© 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. Qwen2.5-7B-IT
    5. 20-MATRYOSHKA-65K
    7. 3770
    Prev
    Next
    INDEX
    Explanations

    reduction cost

    np_max-act-logits · gemini-2.5-flash-lite
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    chanind/qwen2.5-7B-it-layer-20-saes/pile/matryoshka/k-100
    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
    aph
    -0.23
    :[[
    -0.22
    ä¹ı
    -0.21
    haul
    -0.21
    容éĩı
    -0.20
    è§
    -0.20
    Weak
    -0.20
     weak
    -0.20
    ADIO
    -0.19
    isoner
    -0.19
    POSITIVE LOGITS
     Costs
    0.29
    æĪIJæľ¬
    0.29
     costs
    0.28
    cost
    0.27
    Cost
    0.27
     cost
    0.26
     Cost
    0.25
    _cost
    0.24
     componentWillUnmount
    0.23
    Reduc
    0.23
    Activations Density 0.546%

    No Known Activations