© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW 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. 21318
    Prev
    Next
    INDEX
    Explanations

    phrases involving negative stereotypes or criminal connotations associated with Mexicans

    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
    Negative Logits
    inho
    -0.16
    shr
    -0.15
    ãĥ¼ãĥĩ
    -0.15
    iais
    -0.15
    lear
    -0.15
     Shr
    -0.14
    yme
    -0.14
    arget
    -0.14
    uby
    -0.14
    hea
    -0.14
    POSITIVE LOGITS
     anchor
    0.18
     æ½
    0.17
    rap
    0.17
     rap
    0.17
     moo
    0.17
     Rap
    0.16
     Anchor
    0.16
    anium
    0.16
     freel
    0.15
    .sap
    0.15
    Activations Density 0.034%

    No Known Activations