© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI 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. 53881
    Prev
    Next
    INDEX
    Explanations

    references to abuse and hostility, particularly towards individuals identified as Muslim or LGBTQ+

    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
    λει
    -0.14
    åį·
    -0.14
     rij
    -0.14
     plunder
    -0.14
    itnÃŃ
    -0.14
    alar
    -0.14
    745
    -0.14
    rians
    -0.14
     handjob
    -0.14
    pendicular
    -0.14
    POSITIVE LOGITS
     vit
    0.32
     epith
    0.25
     hurl
    0.23
     hate
    0.23
     venom
    0.22
     abuse
    0.22
     bile
    0.22
     threats
    0.22
     threatened
    0.21
     hur
    0.21
    Activations Density 0.144%

    No Known Activations