© 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. 13049
    Prev
    Next
    INDEX
    Explanations

    punctuation marks and their surrounding 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
     honors
    -0.20
    neighbor
    -0.20
     honored
    -0.19
     Theater
    -0.18
     favored
    -0.18
     honoring
    -0.18
    Center
    -0.18
     Center
    -0.17
     theater
    -0.17
     colors
    -0.17
    POSITIVE LOGITS
     Ottawa
    0.36
     RCMP
    0.35
     Toronto
    0.34
     Canadians
    0.34
    CBC
    0.34
     Winnipeg
    0.34
     Canada
    0.34
     CBC
    0.33
    Canada
    0.33
     Canadian
    0.33
    Activations Density 0.053%

    No Known Activations