© 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. 43265
    Prev
    Next
    INDEX
    Explanations

    phrases indicating criticism or lack of communication

    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
    pez
    -0.15
     îł
    -0.14
    ceptar
    -0.14
    UIT
    -0.14
    bew
    -0.14
    alogy
    -0.14
    iece
    -0.14
    EncodingException
    -0.14
    rello
    -0.13
    ponsor
    -0.13
    POSITIVE LOGITS
     consult
    0.42
     consulting
    0.40
     consulted
    0.39
     consultation
    0.39
    consult
    0.37
     Consult
    0.37
    Consult
    0.34
     Consulting
    0.32
     consultar
    0.30
     consulta
    0.30
    Activations Density 0.028%

    No Known Activations