© 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. 25448
    Prev
    Next
    INDEX
    Explanations

    phrases indicating typical or habitual behaviors

    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
     Äijang
    -0.16
     recently
    -0.16
    iven
    -0.16
     actual
    -0.14
    continue
    -0.14
    rieg
    -0.14
    æŃ£åľ¨
    -0.14
     Recently
    -0.14
    errat
    -0.14
    ouri
    -0.14
    POSITIVE LOGITS
     wouldn
    0.27
     reserved
    0.26
     would
    0.24
     Reserved
    0.24
     reserve
    0.23
    Reserved
    0.23
     Reserve
    0.22
    reserved
    0.21
     Wouldn
    0.21
    Would
    0.20
    Activations Density 0.027%

    No Known Activations