© 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. 23329
    Prev
    Next
    INDEX
    Explanations

    ab prefixes leading to abstraction, abundance, or "ab-" words

    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
    ervice
    -0.09
    ģ
    -0.09
    snap
    -0.09
     обÑıзаÑĤелÑĮно
    -0.08
    âĺī
    -0.08
    åijĺ
    -0.08
    飧æĢ§
    -0.08
     Gems
    -0.08
    ering
    -0.08
    aneously
    -0.08
    POSITIVE LOGITS
    stract
    0.13
    andoned
    0.12
    ridged
    0.12
    STRACT
    0.11
    #ab
    0.10
    -ab
    0.10
    straction
    0.09
    raries
    0.09
    olutely
    0.09
    omination
    0.09
    Activations Density 0.059%

    No Known Activations