© 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. 33192
    Prev
    Next
    INDEX
    Explanations

    instances of conditional or promotional phrases related to offers

    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
    icha
    -0.15
    ãĥ¼ãĥ³
    -0.14
    ISBN
    -0.14
    ÑĩиÑĤ
    -0.14
    lemen
    -0.14
    été
    -0.14
    Æ°a
    -0.14
    assage
    -0.14
    ابÙĩ
    -0.13
    ayah
    -0.13
    POSITIVE LOGITS
     stay
    0.38
     receive
    0.33
     Stay
    0.33
    Stay
    0.32
     stays
    0.30
    stay
    0.29
     stayed
    0.29
     receives
    0.28
     staying
    0.28
     Receive
    0.27
    Activations Density 0.042%

    No Known Activations