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

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI 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. 25706
    Prev
    Next
    INDEX
    Explanations

    quantities, specifically the word "two" and variations of "three" and "six."

    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
    /OR
    -0.20
    nock
    -0.16
    ForObject
    -0.15
    æĹıèĩªæ²»
    -0.15
     Redistributions
    -0.15
    ãģĵãģĿ
    -0.14
    ActionBar
    -0.14
    ulton
    -0.14
     ActiveForm
    -0.14
    ORED
    -0.14
    POSITIVE LOGITS
     or
    0.64
    æĪĸ
    0.47
     или
    0.45
     atau
    0.44
     oder
    0.40
    æĪĸèĢħ
    0.39
     hoặc
    0.38
     ÛĮا
    0.38
     nebo
    0.37
     eller
    0.36
    Activations Density 0.054%

    No Known Activations