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. 23597
    Prev
    Next
    INDEX
    Explanations

    expressions related to identity and cultural significance

    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
     Sesso
    -0.14
    riter
    -0.14
     gratuites
    -0.14
    è¶£
    -0.14
    edly
    -0.14
     Bender
    -0.13
    onium
    -0.13
    quan
    -0.13
    orra
    -0.13
    oth
    -0.13
    POSITIVE LOGITS
    Ùħد
    0.16
     завиÑģим
    0.15
    ÑĤаж
    0.15
    ruit
    0.15
    obil
    0.14
    uvo
    0.14
    ováno
    0.14
     createState
    0.14
    eing
    0.14
    unable
    0.14
    Activations Density 0.009%

    No Known Activations