© 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. 63814
    Prev
    Next
    INDEX
    Explanations

    phrases related to causing or preventing harm

    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
     Settlement
    -0.15
    ITA
    -0.15
    ůž
    -0.15
     hra
    -0.15
    責
    -0.15
    itoris
    -0.15
    .loads
    -0.14
    ochen
    -0.14
    ssue
    -0.14
    /**<
    -0.14
    POSITIVE LOGITS
     harm
    0.70
     Harm
    0.58
     harms
    0.41
     damage
    0.40
     harmed
    0.38
     injury
    0.36
     hurt
    0.34
     Damage
    0.31
    damage
    0.31
     DAMAGE
    0.28
    Activations Density 0.036%

    No Known Activations