© 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 EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Gemma-2-27B
    5. 22-GEMMASCOPE-RES-131K
    7. 14581
    Prev
    Next
    INDEX
    Explanations

    is not a conjunction

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

    The neuron detects hedging or disclaimer phrases that use negation to soften a statement (e.g. “not saying it’s a bad…,” “I’m not saying…”).

    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
    both
    -1.26
     both
    -1.23
    BOTH
    -1.20
    Both
    -1.16
     både
    -1.14
     BOTH
    -1.05
     zarówno
    -1.02
    thodox
    -1.00
    ageable
    -0.94
    ทั้ง
    -0.94
    POSITIVE LOGITS
     or
    1.62
     necessarily
    1.57
    necessarily
    1.23
     únicamente
    1.16
     deswegen
    0.96
     necesariamente
    0.95
    ;
    0.94
    或
    0.91
     unbedingt
    0.90
    atakan
    0.89
    Activations Density 0.024%

    No Known Activations