© 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
    2. Julius Han · Gated SAE for Llama3-8B-Instruct
    3. Llama3-8B-IT
    4. Residual Stream
    5. 25-RES-JH
    6. 54801
    Prev
    Next
    INDEX
    Explanations

    instances of the word "up" and its variations in context

    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
    t
    -0.18
    tube
    -0.17
    undler
    -0.17
     upright
    -0.16
    esses
    -0.16
    ر
    -0.16
     up
    -0.15
    esting
    -0.15
    woord
    -0.15
     upbeat
    -0.15
    POSITIVE LOGITS
    /down
    0.36
    datable
    0.28
    pers
    0.27
    ping
    0.26
    ped
    0.25
    sert
    0.24
    wards
    0.23
    turned
    0.23
    dater
    0.21
    gradable
    0.21
    Activations Density 0.176%

    No Known Activations