© 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. 42625
    Prev
    Next
    INDEX
    Explanations

    instances of the word "vile" and its derivatives, often in contexts related to negative behavior or qualities

    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
    all
    -0.21
    ../../../
    -0.20
    elling
    -0.18
    line
    -0.18
    ancellable
    -0.18
    flix
    -0.17
    liner
    -0.17
    ylvania
    -0.16
    eller
    -0.16
    islav
    -0.16
    POSITIVE LOGITS
    ÅĽmy
    0.21
    keiten
    0.21
    antro
    0.20
    ÑģÑĮ
    0.19
    ERCHANT
    0.19
    arity
    0.19
    rious
    0.18
    abeth
    0.18
    Ùĩ
    0.18
    #"
    0.18
    Activations Density 0.013%

    No Known Activations