© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Julius Han · Gated SAE for Llama3-8B-Instruct
    5. Llama3-8B-IT
    7. Residual Stream
    9. 25-RES-JH
    11. 33544
    Prev
    Next
    INDEX
    Explanations

    expressions related to evaluations of situations, particularly emphasizing the concepts of "worst" and "best."

    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
     whom
    -0.17
    /XMLSchema
    -0.16
    Ĵáŀ
    -0.14
    ì¦ĺ
    -0.14
     yours
    -0.14
     Yours
    -0.14
     youre
    -0.14
     Lit
    -0.14
    ckett
    -0.14
    åĩĢ
    -0.14
    POSITIVE LOGITS
     which
    0.36
    which
    0.35
    that
    0.33
     that
    0.32
     ÏĢοÏħ
    0.29
     yang
    0.28
    ãģĮ
    0.28
     Which
    0.28
     WHICH
    0.28
    Which
    0.27
    Activations Density 0.019%

    No Known Activations