© 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. 28358
    Prev
    Next
    INDEX
    Explanations

    instances of the word "when" to focus on conditions or situations

    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
    anik
    -0.17
    ulls
    -0.15
    oid
    -0.15
    qli
    -0.14
    edList
    -0.14
     createState
    -0.14
    جاد
    -0.14
    ilities
    -0.14
    çŃ¾
    -0.14
    .MaxLength
    -0.14
    POSITIVE LOGITS
     faced
    0.22
    /if
    0.20
    EVER
    0.18
     even
    0.18
     facing
    0.17
     done
    0.17
     dealing
    0.16
     confronted
    0.16
     did
    0.16
     every
    0.16
    Activations Density 0.038%

    No Known Activations