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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    © Neuronpedia 2026
    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. 54659
    Prev
    Next
    INDEX
    Explanations

    expressions of effort and hard work

    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
    quist
    -0.14
    inne
    -0.14
    edom
    -0.14
    Sensitive
    -0.13
    ritz
    -0.13
    oya
    -0.13
    wan
    -0.13
    ê
    -0.13
     hacks
    -0.13
    ££
    -0.13
    POSITIVE LOGITS
     hard
    0.36
    hard
    0.29
     harder
    0.28
    -hard
    0.27
     HARD
    0.24
    _hard
    0.24
     Hard
    0.24
    Hard
    0.24
     hardest
    0.23
    硬
    0.22
    Activations Density 0.037%

    No Known Activations