© 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. 42592
    Prev
    Next
    INDEX
    Explanations

    sentences that emphasize the importance of understanding and critical thinking

    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
     they
    -0.17
     their
    -0.16
     THEIR
    -0.16
     Their
    -0.16
    Their
    -0.15
    they
    -0.15
    their
    -0.15
     Due
    -0.15
     due
    -0.15
     They
    -0.14
    POSITIVE LOGITS
     successful
    0.19
     Successful
    0.18
    .scalablytyped
    0.17
     ÑĥÑģпеÑĪ
    0.16
     réuss
    0.16
    avia
    0.16
     yourself
    0.16
    Successful
    0.15
    successful
    0.15
     sucesso
    0.15
    Activations Density 0.018%

    No Known Activations