© 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. 84499
    Prev
    Next
    INDEX
    Explanations

    but still

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

    The neuron activates on adversative discourse markers—words that signal contrast or contradiction (e.g. “but,” “still”).

    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
    trui
    -1.04
    ᐱ
    -1.04
     glück
    -1.03
    тири
    -0.98
     konkuren
    -0.96
     acteur
    -0.95
    蘄
    -0.93
     书法
    -0.93
     alchemist
    -0.93
    ⢱
    -0.92
    POSITIVE LOGITS
     still
    1.48
     etc
    1.24
     general
    1.20
     idea
    1.20
     point
    1.20
     even
    1.14
    それでも
    1.10
     nonetheless
    1.03
     anyway
    1.02
    etc
    1.00
    Activations Density 0.043%

    No Known Activations