© 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. 3-RES-JB
    11. 16582
    Prev
    Next
    INDEX
    Explanations

    references to a sports ball, specifically instances where the ball is being handled or moved

    oai_token-act-pair · gpt-3.5-turbo

    references to "the ball" in a sports context

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 3-res-jb
    Configuration
    jbloom/GPT2-Small-SAEs-Reformatted/blocks.3.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.3.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    3
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Negative Logits
    Rated
    -0.69
     Norn
    -0.64
     plur
    -0.62
     prevail
    -0.62
    ä
    -0.62
    ths
    -0.61
    rians
    -0.59
    å°Ĩ
    -0.57
     Borders
    -0.57
     Procedures
    -0.57
    POSITIVE LOGITS
    ball
    0.98
    istics
    0.92
    park
    0.90
    oons
    0.90
    handler
    0.87
    inic
    0.85
    ghazi
    0.84
    ozo
    0.81
    istically
    0.81
    haus
    0.81
    Activations Density 0.011%

    No Known Activations