© 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 EvalsExportsAPI 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. 4-RES-JB
    11. 16649
    Prev
    Next
    INDEX
    Explanations

    expressions indicating emphasis or strong feelings, often related to disbelief or surprise

    oai_token-act-pair · gpt-3.5-turbo

    expressions of emphasis or seriousness

    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
    •layer 4 - surprise

    No Comments

    Negative Logits
    ifully
    -0.91
    tein
    -0.82
    atu
    -0.72
    orem
    -0.71
    anooga
    -0.70
    iful
    -0.69
    orius
    -0.68
    eer
    -0.67
    ramid
    -0.66
    winner
    -0.66
    POSITIVE LOGITS
    serious
    0.79
     entertained
    0.78
     considering
    0.77
     bothering
    0.73
     seriously
    0.73
    minent
    0.72
     enough
    0.71
    sted
    0.69
     consider
    0.68
    mond
    0.68
    Activations Density 0.013%

    No Known Activations