalized topic headings or single-word section titles (e.g. “Bee,” “Thermography,” “SIM,” “Olympic”), effectively spotting the main heading or title of a section.

    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
     provavelmente
    -1.55
     cómodo
    -1.53
     its
    -1.47
    ↵
    -1.46
     cheio
    -1.45
    -1.45
     trám
    -1.44
    		
    -1.44
                          
    -1.40
          
    -1.38
    POSITIVE LOGITS
     in
    1.95
     for
    1.83
     so
    1.77
     during
    1.69
     at
    1.57
     over
    1.54
     from
    1.46
     on
    1.38
     after
    1.37
     by
    1.32
    Activations Density 0.160%
    No Known Activations