© 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. 22122
    Prev
    Next
    INDEX
    Explanations

    references to geopolitical conflicts or tensions

    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
    -threat
    -0.17
    ynchronously
    -0.16
    Ø´ÙħاÙĦÛĮ
    -0.16
    inear
    -0.15
    avax
    -0.15
     поба
    -0.15
    atform
    -0.15
    RenderingContext
    -0.14
    ibur
    -0.14
    OrFail
    -0.14
    POSITIVE LOGITS
     web
    0.20
     webs
    0.20
     loop
    0.19
     cycle
    0.18
    web
    0.18
     lim
    0.17
     Caught
    0.17
     vice
    0.17
    2
    0.17
     traffic
    0.17
    Activations Density 0.025%

    No Known Activations