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

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI 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. 44343
    Prev
    Next
    INDEX
    Explanations

    terms related to public relations and organizational communication

    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
    лаб
    -0.16
    orrent
    -0.14
     meslek
    -0.14
    urence
    -0.14
    imité
    -0.14
    ragen
    -0.13
    ucid
    -0.13
    wner
    -0.13
    lum
    -0.13
    áºŃu
    -0.13
    POSITIVE LOGITS
     PR
    0.78
    PR
    0.65
     press
    0.54
    _PR
    0.47
     Press
    0.47
     communications
    0.46
    (PR
    0.44
     Communications
    0.42
     media
    0.41
     publicity
    0.40
    Activations Density 0.034%

    No Known Activations