© 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. Qwen3-32B
    5. 32-RESID-BATCHTOPK-65K
    7. 9084
    Prev
    Next
    INDEX
    Explanations

    found text indicating a start

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

    The neuron detects document metadata and boilerplate — headings, bylines/author lines, dates, copyright/contact info and other structural header elements.

    oai_token-act-pair · gpt-5-miniTriggered by @yooniel31
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    adamkarvonen/qwen3-32b-saes/saes_Qwen_Qwen3-32B_batch_top_k/resid_post_layer_32
    Prompts (Dashboard)
    16,384 prompts, 128 tokens each
    Dataset (Dashboard)
    monology/pile-uncopyrighted
    No Configuration Found
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     \`
    -0.09
    æĬĬèĩªå·±çļĦ
    -0.08
    ãĢı
    -0.08
    æĪijå¿ĥéĩĮ
    -0.07
    #{
    -0.07
    ãģĤãĤĭãģĦãģ¯
    -0.07
    ä¹ŁæŃ£æĺ¯
    -0.07
    æ¯Ķå¦Ĥ说
    -0.07
    åıªä¸įè¿ĩæĺ¯
    -0.07
    è¿Ļç§įäºĭæĥħ
    -0.07
    POSITIVE LOGITS
    ãĤ¢ãĤ¤ãĥĨ
    0.09
    ãĤ¿ãĤ¤ãĥŁ
    0.09
    jsc
    0.09
    ãĤ¤ãĥ³ãĤ¿ãĥ¼ãĥį
    0.08
    ç¶ļãģįãĤĴèªŃ
    0.08
    ãĪ¯
    0.08
    ÙĤÙĩ
    0.08
    YRO
    0.08
    esters
    0.08
    åįķä¸Ģ
    0.08
    Activations Density 16.056%

    No Known Activations