© 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. 2728
    Prev
    Next
    INDEX
    Explanations

    statements indicating certain information is known or well-known

    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
    213
    +0.14
    0.7%
    383
    +0.11
    0.5%
    725
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    213
    +0.14
    0.03
    383
    +0.11
    0.03
    725
    +0.10
    0.03
    Negative Logits
    otion
    -0.81
    erva
    -0.81
    rentice
    -0.80
    aminer
    -0.75
    reau
    -0.73
    boa
    -0.73
    ovie
    -0.73
    onde
    -0.72
    anco
    -0.71
    otos
    -0.69
    POSITIVE LOGITS
    lege
    0.86
    л
    0.86
    cut
    0.84
    ledged
    0.82
    saf
    0.73
     beforehand
    0.72
     aliases
    0.70
    sie
    0.69
     falsehood
    0.68
    hed
    0.68
    Activations Density 0.030%

    No Known Activations