© 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. 24545
    Prev
    Next
    INDEX
    Explanations

    titles and topics related to diverse subjects, particularly categories of articles and content types

    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
    53
    +0.12
    0.6%
    371
    +0.11
    0.5%
    398
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    398
    +0.12
    0.02
    411
    +0.11
    0.01
    53
    +0.10
    0.01
    Negative Logits
     later
    -0.68
     depended
    -0.67
     subsequ
    -0.66
    IELD
    -0.65
    uers
    -0.65
    rine
    -0.65
    Ts
    -0.64
    oche
    -0.63
    ORS
    -0.62
    DERR
    -0.62
    POSITIVE LOGITS
     POLIT
    0.82
     Discrimination
    0.79
     Miscellaneous
    0.77
     Unc
    0.72
     Diseases
    0.72
    education
    0.71
    crime
    0.70
     Antiqu
    0.69
     Architecture
    0.69
     Tags
    0.68
    Activations Density 0.019%

    No Known Activations