© 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. 8161
    Prev
    Next
    INDEX
    Explanations

    legal or ethical considerations

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

    finds words and tokens expressing possibility or impossibility and the idea that something is difficult (e.g., "possible", "impossible", "task", "monstrous").

    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
    éĴ´
    -0.09
    夺
    -0.08
    çĪ½
    -0.08
    çŃ¾çº¦
    -0.08
    èĢģçĪ·åŃIJ
    -0.08
    _drv
    -0.08
     Sor
    -0.08
    就没
    -0.08
    esc
    -0.08
    æĮ¡ä½ı
    -0.08
    POSITIVE LOGITS
    åIJĪæ³ķ
    0.12
    æ³ķå¾ĭæ³ķè§Ħ
    0.10
    éĿŀæ³ķ
    0.10
    è¿Ļæł·åģļ
    0.10
     ethical
    0.10
    ä»ħä¾Ľ
    0.10
     legality
    0.09
    æ³ķå¾ĭ
    0.09
    亲æīĭ
    0.09
     Eth
    0.09
    Activations Density 0.051%

    No Known Activations