© 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. 13438
    Prev
    Next
    INDEX
    Explanations

    phrases and terms related to "considerations" or "factors" in discussions, particularly in decision-making contexts

    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
    ανα
    -0.17
     còn
    -0.17
    ych
    -0.17
    ura
    -0.16
    HEST
    -0.16
    /her
    -0.15
    339
    -0.15
    lord
    -0.15
    ť
    -0.15
    uese
    -0.15
    POSITIVE LOGITS
    ately
    0.30
    ate
    0.26
    ably
    0.23
    ation
    0.20
    ering
    0.19
    ations
    0.18
    aires
    0.18
    ance
    0.17
    aire
    0.17
     carefully
    0.17
    Activations Density 0.026%

    No Known Activations