© 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. 54121
    Prev
    Next
    INDEX
    Explanations

    the word "can" in various contexts, indicating potential or capability

    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
    atron
    -0.15
    irs
    -0.14
    robe
    -0.14
    à¸ĵะ
    -0.13
    .Îij
    -0.13
    edom
    -0.13
    rottle
    -0.13
    ãĤ´ãĥª
    -0.13
    овиÑĩ
    -0.13
    ailer
    -0.13
    POSITIVE LOGITS
    't
    0.33
    ’t
    0.31
     sometimes
    0.30
     be
    0.28
     often
    0.25
    sometimes
    0.23
    berra
    0.22
     Sometimes
    0.22
    opies
    0.22
    often
    0.21
    Activations Density 0.194%

    No Known Activations