Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Julius Han · Gated SAE for Llama3-8B-Instruct
    3. Llama3-8B-IT
    4. Residual Stream
    5. 25-RES-JH
    6. 23475
    Prev
    Next
    INDEX
    Explanations

    the definite article "the" and other frequently occurring function words that help establish connections in the text

    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
     stuff
    -0.14
    ars
    -0.14
     that
    -0.14
    3
    -0.14
     things
    -0.14
     personal
    -0.14
     finally
    -0.14
     hundreds
    -0.13
     consistently
    -0.13
    anto
    -0.13
    POSITIVE LOGITS
    ä¸įè¶³
    0.14
    ecure
    0.14
    tesy
    0.14
    öh
    0.13
     pÅĻeklad
    0.13
    erland
    0.13
    ÃĹ</
    0.13
    ̣
    0.13
    itably
    0.12
    prit
    0.12
    Activations Density 0.009%

    No Known Activations