© 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. 22-GEMMASCOPE-RES-131K
    7. 129490
    Prev
    Next
    INDEX
    Explanations

    The neuron strongly activates on long, domain-specific section or link titles (e.g. “representation,” “protocols,” “repository,” “structure”) typically found in API or technical documentation.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408

    introduction and summary sections

    np_acts-logits-general · gemini-2.5-flash-lite
    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
     not
    -1.12
     in
    -1.00
     frumo
    -0.97
     From
    -0.93
     However
    -0.86
     The
    -0.85
     There
    -0.84
     Therefore
    -0.84
     needs
    -0.81
     Only
    -0.81
    POSITIVE LOGITS
    craina
    1.13
     œufs
    1.09
     arrière
    1.06
    astră
    1.04
     poêle
    1.02
    atiques
    1.01
    Mvh
    1.00
     asiatique
    0.99
     végét
    0.97
     rése
    0.94
    Activations Density 0.022%

    No Known Activations