© 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. 42195
    Prev
    Next
    INDEX
    Explanations

    instances of the word "enough."

    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
     ê´Ģíķľ
    -0.15
    _ENCODE
    -0.14
    ContentPane
    -0.14
    acco
    -0.14
    è«
    -0.13
    oad
    -0.13
    FFFFFFFF
    -0.13
    ulis
    -0.13
    acia
    -0.13
    impl
    -0.13
    POSITIVE LOGITS
     to
    0.36
     alone
    0.24
     Äijá»ĥ
    0.23
     for
    0.22
     Alone
    0.21
    alone
    0.21
     reason
    0.20
     ÑĩÑĤобÑĭ
    0.20
    to
    0.18
    reason
    0.18
    Activations Density 0.038%

    No Known Activations