© 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. 41831
    Prev
    Next
    INDEX
    Explanations

    statements expressing feelings of injustice or unfairness

    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
     Useful
    -0.17
     dangerous
    -0.17
     useless
    -0.17
    ymous
    -0.16
    _deinit
    -0.15
    athan
    -0.15
     risks
    -0.15
     useful
    -0.14
     harmless
    -0.14
    MMdd
    -0.14
    POSITIVE LOGITS
     fairness
    0.50
     unfair
    0.49
     fair
    0.49
     Fair
    0.44
    fair
    0.43
    Fair
    0.43
     injustice
    0.39
     justice
    0.35
     неÑģп
    0.33
     faire
    0.32
    Activations Density 0.036%

    No Known Activations