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. 23339
    Prev
    Next
    INDEX
    Explanations

    instances of punctuation, particularly commas, indicating pauses or separations in thoughts

    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
    ...\
    -0.15
     ifndef
    -0.14
    bersome
    -0.14
    éĺħ读次æķ°
    -0.13
    èo
    -0.13
    -chan
    -0.13
    ymoon
    -0.13
    ldkf
    -0.13
    	ROM
    -0.13
    nam
    -0.13
    POSITIVE LOGITS
     but
    0.24
     
    0.21
     which
    0.19
     it
    0.17
    000
    0.16
     he
    0.16
     we
    0.16
     and
    0.15
     as
    0.15
    if
    0.14
    Activations Density 0.256%

    No Known Activations