© 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. 24529
    Prev
    Next
    INDEX
    Explanations

    references to gossip and negativity within a community context, particularly related to the theater and Broadway

    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
    .mapbox
    -0.16
     pokoj
    -0.14
    Gro
    -0.14
    206
    -0.14
    æŃ·
    -0.13
    _IOCTL
    -0.13
    ÏĦÏī
    -0.13
    è¬Ŀ
    -0.13
    ève
    -0.13
    ems
    -0.13
    POSITIVE LOGITS
     rumor
    0.49
     Rum
    0.48
     gossip
    0.46
     rumors
    0.45
     rum
    0.45
    rum
    0.41
    ossip
    0.38
     grape
    0.33
     whispers
    0.33
     whispered
    0.31
    Activations Density 0.042%

    No Known Activations