© 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. Qwen3-32B
    5. 32-RESID-BATCHTOPK-65K
    7. 21462
    Prev
    Next
    INDEX
    Explanations

    Roman history (Emperor, Roman, Caesar, Hadrian's Wall)

    np_acts-logits-general · gemini-2.5-flash-lite
    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
    åĩĢåĢ¼
    -0.10
     dare
    -0.08
     competitive
    -0.08
    æ¢Ń
    -0.08
    å¤ĸ表
    -0.07
    紧紧
    -0.07
    ichten
    -0.07
    让åѦçĶŁ
    -0.07
    éŃĦ
    -0.07
    æİ©
    -0.07
    POSITIVE LOGITS
    ç½Ĺ马
    0.12
    æµ´
    0.11
     Caesar
    0.10
    士åħµ
    0.10
    é©¿ç«Ļ
    0.10
    íıIJ
    0.10
    tü
    0.10
     Roman
    0.09
     Imperial
    0.09
    éĺ³åı°
    0.09
    Activations Density 0.057%

    No Known Activations