© 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 · Feature Splitting for GPT2-Small
    5. GPT2-Small
    7. Residual Feature Splitting 24576
    9. 8-RES_FS24576-JB
    11. 6438
    Prev
    Next
    INDEX
    Explanations

    the words "Donald" and "and" together

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs24576-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    24,576
    Data Type
    torch.float32
    Hook Point
    blocks.8.hook_resid_pre
    Architecture
    standard
    Context Size
    128
    Dataset
    Skylion007/openwebtext
    Hook Point Layer
    8
    Activation Function
    relu
    Embeds
    IFrame
    Link
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    430
    +0.11
    0.5%
    543
    +0.11
    0.5%
    659
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    543
    +0.11
    0.04
    659
    +0.11
    0.03
    265
    +0.10
    0.03
    Negative Logits
    wcs
    -0.81
    uggest
    -0.77
    ancial
    -0.75
    artifacts
    -0.74
    mining
    -0.73
    different
    -0.71
    minist
    -0.70
    system
    -0.69
    pmwiki
    -0.68
    actory
    -0.67
    POSITIVE LOGITS
     Sons
    1.05
     Associates
    0.92
     Mary
    0.91
     Morty
    0.90
    ERSON
    0.87
     Juliet
    0.86
     Catherine
    0.86
     Buster
    0.86
     Anna
    0.86
     Donna
    0.86
    Activations Density 0.090%

    No Known Activations