© 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. 33462
    Prev
    Next
    INDEX
    Explanations

    quinine, medical measurements

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

    This neuron activates for years in scientific or academic publications, particularly when they appear as dates in the format "2005", "1994-2006", etc.

    oai_token-act-pair · claude-3-7-sonnet-20250219Triggered by @neilrathi

    This neuron fires on technical “Methods/Results” language in scientific writing—words signaling detailed experimental or analytical reporting (e.g. “According,” “parasitemia,” “replacements,” “Analysis,” etc.).

    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
     it
    -0.96
    ittar
    -0.94
    plona
    -0.91
    時間が
    -0.90
    vez
    -0.88
     Assessing
    -0.88
     will
    -0.85
     decorazione
    -0.85
    私には
    -0.85
     Wouldn
    -0.84
    POSITIVE LOGITS
     titul
    1.03
    police
    1.02
     postale
    1.00
    deutschen
    0.97
     territoriale
    0.96
    ymbol
    0.93
     plafon
    0.93
    כש
    0.92
     formules
    0.92
    (…)
    0.91
    Activations Density 0.054%

    No Known Activations