© 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. 30425
    Prev
    Next
    INDEX
    Explanations

    instances of the word "to" and variations, suggesting a focus on purpose or intention in the text

    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
    ladu
    -0.16
    vek
    -0.15
    IIIK
    -0.14
    _mC
    -0.14
    ofday
    -0.14
    ringe
    -0.14
    obierno
    -0.14
    ɵ
    -0.14
    amina
    -0.14
    족
    -0.14
    POSITIVE LOGITS
     otherwise
    0.46
     already
    0.38
     Otherwise
    0.36
    otherwise
    0.34
     OTHERWISE
    0.33
     Already
    0.32
    already
    0.32
     proceedings
    0.30
     existing
    0.29
    Otherwise
    0.29
    Activations Density 0.060%

    No Known Activations