© 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. 43478
    Prev
    Next
    INDEX
    Explanations

    phrases related to the need for comprehensive information and understanding in discussions

    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
     detect
    -0.16
     Detect
    -0.16
    ichel
    -0.15
    ussen
    -0.15
     detection
    -0.14
    iler
    -0.14
    _dark
    -0.14
    æ
    -0.14
    üre
    -0.14
    jong
    -0.14
    POSITIVE LOGITS
     facts
    0.34
    facts
    0.32
     Facts
    0.30
     full
    0.26
     complete
    0.26
     information
    0.23
     incomplete
    0.23
     factual
    0.22
     fuller
    0.22
    complete
    0.21
    Activations Density 0.047%

    No Known Activations