© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    1. Home
    3. Joseph Bloom · Open Source Sparse Autoencoders for all Residual Stream Layers of GPT2-Small
    5. GPT2-Small
    7. Residual Stream
    9. 0-RES-JB
    11. 6595
    Prev
    Next
    INDEX
    Explanations

    phrases related to titles or names using the term "Duke"

    oai_token-act-pair · gpt-3.5-turbo

    mentions of royalty, specifically the terms "Duke" and "Duchess."

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 0-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.0.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.0.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    0
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     tal
    -0.80
    Ñģ
    -0.74
    }{
    -0.72
     acron
    -0.72
    xml
    -0.71
    pers
    -0.68
     techn
    -0.67
    Ñĥ
    -0.67
    ital
    -0.66
    ~~~~
    -0.66
    POSITIVE LOGITS
     Duke
    3.97
     Duchess
    1.93
     Marqu
    1.70
     Durham
    1.55
     Earl
    1.48
     Vanderbilt
    1.39
     Prince
    1.34
    uke
    1.28
     Baron
    1.27
     Clemson
    1.23
    Activations Density 0.020%

    No Known Activations