© 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. 35848
    Prev
    Next
    INDEX
    Explanations

    certainly, clearly, definitelyThis is a phrase of 3 words, which is within the 3-20 word limit. It directly reflects the most frequent and prominent words in MAX_ACTIVATING_TOKENS. It captures the core idea of certainty or emphasis.certainly, clearly, definitely

    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
    åħ¨æĺ¯
    -0.10
    ä¸įçķĻ
    -0.09
    好似
    -0.08
    æĸ¹æīį
    -0.08
    пÑĥÑģка
    -0.08
    ä»Ģä¹Īåij¢
    -0.08
    éĢł
    -0.08
     Dess
    -0.08
     mast
    -0.08
    çļĦå¿ĥ
    -0.08
    POSITIVE LOGITS
    æĶ¶æķĽ
    0.11
    ãģ¨ãģĦãģ£ãģ¦
    0.10
    lix
    0.09
    ä¸ĢèĪ¬æĿ¥è¯´
    0.09
    oyer
    0.09
    ä¼Ĭæĭī
    0.09
    æľīåı¯èĥ½
    0.09
    åİŁåĪĻä¸Ĭ
    0.08
    ìĭľ
    0.08
    ä¼ļå½±åĵį
    0.08
    Activations Density 0.032%

    No Known Activations