© 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. 43962
    Prev
    Next
    INDEX
    Explanations

    statistical data and references to numbers

    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
    hin
    -0.15
     olab
    -0.14
     sometimes
    -0.14
    .Packet
    -0.14
    ãģĦãģ®
    -0.14
     Sometimes
    -0.14
    òn
    -0.14
    ////////////////////////////////////////////////////////////////
    -0.13
     occasionally
    -0.13
     вообÑīе
    -0.13
    POSITIVE LOGITS
     likely
    1.04
     probably
    0.91
    likely
    0.86
     Likely
    0.79
    probably
    0.77
     Probably
    0.75
    Probably
    0.71
     probable
    0.69
     most
    0.53
     likelihood
    0.53
    Activations Density 0.102%

    No Known Activations