© 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. 25510
    Prev
    Next
    INDEX
    Explanations

    code structure and data configuration

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

    The neuron fires strongly on identifiers used as object‐property or JSON keys (whether camelCase or snake_case) – i.e. it’s picking out key names in key–value definitions.

    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
    -1.11
     will
    -1.09
     Part
    -1.08
     each
    -1.08
    Part
    -1.00
     Each
    -0.98
     part
    -0.97
     still
    -0.96
     includes
    -0.96
     other
    -0.94
    POSITIVE LOGITS
     mediu
    1.18
     nerfs
    1.13
    grilled
    1.12
     maintenant
    1.12
    became
    1.11
     désormais
    1.10
    warz
    1.09
    ŕ
    1.09
    vertis
    1.07
     bril
    1.04
    Activations Density 0.024%

    No Known Activations