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

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI 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. 11-RES-JB
    11. 6190
    Prev
    Next
    INDEX
    Explanations

    names of athletes

    oai_token-act-pair · gpt-3.5-turbo

    nouns and proper names associated with individuals or players in sports

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 11-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.11.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.11.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    11
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
     Wendy
    -0.73
     Feminist
    -0.70
     Regions
    -0.70
     Rowling
    -0.66
     psychologists
    -0.66
     Governments
    -0.66
     Gloria
    -0.66
     psychiatrists
    -0.65
     publishers
    -0.64
     capitalists
    -0.63
    POSITIVE LOGITS
     Jr
    1.20
     (.
    0.98
     III
    0.98
    iard
    0.94
     sidelined
    0.87
    sson
    0.86
    kamp
    0.85
     offensively
    0.84
    bach
    0.84
     Sr
    0.83
    Activations Density 0.176%

    No Known Activations