© 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. 22-GEMMASCOPE-RES-131K
    7. 33942
    Prev
    Next
    INDEX
    Explanations

    intermediary entities and components

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

    The neuron fires most strongly on uncommon, content-heavy words—especially technical or domain-specific terms and nouns—rather than on common function words.

    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
     even
    -0.91
     particularly
    -0.89
     afec
    -0.87
     want
    -0.86
     started
    -0.85
     destroying
    -0.84
    ünstler
    -0.83
     by
    -0.82
     time
    -0.81
    老太太
    -0.81
    POSITIVE LOGITS
     intermediary
    1.20
    然后再
    0.93
    撵
    0.92
     pozo
    0.91
    然後
    0.91
     mauvaises
    0.91
     who
    0.89
     Quien
    0.87
     cascading
    0.86
     courants
    0.86
    Activations Density 0.097%

    No Known Activations