© 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. 22721
    Prev
    Next
    INDEX
    Explanations

    terms related to arbitrary decisions or actions

    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
    635
    +0.12
    0.6%
    516
    +0.11
    0.5%
    237
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    516
    +0.12
    0.01
    151
    +0.11
    0.01
    288
    +0.10
    0.01
    Negative Logits
    iosis
    -0.95
    ector
    -0.82
    ership
    -0.82
    iquette
    -0.81
    icans
    -0.81
    din
    -0.79
    igi
    -0.78
    tha
    -0.77
    nen
    -0.77
    ilitating
    -0.76
    POSITIVE LOGITS
     confinement
    0.85
     whims
    0.85
     recomp
    0.76
     detain
    0.75
     arbitrary
    0.73
     detention
    0.72
     det
    0.71
     gratification
    0.69
     whim
    0.69
     guiActiveUn
    0.68
    Activations Density 0.033%

    No Known Activations