© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW Community BlogPrivacy & TermsContact
    1. Home
    3. Qwen3-32B
    5. MLP Neurons - Layer 58
    7. 2142
    Prev
    Next
    Layer 57Layer 59
    INDEX
    Explanations

    Christmas context

    np_max-act · claude-sonnet-5-5
    Triggered by @paultiplady

    decorating

    np_max-act-logits · claude-sonnet-5-5
    Triggered by @paultiplady

    Christmas greetings Method used: 3. The top logits are happy/merry words in several languages, and the activating texts are about Christmas and December 25

    np_min-act-logits · claude-sonnet-5-5Neg Acts
    Triggered by @paultiplady

    Words in Christmas or holiday-tradition text that come right after a point where a list of activities or customs is being set up or continued, such as verbs and commas or "and" introducing the next practice (e.g., "brought," "includes," "enjoyed," "put away," "take down"), so the neuron seems to predict that a holiday-related item or action comes next.

    oai_token-act-pair · claude-sonnet-5-5
    Triggered by @paultiplady

    Function words and punctuation (commas, "and", "of", "than", "with", periods) that link items in lists or descriptions of Christmas and holiday traditions, decorations, and activities, including connecting verbs within such enumerations.

    eleuther_acts_top20 · claude-sonnet-5-5
    Triggered by @paultiplady
    New Auto-Interp
    Top Features by Cosine Similarity
    Embeds
    IFrame
    Link
    Loading dashboard...