© 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. 54924
    Prev
    Next
    INDEX
    Explanations

    references to the concept of "newness."

    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
    aju
    -0.16
    irs
    -0.14
    avar
    -0.14
    epar
    -0.13
    ÄĻ
    -0.13
     almost
    -0.13
     congr
    -0.13
    aris
    -0.13
    ottle
    -0.13
    리ì¦Ī
    -0.13
    POSITIVE LOGITS
     nor
    0.48
     either
    0.38
    nor
    0.36
     anymore
    0.34
    Nor
    0.32
    either
    0.31
     EITHER
    0.31
     Nor
    0.29
     Either
    0.28
    Either
    0.27
    Activations Density 0.174%

    No Known Activations