© 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. 29619
    Prev
    Next
    INDEX
    Explanations

    conversational phrases and expressions that convey a sense of reasoning or justification

    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
     whichever
    -0.15
    noch
    -0.15
    ÄĻk
    -0.15
    ãĢħ
    -0.15
    enaire
    -0.13
    ylene
    -0.13
    .asp
    -0.13
    de
    -0.13
    dee
    -0.13
    λιά
    -0.12
    POSITIVE LOGITS
     why
    0.66
    why
    0.52
     Why
    0.51
     who
    0.48
    Why
    0.47
    who
    0.39
     WHY
    0.39
    为ä»Ģä¹Ī
    0.37
     pourquoi
    0.36
     Who
    0.36
    Activations Density 0.026%

    No Known Activations