© 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. 44416
    Prev
    Next
    INDEX
    Explanations

    conjunctions and transitions in sentences, particularly 'but'

    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
    odega
    -0.16
    UNUSED
    -0.15
    kaar
    -0.14
    odesk
    -0.14
    okit
    -0.14
    @qq
    -0.14
    ellas
    -0.14
    quina
    -0.14
    plusplus
    -0.14
    ickness
    -0.14
    POSITIVE LOGITS
     until
    0.58
     Until
    0.54
    Until
    0.50
     for
    0.45
    until
    0.45
     till
    0.45
     meanwhile
    0.42
     meantime
    0.38
     Meanwhile
    0.38
     hasta
    0.37
    Activations Density 0.043%

    No Known Activations