© 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. 22243
    Prev
    Next
    INDEX
    Explanations

    words and phrases indicating opposition or negativity towards a subject

    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
    j
    -0.17
    chan
    -0.17
    aset
    -0.17
    æİĽ
    -0.16
    van
    -0.16
    à¹Ģà¸Ń
    -0.15
    vis
    -0.15
    entifier
    -0.15
    ji
    -0.15
    vol
    -0.15
    POSITIVE LOGITS
    logue
    0.20
    plate
    0.18
    heroes
    0.18
    wner
    0.18
    ennes
    0.17
    yte
    0.17
    reeze
    0.17
    adir
    0.17
    establish
    0.17
    wand
    0.17
    Activations Density 0.016%

    No Known Activations