© 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. 22859
    Prev
    Next
    INDEX
    Explanations

    names of academic or research institutions and centers

    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
    145
    +0.11
    0.5%
    386
    +0.10
    0.4%
    288
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    146
    +0.11
    0.04
    247
    +0.10
    0.04
    267
    +0.10
    0.04
    Negative Logits
    inventoryQuantity
    -0.70
     embodiments
    -0.68
    NetMessage
    -0.68
    dated
    -0.67
     ages
    -0.67
     moms
    -0.63
     pleas
    -0.62
    theless
    -0.62
     wages
    -0.62
    ardless
    -0.60
    POSITIVE LOGITS
     Prize
    1.06
     Collection
    0.95
     Consortium
    0.93
     School
    0.92
     Cooperative
    0.88
     Scholarship
    0.86
     Partners
    0.84
     Laboratory
    0.83
     Publishers
    0.83
     Laboratories
    0.83
    Activations Density 0.181%

    No Known Activations