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. 23417
    Prev
    Next
    INDEX
    Explanations

    references to alternative medicine and quackery

    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
    i
    -0.23
    ÛĮ
    -0.21
    и
    -0.21
    iem
    -0.20
    iard
    -0.20
    Ùĩ
    -0.19
    ing
    -0.19
    ÙĬ
    -0.18
    osu
    -0.18
    auss
    -0.18
    POSITIVE LOGITS
    itionally
    0.20
    pole
    0.20
    urai
    0.19
    ossal
    0.19
    hy
    0.18
    oin
    0.18
    ges
    0.18
    deo
    0.18
    dest
    0.18
    venture
    0.18
    Activations Density 0.022%

    No Known Activations