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. 23661
    Prev
    Next
    INDEX
    Explanations

    negative attributes associated with failure and injustice

    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
    qui
    -0.15
    ált
    -0.15
     somebody
    -0.14
    (optional
    -0.14
    iola
    -0.14
    atcher
    -0.14
    oda
    -0.14
    linkplain
    -0.14
    antar
    -0.14
    ran
    -0.13
    POSITIVE LOGITS
     regardless
    0.67
    Regardless
    0.64
     Regardless
    0.61
     whether
    0.54
    whether
    0.46
    ardless
    0.44
    Whether
    0.44
     Whether
    0.43
     WHETHER
    0.38
     whatever
    0.38
    Activations Density 0.085%

    No Known Activations