© 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. 13460
    Prev
    Next
    INDEX
    Explanations

    words and phrases indicating falsehood or deception related to narratives or actions

    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
    isContained
    -0.17
    .SizeType
    -0.17
    cmpeq
    -0.16
    -lnd
    -0.15
    lacak
    -0.15
    .LookAndFeel
    -0.15
    IIIK
    -0.15
    çĽijåIJ¬é¡µéĿ¢
    -0.15
    incinn
    -0.15
    -REAL
    -0.15
    POSITIVE LOGITS
    âĢ¦↵
    0.16
    Ø
    0.14
    amp
    0.13
    _
    0.12
    âĢ¦”
    0.12
    alt
    0.12
    Ñĸ
    0.12
     ----------------------------------------------------------------------------↵
    0.12
     Dedicated
    0.12
    ...↵
    0.12
    Activations Density 0.051%

    No Known Activations