© 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 EvalsExportsAPI 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. 43367
    Prev
    Next
    INDEX
    Explanations

    instances of the word "escape" or its variations related to evading capture or danger

    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
    yen
    -0.16
    helm
    -0.16
    locker
    -0.15
    Å¡it
    -0.15
     EntryPoint
    -0.14
    hud
    -0.14
    sus
    -0.14
    ukan
    -0.14
    .ws
    -0.14
    uld
    -0.14
    POSITIVE LOGITS
     from
    0.25
     detection
    0.23
     notice
    0.23
    from
    0.20
     Detection
    0.19
     capture
    0.18
     NOTICE
    0.17
     bounds
    0.17
     khá»ıi
    0.16
     Notice
    0.16
    Activations Density 0.012%

    No Known Activations