© 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
    2. Gemma-3-12B
    3. 24-GEMMASCOPE-2-RES-16K
    4. 8573
    Prev
    Next
    INDEX
    Explanations

    Okay, let's break downThe neuron seems to be activated by phrases that introduce an explanation or a breakdown of information, often following an initial statement or question. The `let's break down` and `Okay, let's break down` examples are strong indicators. The token `let` also appears frequently in `MAX_ACTIVATING_TOKENS`, potentially hinting at the start of such explanatory sequences, which are then often followed by specific types of information or follow-up questions.Based on this, the neuron appears to be associated with the act of providing explanations or dissecting information.**Concise Explanation:**`Okay, let's break down`

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

    The neuron strongly responds to semantically rich, topic-bearing content words (especially nouns) rather than to function words.

    oai_token-act-pair · o4-miniTriggered by @jyhe0408
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    google/gemma-scope-2-12b-pt/resid_post/layer_24_width_16k_l0_medium
    Prompts (Dashboard)
    392,802 prompts, 256 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     Олександр
    0.83
    ফ
    0.76
    AKER
    0.72
     চাহিয়া
    0.71
     மீட்ப
    0.71
     सलूशन
    0.71
    રિક
    0.70
    detection
    0.70
     Ростов
    0.70
    ombies
    0.70
    POSITIVE LOGITS
    芫
    0.88
    就开始
    0.73
    ┐
    0.70
    ñana
    0.70
    ১৪
    0.69
    すすめ
    0.69
     hatchery
    0.68
     spad
    0.66
    apun
    0.66
    挑
    0.65
    Activations Density 0.000%

    No Known Activations