Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsBlogSlackPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Julius Han · Gated SAE for Llama3-8B-Instruct
    3. Llama3-8B-IT
    4. Residual Stream
    5. 25-RES-JH
    6. 15
    Prev
    Next
    INDEX
    Explanations

    references to security policies and audits in a technical context

    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
     Gecko
    -0.16
    iband
    -0.15
    oplayer
    -0.15
    aylight
    -0.14
    enger
    -0.14
    à¹īà¸ĩ
    -0.14
    ZE
    -0.14
    ÑĢава
    -0.14
     Ulus
    -0.14
    amaha
    -0.14
    POSITIVE LOGITS
     Controls
    0.35
     controls
    0.32
    controls
    0.31
    Controls
    0.31
    _controls
    0.28
    -controls
    0.25
     risk
    0.23
    .controls
    0.22
     Ris
    0.22
     risks
    0.21
    Activations Density 0.022%

    No Known Activations