© 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 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. 43412
    Prev
    Next
    INDEX
    Explanations

    phrases that highlight key issues or concerns within a political context

    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
     predomin
    -0.13
    TargetException
    -0.13
    alls
    -0.13
    lier
    -0.13
     Necessary
    -0.13
    otime
    -0.13
     zas
    -0.13
     Vys
    -0.13
    _il
    -0.12
    λα
    -0.12
    POSITIVE LOGITS
     most
    0.58
     biggest
    0.53
     greatest
    0.44
    most
    0.40
     main
    0.37
     MOST
    0.34
    æľĢ大
    0.34
     ê°Ģìŀ¥
    0.33
    -most
    0.33
    æľĢ
    0.31
    Activations Density 0.133%

    No Known Activations