© 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 EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    2. Gemma-3-12B
    3. 24-GEMMASCOPE-2-RES-16K
    4. 2773
    Prev
    Next
    INDEX
    Explanations

    These lists contain elements from Korean, Chinese, and Japanese languages, often functioning as grammatical particles or descriptors. The `TOKENS_AFTER_MAX_ACTIVATING_TOKEN` list includes words like "factors," "content," "view," "aspect," "model," "main," and "method."The pattern suggests the neuron is recognizing descriptors or particles commonly found in lists or enumerations across East Asian languages, often preceding explanatory terms like "factors," "content," or "aspects."Therefore, a good short explanation would be:East Asian descriptive particles before list items

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

    This neuron detects non-Latin (East Asian) script tokens, e.g. Korean or Japanese 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
    由于
    0.51
    哪些
    0.50
    的需求
    0.48
    athar
    0.48
    的看着
    0.47
    看着
    0.46
    യില
    0.46
    的原因
    0.45
    在哪
    0.45
     населения
    0.45
    POSITIVE LOGITS
     것도
    0.75
     것이
    0.75
    ことは
    0.66
     şeyi
    0.64
     것은
    0.63
     것으로
    0.63
    ことを
    0.62
    ような
    0.61
    ことが
    0.61
     şey
    0.60
    Activations Density 0.003%

    No Known Activations