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. 23196
    Prev
    Next
    INDEX
    Explanations

    phrases related to the concept of fitting or compatibility

    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
    xba
    -0.15
    hua
    -0.14
    aeper
    -0.14
    edom
    -0.14
    innen
    -0.13
    że
    -0.13
    undi
    -0.13
    bai
    -0.13
    енка
    -0.13
    à¸ĵ
    -0.13
    POSITIVE LOGITS
     into
    0.28
     Into
    0.22
    Into
    0.21
     together
    0.21
     snug
    0.21
     onto
    0.21
    into
    0.20
     perfectly
    0.19
     INTO
    0.19
    _into
    0.18
    Activations Density 0.041%

    No Known Activations