Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsBlogSlackPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Julius Han · Gated SAE for Llama3-8B-Instruct
    3. Llama3-8B-IT
    4. Residual Stream
    5. 25-RES-JH
    6. 0
    Prev
    Next
    INDEX
    Explanations

    references to information being available to the public

    oai_token-act-pair · gpt-4o-miniTriggered by @bot

    instances where information or content is made publicly available.

    oai_token-act-pair · gpt-4oTriggered by @faresfawzi
    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
    •Which is greater, 9.8 or 9.11?

    No Comments

    Negative Logits
    oll
    -0.20
    crow
    -0.15
    enez
    -0.15
    cio
    -0.14
    PublicKey
    -0.14
    inton
    -0.14
    ÙĪÙĬÙĥ
    -0.14
    çak
    -0.14
    ample
    -0.14
    ars
    -0.14
    POSITIVE LOGITS
    ally
    0.38
    ALLY
    0.28
     "-//
    0.26
    -private
    0.25
    -facing
    0.25
    /private
    0.23
    ized
    0.22
    /Private
    0.22
    aly
    0.21
    -domain
    0.20
    Activations Density 0.032%

    No Known Activations