© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI 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. 22739
    Prev
    Next
    INDEX
    Explanations

    instances of the word "roll" and its variations 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
    riel
    -0.17
    liers
    -0.17
    949
    -0.15
    810
    -0.15
    ebe
    -0.15
    fir
    -0.15
    ãĥ¯ãĥ¼
    -0.14
    جÙĪÛĮ
    -0.14
    oba
    -0.14
    utron
    -0.14
    POSITIVE LOGITS
     Roll
    0.44
     roll
    0.42
     rolling
    0.42
    Roll
    0.41
    -roll
    0.40
     rolls
    0.39
     Rolling
    0.38
     rolled
    0.38
    .roll
    0.36
    .Roll
    0.36
    Activations Density 0.014%

    No Known Activations