© 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. 21804
    Prev
    Next
    INDEX
    Explanations

    expressions of positive emotions and feelings of happiness

    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
    ĩ
    -0.14
    NullOr
    -0.13
    alach
    -0.13
     Roths
    -0.13
    /classes
    -0.13
    ilan
    -0.13
    155
    -0.13
    orro
    -0.13
    uala
    -0.13
    726
    -0.13
    POSITIVE LOGITS
     to
    0.34
     that
    0.33
    that
    0.25
     rằng
    0.21
    	that
    0.20
    	to
    0.19
     bahwa
    0.18
     että
    0.18
     they
    0.18
     ÑĩÑĤо
    0.17
    Activations Density 0.076%

    No Known Activations