© 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 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. 43116
    Prev
    Next
    INDEX
    Explanations

    negative expressions or phrases regarding official reports and conditions

    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
     báºŃc
    -0.14
     Whatever
    -0.13
    hazi
    -0.13
    itest
    -0.13
     hopefully
    -0.13
    compareTo
    -0.13
     æ¨
    -0.13
    mast
    -0.13
    маÑĢ
    -0.12
    ارد
    -0.12
    POSITIVE LOGITS
     anymore
    0.24
     properly
    0.23
     yet
    0.22
    proper
    0.22
     adequately
    0.19
    ching
    0.19
     originally
    0.19
     able
    0.19
    ched
    0.19
    oriously
    0.18
    Activations Density 0.320%

    No Known Activations