© 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. 43089
    Prev
    Next
    INDEX
    Explanations

    phrases indicating uncertainty or lack of clarity in definitions and claims

    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
     conclusive
    -0.15
    ONA
    -0.14
    -toggler
    -0.14
    urdy
    -0.14
     correct
    -0.14
     truths
    -0.13
    ÏĨα
    -0.13
     smooth
    -0.13
    _ASSUME
    -0.12
    ahr
    -0.12
    POSITIVE LOGITS
     specific
    0.39
     specificity
    0.39
    åħ·ä½ĵ
    0.38
     Specific
    0.37
     concrete
    0.36
     specify
    0.35
     specifies
    0.35
     specifics
    0.35
    Specific
    0.34
     specified
    0.34
    Activations Density 0.036%

    No Known Activations