© 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. 24656
    Prev
    Next
    INDEX
    Explanations

    forms of the verb "to be."

    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
    stin
    -0.16
    ton
    -0.16
    ville
    -0.15
     ìĿ¼ìĿ´
    -0.14
    lyph
    -0.14
    forth
    -0.14
    ucht
    -0.14
    holm
    -0.14
    ard
    -0.13
    jin
    -0.13
    POSITIVE LOGITS
    /is
    0.30
    /w
    0.29
     supposed
    0.27
    nt
    0.20
    /how
    0.20
     gonna
    0.20
     likely
    0.18
    otope
    0.18
     able
    0.17
     perceived
    0.17
    Activations Density 0.080%

    No Known Activations