© 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. 44356
    Prev
    Next
    INDEX
    Explanations

    phrases related to hope and potential change

    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
    лад
    -0.15
    getSingleton
    -0.15
    umper
    -0.15
    ži
    -0.14
    abay
    -0.14
    Å¡t
    -0.14
    бÑĭ
    -0.14
    Resolve
    -0.14
    ãĤīãģļ
    -0.13
    ?action
    -0.13
    POSITIVE LOGITS
     slowly
    0.26
     gradually
    0.25
     eventual
    0.24
     chip
    0.22
     eventually
    0.21
     gradual
    0.21
     Slow
    0.20
     Begins
    0.20
    slow
    0.19
     begins
    0.19
    Activations Density 0.034%

    No Known Activations