© 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
    3. Qwen3-32B
    5. 32-RESID-BATCHTOPK-65K
    7. 18326
    Prev
    Next
    INDEX
    Explanations

    thinking 'not' or 'wrong'

    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
    rious
    -0.09
    æĮīè§Ħå®ļ
    -0.09
    .biz
    -0.09
    led
    -0.09
    çļĦæľīæķĪ
    -0.08
    çĸ±
    -0.08
    尺度
    -0.08
     cele
    -0.08
     adm
    -0.08
     organ
    -0.08
    POSITIVE LOGITS
    å®ŀéĻħæĥħåĨµ
    0.13
    éĶĻäºĨ
    0.12
    æ®Ĭ
    0.12
    äºĭå®ŀ
    0.12
     untrue
    0.11
     reality
    0.11
    ä¸įçĦ¶
    0.10
     Nope
    0.10
     WRONG
    0.10
    ãģĿãģĨ
    0.10
    Activations Density 0.086%

    No Known Activations