© 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 · Feature Splitting for GPT2-Small
    5. GPT2-Small
    7. Residual Feature Splitting 24576
    9. 8-RES_FS24576-JB
    11. 22644
    Prev
    Next
    INDEX
    Explanations

    references to internal processes or investigations within organizations

    oai_token-act-pair · gpt-4o-miniTriggered 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
    288
    +0.13
    0.6%
    285
    +0.11
    0.5%
    455
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    613
    +0.13
    0.02
    285
    +0.11
    0.02
    332
    +0.11
    0.02
    Negative Logits
    gger
    -0.81
    apo
    -0.79
    oÄŁ
    -0.78
    some
    -0.78
    ammy
    -0.77
    nir
    -0.75
    ky
    -0.74
    eful
    -0.72
    vich
    -0.72
    alla
    -0.71
    POSITIVE LOGITS
     combustion
    1.41
     workings
    1.16
    ized
    1.15
    ization
    1.04
    izing
    0.96
    ised
    0.95
     organs
    0.92
    izations
    0.91
     affairs
    0.90
    izable
    0.86
    Activations Density 0.011%

    No Known Activations