© 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. 104412
    Prev
    Next
    INDEX
    Explanations

    apologize, redemption, remorse

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

    The neuron detects words and word‐parts related to apologies (e.g. “apologize,” “apologies,” “unapologetic,” 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
     are
    -1.63
     =
    -1.15
     of
    -1.13
     /
    -1.12
    .
    -1.07
    vegli
    -1.04
    preprint
    -1.02
    *
    -1.01
    と思ったら
    -1.01
     -
    -1.00
    POSITIVE LOGITS
    paket
    1.40
     maravillas
    1.28
     remorse
    1.27
    offering
    1.23
    össä
    1.23
    Szia
    1.23
    kamera
    1.23
     acciones
    1.23
     criaturas
    1.22
    موقع
    1.22
    Activations Density 0.022%

    No Known Activations