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

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI 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. 13370
    Prev
    Next
    INDEX
    Explanations

    instances of the word "paradox" and its variations

    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
    å¹ķ
    -0.18
    anik
    -0.15
    νÏĮ
    -0.14
    unde
    -0.14
    deme
    -0.14
    neh
    -0.13
    ativo
    -0.13
    itted
    -0.13
    bread
    -0.13
    ัà¸ķà¸ĸ
    -0.13
    POSITIVE LOGITS
    ical
    0.35
    ically
    0.25
     Resolution
    0.22
     resolution
    0.22
     resolutions
    0.20
     resolved
    0.19
    ICAL
    0.19
    -resolution
    0.19
    Resolution
    0.18
    resolved
    0.18
    Activations Density 0.017%

    No Known Activations