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

    Neuronpedia

    APICircuit TracerNEWSteerSAE EvalsExportsSlackBlogPrivacy & TermsContact
    © Neuronpedia 2025
    Privacy & TermsBlogGitHubSlackTwitterContact
    1. Home
    2. Joseph Bloom · Open Source Sparse Autoencoders for all Residual Stream Layers of GPT2-Small
    3. GPT2-Small
    4. Residual Stream
    5. 4-RES-JB
    6. 19324
    Prev
    Next
    INDEX
    Explanations

    violence-related actions involving physical contact such as hitting, stabbing, or pushing

    oai_token-act-pair · gpt-3.5-turbo

    instances of violent physical actions or assaults involving objects

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 4-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.hook_resid_pre
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    24,576
    Data Type
    torch.float32
    Hook Point
    blocks.4.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    4
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    sylvania
    -0.72
    omsky
    -0.72
    ajor
    -0.71
    ieri
    -0.67
    achusetts
    -0.66
    obook
    -0.64
    anty
    -0.62
    mart
    -0.62
     Reconstruction
    -0.61
    sburg
    -0.61
    POSITIVE LOGITS
    stood
    1.26
     regard
    1.21
     regards
    1.15
    standing
    1.14
     impunity
    1.08
    draw
    1.03
    holding
    0.94
     respect
    0.94
    drawn
    0.91
     dignity
    0.91
    Activations Density 0.226%

    No Known Activations