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. 23197
    Prev
    Next
    INDEX
    Explanations

    phrases indicating offers or statements of confirmation

    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
    .
    -0.15
    ex
    -0.15
     forgotten
    -0.15
     unused
    -0.15
    ara
    -0.15
     manual
    -0.14
     vocal
    -0.14
    idal
    -0.14
     full
    -0.14
     argument
    -0.14
    POSITIVE LOGITS
     crypt
    0.38
     vague
    0.38
     generic
    0.35
    -generic
    0.35
     simply
    0.33
    generic
    0.32
     vag
    0.31
     general
    0.30
    _generic
    0.29
    crypt
    0.29
    Activations Density 0.030%

    No Known Activations