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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports 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. 29079
    Prev
    Next
    INDEX
    Explanations

    negations and phrases indicating disagreement or rejection

    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
     be
    -0.23
     بتÙĪاÙĨ
    -0.18
    (be
    -0.18
    be
    -0.17
     Be
    -0.16
    .xmlbeans
    -0.15
    Be
    -0.15
    ature
    -0.15
    аÑĤа
    -0.15
    -be
    -0.15
    POSITIVE LOGITS
     exist
    0.34
     exists
    0.29
     Exist
    0.28
    exist
    0.24
     matter
    0.24
     existed
    0.23
     Exists
    0.22
     belong
    0.21
    Exist
    0.21
     existence
    0.21
    Activations Density 0.087%

    No Known Activations