© 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. 22468
    Prev
    Next
    INDEX
    Explanations

    references to safety and regulatory measures related to technology and environmental conditions

    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
    ãģ¾ãģļ
    -0.16
    izr
    -0.15
     arada
    -0.15
    AndView
    -0.15
     eskort
    -0.14
     withd
    -0.14
    /libs
    -0.14
     ÏħÏĢÏĮ
    -0.14
    iswa
    -0.14
    inç
    -0.14
    POSITIVE LOGITS
     which
    0.58
    which
    0.52
     Which
    0.42
    Which
    0.38
     WHICH
    0.38
     something
    0.37
     wich
    0.32
    something
    0.29
     Something
    0.26
     commonly
    0.26
    Activations Density 0.207%

    No Known Activations