© 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. 35517
    Prev
    Next
    INDEX
    Explanations

    discussion quotes

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

    The neuron responds to tokens used in chat or forum transcript formatting—things like angle-bracketed user/timestamp tags and dialogue punctuation (e.g. “<live>,” numeric timestamps, question marks).

    oai_token-act-pair · o4-miniTriggered by @jyhe0408

    quotation marks at the beginning of quoted dialogue or speech.

    oai_token-act-pair · claude-4-5-sonnetTriggered by @jyhe0408

    direct, conversational dialogue—especially quoted speech with speaker turns, questions, and interjections.

    oai_token-act-pair · gpt-5Triggered 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
     and
    -1.10
     telefone
    -0.94
    bores
    -0.89
     cesse
    -0.87
     but
    -0.85
     Bí
    -0.84
     Existe
    -0.83
     Innenstadt
    -0.83
    すべき
    -0.82
     effectivement
    -0.81
    POSITIVE LOGITS
     смысле
    1.14
    例えば
    0.94
    цах
    0.89
     vielfach
    0.89
     словно
    0.89
     allgemeinen
    0.86
    Specifically
    0.85
     früheren
    0.85
     sogenannte
    0.85
     Vielzahl
    0.84
    Activations Density 0.014%

    No Known Activations