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. 31390
    Prev
    Next
    INDEX
    Explanations

    specific capitalized letters, which likely indicate the presence of names, brands, or important entities

    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
    $MESS
    -0.16
    Ø©
    -0.15
    oultry
    -0.14
    ENCIL
    -0.13
    .Generation
    -0.13
    complete
    -0.13
    zelf
    -0.13
    SW
    -0.13
    decess
    -0.12
    acji
    -0.12
    POSITIVE LOGITS
    etheless
    0.37
    adays
    0.34
    odore
    0.32
    bsites
    0.23
    atomy
    0.23
    tlement
    0.23
    ÑįÑĤомÑĥ
    0.22
    gether
    0.22
    atre
    0.21
    xiety
    0.20
    Activations Density 0.414%

    No Known Activations