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

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & 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. 23376
    Prev
    Next
    INDEX
    Explanations

    adjectives and descriptive terms related to size, treatment, and preparation

    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
    ages
    -0.15
    /place
    -0.15
    antly
    -0.14
    hton
    -0.14
    ish
    -0.14
    ised
    -0.14
    ìłģìľ¼ë¡ľ
    -0.14
    Wunused
    -0.14
    44
    -0.14
     äºĨ
    -0.14
    POSITIVE LOGITS
    ness
    0.30
    -up
    0.28
    -out
    0.26
    own
    0.23
    /un
    0.23
    -down
    0.23
    owns
    0.22
    -back
    0.22
    -Up
    0.21
    epend
    0.21
    Activations Density 0.333%

    No Known Activations