© 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. 41853
    Prev
    Next
    INDEX
    Explanations

    instances of the word "when" used in various contexts

    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
    krv
    -0.16
    acs
    -0.16
    ekk
    -0.15
    ì°°
    -0.14
    anela
    -0.14
    andro
    -0.13
    isoft
    -0.13
    ÑĢава
    -0.13
     sposób
    -0.13
    Ñģам
    -0.13
    POSITIVE LOGITS
    soever
    0.31
    /if
    0.29
     you
    0.23
    EVER
    0.23
    eve
    0.21
     compared
    0.20
     they
    0.20
    ver
    0.20
     there
    0.19
    -либо
    0.19
    Activations Density 0.277%

    No Known Activations