© 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. 54741
    Prev
    Next
    INDEX
    Explanations

    references to child-related crimes and abuse

    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
    èĥİ
    -0.17
    SSIP
    -0.15
     intros
    -0.14
     xsi
    -0.14
    mbH
    -0.14
    ertype
    -0.14
    íĥģ
    -0.13
     tatto
    -0.13
    upertino
    -0.13
    каÑĤ
    -0.13
    POSITIVE LOGITS
     abuse
    0.25
     mole
    0.24
     sex
    0.22
     sexual
    0.21
     exploitation
    0.21
     pornography
    0.20
     mol
    0.19
     molest
    0.19
    -ab
    0.19
     Abuse
    0.18
    Activations Density 0.014%

    No Known Activations