© 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. 25753
    Prev
    Next
    INDEX
    Explanations

    legal terminology related to negligence and responsibility

    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
    itimate
    -0.16
    appropri
    -0.15
    zano
    -0.14
    çIJ
    -0.14
     misunder
    -0.13
    Wheel
    -0.13
    mute
    -0.13
    목
    -0.13
     wym
    -0.13
    ippi
    -0.13
    POSITIVE LOGITS
     Neg
    0.51
    Neg
    0.44
    neg
    0.43
     care
    0.40
     neg
    0.39
     negligence
    0.35
     NEGLIGENCE
    0.34
     reck
    0.34
     neglect
    0.33
    _neg
    0.31
    Activations Density 0.073%

    No Known Activations