© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    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. 42171
    Prev
    Next
    INDEX
    Explanations

    phrases emphasizing authority or directive language

    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
    kek
    -0.15
    dda
    -0.15
    _into
    -0.15
    itten
    -0.14
    ada
    -0.14
     Shel
    -0.14
    :focus
    -0.14
    ilen
    -0.14
    à¹ĥห
    -0.14
    ól
    -0.14
    POSITIVE LOGITS
     bone
    0.20
     tune
    0.20
    enny
    0.19
     accom
    0.19
     maxx
    0.19
     letter
    0.18
    Bone
    0.17
     Bone
    0.17
    bone
    0.17
    inize
    0.16
    Activations Density 0.033%

    No Known Activations