© 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. 23000
    Prev
    Next
    INDEX
    Explanations

    negative phrases or terms indicating a lack of solutions or recognition for issues

    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
    enga
    -0.14
    iaz
    -0.14
    asio
    -0.14
     importantly
    -0.14
    à¹Īาà¸ĩà¹Ĩ
    -0.13
    pNet
    -0.13
    eÅŁit
    -0.13
    ego
    -0.12
    ä¸įåIJĮçļĦ
    -0.12
    ãģıãģł
    -0.12
    POSITIVE LOGITS
     direct
    0.27
     explicit
    0.27
     equivalent
    0.25
     official
    0.25
     clear
    0.25
     dedicated
    0.23
     reliable
    0.23
     precise
    0.22
     equivalents
    0.22
     officially
    0.22
    Activations Density 0.236%

    No Known Activations