© 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. 13079
    Prev
    Next
    INDEX
    Explanations

    instances of negation or stopping in a context that implies comparison or contrast

    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
     $("#"
    -0.14
    ÑĢол
    -0.14
    LC
    -0.14
    ofilm
    -0.14
    oul
    -0.14
     baseline
    -0.14
    anner
    -0.14
    omy
    -0.14
    gs
    -0.14
    est
    -0.14
    POSITIVE LOGITS
     there
    0.30
     stops
    0.27
     stop
    0.25
     THERE
    0.24
     Stop
    0.24
     Stops
    0.24
     stopped
    0.24
    stop
    0.24
    .stop
    0.23
    stopped
    0.23
    Activations Density 0.013%

    No Known Activations