© 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. 54701
    Prev
    Next
    INDEX
    Explanations

    adjectives and descriptors related to negative traits or behaviors

    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
    สม
    -0.15
    edium
    -0.14
    antino
    -0.14
     SKIP
    -0.14
    ær
    -0.14
     NSK
    -0.13
    änn
    -0.13
    zing
    -0.13
     Rout
    -0.13
    indo
    -0.13
    POSITIVE LOGITS
    gre
    0.51
     Gre
    0.49
     greedy
    0.47
     greed
    0.45
    Gre
    0.42
     GRE
    0.41
     selfish
    0.35
     rap
    0.34
    GRE
    0.33
     gre
    0.31
    Activations Density 0.040%

    No Known Activations