© 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. 25400
    Prev
    Next
    INDEX
    Explanations

    instances of the word "this"

    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
     mne
    -0.14
    145
    -0.14
    -0.13
    ALAR
    -0.13
     beginnings
    -0.13
    ous
    -0.13
    ful
    -0.13
    180
    -0.12
     grains
    -0.12
    umb
    -0.12
    POSITIVE LOGITS
     week
    0.25
    maal
    0.18
     morning
    0.18
     year
    0.17
     day
    0.17
     particular
    0.17
     far
    0.17
     keer
    0.17
     time
    0.16
     inning
    0.16
    Activations Density 0.084%

    No Known Activations