© 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. 12897
    Prev
    Next
    INDEX
    Explanations

    references to the audience or direct address to the reader

    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
     themselves
    -0.25
     usted
    -0.18
    ly
    -0.18
     itself
    -0.17
    fn
    -0.14
    edly
    -0.14
    ica
    -0.14
    247
    -0.14
    ÃŃm
    -0.14
    .codehaus
    -0.14
    POSITIVE LOGITS
    /us
    0.39
     guys
    0.33
    -même
    0.30
    åĢij
    0.28
    SELF
    0.27
    uu
    0.27
    ’re
    0.24
    /me
    0.24
    -know
    0.22
    /her
    0.21
    Activations Density 0.098%

    No Known Activations