© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports 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. 63355
    Prev
    Next
    INDEX
    Explanations

    references to the word "back."

    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
    avax
    -0.17
    eka
    -0.15
    raž
    -0.15
    ugas
    -0.15
    zyst
    -0.15
    anges
    -0.15
    uco
    -0.14
    uld
    -0.14
    инов
    -0.14
     antlr
    -0.14
    POSITIVE LOGITS
    slash
    0.28
    side
    0.28
    country
    0.25
    wards
    0.24
    scatter
    0.24
    gam
    0.22
    woods
    0.22
    ends
    0.21
    hand
    0.21
    hoe
    0.21
    Activations Density 0.048%

    No Known Activations