© 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 · Feature Splitting for GPT2-Small
    5. GPT2-Small
    7. Residual Feature Splitting 24576
    9. 8-RES_FS24576-JB
    11. 19986
    Prev
    Next
    INDEX
    Explanations

    mentions of political events or details about people

    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
    481
    +0.21
    1.0%
    447
    +0.13
    0.6%
    133
    +0.12
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    409
    +0.21
    0.03
    133
    +0.13
    0.04
    278
    +0.12
    0.03
    Negative Logits
    ilk
    -0.83
     Brist
    -0.79
     Arn
    -0.73
    Wil
    -0.70
    ibly
    -0.69
     Ort
    -0.68
     Slater
    -0.67
     zinc
    -0.67
    Morgan
    -0.67
     kW
    -0.67
    POSITIVE LOGITS
     event
    1.01
     tournament
    0.99
     Finals
    0.99
     finale
    0.97
     festival
    0.94
     outing
    0.94
    iversary
    0.93
     marathon
    0.90
     conference
    0.90
    athon
    0.89
    Activations Density 0.317%

    No Known Activations