© 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. 22348
    Prev
    Next
    INDEX
    Explanations

    phrases related to the reporting and acknowledgment of sexual assault

    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
    untime
    -0.16
    oyer
    -0.16
    ürn
    -0.15
    elman
    -0.15
    ucken
    -0.15
    ģında
    -0.15
    ivot
    -0.14
     ä¹
    -0.14
    寸
    -0.14
    -pocket
    -0.14
    POSITIVE LOGITS
     coming
    0.47
     come
    0.42
    coming
    0.39
     Coming
    0.39
    Coming
    0.37
     comes
    0.36
     Come
    0.36
    come
    0.35
     came
    0.35
     speak
    0.33
    Activations Density 0.038%

    No Known Activations