© 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. 24726
    Prev
    Next
    INDEX
    Explanations

    phrases indicating consideration or inclusion of factors in analysis

    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
    ened
    -0.17
    klä
    -0.17
     nackte
    -0.15
    iske
    -0.15
    ening
    -0.15
    ward
    -0.15
    èľ
    -0.15
     Webseite
    -0.14
    nor
    -0.14
    Ñħов
    -0.14
    POSITIVE LOGITS
     account
    0.66
    account
    0.45
     Account
    0.45
     consideration
    0.44
    Account
    0.39
     accounts
    0.37
    _account
    0.37
    -account
    0.36
     cuenta
    0.36
     accounted
    0.35
    Activations Density 0.012%

    No Known Activations