© 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. 55153
    Prev
    Next
    INDEX
    Explanations

    expressions of personal struggle and the use of the first person perspective

    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
    lor
    -0.15
    fp
    -0.14
    oldown
    -0.14
    f
    -0.14
    Å©
    -0.14
     Flores
    -0.14
    esti
    -0.13
     Baron
    -0.13
    ukan
    -0.13
    pectral
    -0.13
    POSITIVE LOGITS
    'll
    0.45
    ’ll
    0.43
     will
    0.42
     shall
    0.34
    will
    0.34
     WILL
    0.30
     sẽ
    0.29
     akan
    0.29
     SHALL
    0.25
    ä¼ļ
    0.25
    Activations Density 0.065%

    No Known Activations