© 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. 15378
    Prev
    Next
    INDEX
    Explanations

    expressions of personal reflection and emotional sentiment related to experiences

    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
     ↵
    -0.17
     sez
    -0.16
     ""
    -0.16
     "...
    -0.15
     begin
    -0.15
    aus
    -0.15
     insist
    -0.15
     demonstrate
    -0.15
     dashed
    -0.15
     "
    -0.15
    POSITIVE LOGITS
     —↵↵
    0.15
     derec
    0.15
    [out
    0.15
    cház
    0.15
    ROC
    0.15
    ponge
    0.14
    .MixedReality
    0.14
    ÙħÙĪد
    0.14
     kli
    0.14
    ARIANT
    0.14
    Activations Density 0.619%

    No Known Activations