© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports 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. 54969
    Prev
    Next
    INDEX
    Explanations

    phrases expressing caution or advisories

    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
    Äįky
    -0.16
    plib
    -0.15
     Cumberland
    -0.15
    æľĿ
    -0.15
    .lu
    -0.15
    imore
    -0.15
    EOS
    -0.14
    ãĤĮãģ°
    -0.14
    ignKey
    -0.14
    заб
    -0.14
    POSITIVE LOGITS
     careful
    0.53
     caution
    0.48
    care
    0.46
     care
    0.43
     Care
    0.43
     carefully
    0.42
    Care
    0.42
    -care
    0.41
     CARE
    0.37
     cautious
    0.36
    Activations Density 0.048%

    No Known Activations