© 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. 22953
    Prev
    Next
    INDEX
    Explanations

    conversational cues and phrases indicating actions or instructions

    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
    COMPARE
    -0.15
    auer
    -0.14
    Arrange
    -0.14
    onder
    -0.13
    ç½
    -0.13
     ç¿
    -0.13
    mán
    -0.13
    reshape
    -0.13
     Decide
    -0.12
    Unlock
    -0.12
    POSITIVE LOGITS
     delegate
    0.22
     never
    0.22
     Delegate
    0.21
     stay
    0.21
     Surround
    0.21
    Stay
    0.19
    Delegate
    0.19
     Stay
    0.19
     surround
    0.19
    never
    0.18
    Activations Density 0.066%

    No Known Activations