© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW 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. 12083
    Prev
    Next
    INDEX
    Explanations

    instances of abrupt changes or transformations

    oai_token-act-pair · gpt-4o-mini
    Triggered 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
    Negative Logits
     eventually
    -0.32
     eventual
    -0.30
     ultimately
    -0.26
     soon
    -0.24
     Eventually
    -0.24
    soon
    -0.21
    Eventually
    -0.21
     finally
    -0.20
     Ultimately
    -0.19
     ultimate
    -0.19
    POSITIVE LOGITS
     become
    0.22
    heimer
    0.21
    aneously
    0.20
     became
    0.19
     everywhere
    0.19
    bec
    0.18
     becomes
    0.18
    Become
    0.18
    aneous
    0.17
    sworth
    0.16
    Activations Density 0.025%

    No Known Activations