© 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. 13485
    Prev
    Next
    INDEX
    Explanations

    expressions related to danger and fatal outcomes

    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
     slight
    -0.18
     seemed
    -0.17
     apparently
    -0.16
     must
    -0.16
     almost
    -0.16
     MUST
    -0.16
     blah
    -0.15
     Gad
    -0.15
     weak
    -0.14
     immense
    -0.14
    POSITIVE LOGITS
    elsing
    0.17
     tô
    0.17
     serious
    0.16
     EITHER
    0.16
    .Guna
    0.15
    lemn
    0.15
    ddit
    0.15
    лоп
    0.15
     поÑĤен
    0.15
    cona
    0.15
    Activations Density 0.029%

    No Known Activations