© 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 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. 14076
    Prev
    Next
    INDEX
    Explanations

    conjunctions and terms related to conditional reasoning or contrasts

    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
     drew
    -0.19
    came
    -0.18
     gave
    -0.18
     withdrew
    -0.17
    Generated
    -0.17
    woke
    -0.17
     deduct
    -0.17
     grew
    -0.17
     wrench
    -0.16
     conven
    -0.16
    POSITIVE LOGITS
     kept
    0.21
     taken
    0.20
     done
    0.20
     worsh
    0.17
     worden
    0.17
     thrown
    0.17
    ãģķãĤĮ
    0.17
    angkan
    0.16
     put
    0.16
     used
    0.16
    Activations Density 0.026%

    No Known Activations