© 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. 25249
    Prev
    Next
    INDEX
    Explanations

    references to articles and posts related to content or information sharing

    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
     поба
    -0.14
    oller
    -0.13
    739
    -0.13
     Yer
    -0.12
    ollo
    -0.12
    jes
    -0.12
     Alta
    -0.12
    инок
    -0.12
    ometr
    -0.12
     inset
    -0.12
    POSITIVE LOGITS
     will
    0.20
     was
    0.18
     contains
    0.16
     shall
    0.15
     belongs
    0.15
     is
    0.15
     has
    0.15
     assumes
    0.14
     WILL
    0.14
    isco
    0.14
    Activations Density 0.101%

    No Known Activations