© 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. 28161
    Prev
    Next
    INDEX
    Explanations

    adverbs that indicate degrees of certainty or emphasis

    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
     are
    -0.20
     blown
    -0.20
     пода
    -0.17
     eaten
    -0.15
     aren
    -0.15
     bitten
    -0.15
     beaten
    -0.15
    _tF
    -0.15
     resemble
    -0.14
     withdrawn
    -0.14
    POSITIVE LOGITS
     took
    0.25
     went
    0.23
     gave
    0.23
     did
    0.22
    took
    0.21
     came
    0.21
     became
    0.21
     saw
    0.20
     fell
    0.18
     threw
    0.18
    Activations Density 0.438%

    No Known Activations