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. 23307
    Prev
    Next
    INDEX
    Explanations

    descriptions of cruelty or harshness in various contexts

    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
    laus
    -0.15
    chten
    -0.14
    .scalablytyped
    -0.14
    -issue
    -0.14
    .deepcopy
    -0.14
    gon
    -0.13
    adem
    -0.13
     pute
    -0.13
    eated
    -0.13
    ãĤħ
    -0.13
    POSITIVE LOGITS
    lest
    0.31
    ly
    0.26
     treatment
    0.23
     Treatment
    0.22
     towards
    0.22
     fate
    0.21
     toward
    0.21
    er
    0.20
    -hearted
    0.20
    Treatment
    0.19
    Activations Density 0.018%

    No Known Activations