© 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. 16205
    Prev
    Next
    INDEX
    Explanations

    words related to redistricting and gerrymandering

    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
    481
    +0.18
    0.8%
    137
    +0.12
    0.6%
    218
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.18
    0.01
    137
    +0.12
    0.01
    409
    +0.11
    0.01
    Negative Logits
     Chiefs
    -0.67
    ournal
    -0.66
     LIFE
    -0.65
    Gi
    -0.65
    theless
    -0.62
    ournals
    -0.60
     Tonight
    -0.60
     Spit
    -0.60
     DRAG
    -0.60
     witnesses
    -0.59
    POSITIVE LOGITS
    ricting
    1.30
    ribut
    1.26
    ricted
    1.05
    abled
    0.93
    etrical
    0.93
    hee
    0.92
    etric
    0.89
     redist
    0.87
     redistributed
    0.86
    artisan
    0.85
    Activations Density 0.021%

    No Known Activations