© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 28008
    Prev
    Next
    INDEX
    Explanations

    introduces contrast or exception

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

    The neuron is acting as a discourse‐marker detector – it lights up on transition words and sentence‐starters like “Now,” “However,” “if,” “as,” “you,” “can,” 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
     suitable
    -1.08
     anything
    -1.02
     even
    -1.02
    buttonStyle
    -0.99
     appropriate
    -0.99
     definitely
    -0.98
     easily
    -0.97
     always
    -0.96
    しましょう
    -0.96
     any
    -0.96
    POSITIVE LOGITS
     totiž
    1.26
    なんと
    1.19
     nämlich
    1.14
    Instead
    1.14
    !!!!!!!!!
    1.13
    Ĭ
    1.09
     vaksin
    1.06
     Février
    1.05
     -¿
    1.04
    ‼️
    1.02
    Activations Density 0.215%

    No Known Activations