© 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. Julius Han · Gated SAE for Llama3-8B-Instruct
    5. Llama3-8B-IT
    7. Residual Stream
    9. 25-RES-JH
    11. 22590
    Prev
    Next
    INDEX
    Explanations

    references to gendered terms, particularly highlighting women and their association in various contexts

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Configuration
    Juliushanhanhan/llama-3-8b-it-res/blocks.25.hook_resid_post
    Features
    65,536
    Data Type
    float32
    Hook Name
    blocks.25.hook_resid_post
    Hook Layer
    25
    Architecture
    gated
    Context Size
    1,024
    Dataset
    Juliushanhanhan/openwebtext-1b-llama3-tokenized-cxt-1024
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    quia
    -0.19
    Girl
    -0.18
    girl
    -0.17
     Girl
    -0.17
     wife
    -0.16
    ewolf
    -0.16
    eyse
    -0.15
    ÄĻp
    -0.15
    istrov
    -0.15
     Girlfriend
    -0.15
    POSITIVE LOGITS
     gentlemen
    0.69
     gentleman
    0.65
    gent
    0.63
     Gent
    0.63
     gent
    0.62
     Gentle
    0.57
     gentle
    0.45
     ladies
    0.35
     Mess
    0.34
     Ladies
    0.34
    Activations Density 0.009%

    No Known Activations