© 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. 25050
    Prev
    Next
    INDEX
    Explanations

    mentions of "state machines" and related concepts

    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
    _states
    -0.15
    æݪ
    -0.15
    inne
    -0.14
    _stmt
    -0.14
    ожеÑĤ
    -0.14
    sak
    -0.14
    ÏĢο
    -0.14
    anmar
    -0.14
    ulk
    -0.14
    neau
    -0.14
    POSITIVE LOGITS
    -of
    0.35
    craft
    0.33
    ful
    0.33
    hood
    0.28
    MENTS
    0.27
    fully
    0.26
    less
    0.24
    fulness
    0.24
    MENT
    0.23
    wide
    0.22
    Activations Density 0.048%

    No Known Activations