© 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. 50046
    Prev
    Next
    INDEX
    Explanations

    fill level, combine vehicle, stainings, wrap content, surround stimuli, chewers, compute blades, drill pipe

    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
    ized
    -0.15
    ities
    -0.11
    ize
    -0.11
    naire
    -0.11
    ãģĦãĤĭ
    -0.10
    ocs
    -0.09
    å®Įæķ´çļĦ
    -0.09
    ity
    -0.09
    izing
    -0.09
    izes
    -0.09
    POSITIVE LOGITS
    ings
    0.29
    able
    0.28
    ABLE
    0.22
    ables
    0.20
    ers
    0.20
    ÂŃing
    0.20
    ingly
    0.20
    ability
    0.18
    ableObject
    0.18
    ments
    0.17
    Activations Density 1.366%

    No Known Activations