© 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 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. 14866
    Prev
    Next
    INDEX
    Explanations

    phrases indicating functionality and ease of use

    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
    ستاÙĨ
    -0.15
    sing
    -0.15
    ylie
    -0.14
    ิà¸ĩ
    -0.14
     Saves
    -0.13
    eba
    -0.13
     cle
    -0.13
    ullan
    -0.13
    egis
    -0.13
     depending
    -0.13
    POSITIVE LOGITS
     remedy
    0.53
     remedies
    0.47
     rect
    0.46
     remed
    0.42
     overcome
    0.40
    解åĨ³
    0.39
     circum
    0.39
     Remed
    0.38
     solve
    0.36
     Rect
    0.35
    Activations Density 0.051%

    No Known Activations