© 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. 9402
    Prev
    Next
    INDEX
    Explanations

    names of individuals, particularly with Arabic or Eastern European origin

    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
    564
    +0.12
    0.5%
    298
    +0.11
    0.5%
    430
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    299
    +0.12
    0.02
    683
    +0.11
    0.02
    564
    +0.10
    0.02
    Negative Logits
     Goose
    -0.73
    bats
    -0.70
     gum
    -0.63
     err
    -0.60
     cough
    -0.60
     Turtle
    -0.59
     fetch
    -0.58
     Hallow
    -0.58
     Zot
    -0.57
     fault
    -0.56
    POSITIVE LOGITS
    ibo
    0.96
    iaz
    0.90
    orio
    0.82
    agus
    0.79
    chenko
    0.76
    ippi
    0.76
    eni
    0.75
    amph
    0.75
    imus
    0.74
    ream
    0.74
    Activations Density 0.054%

    No Known Activations