© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Joseph Bloom · Feature Splitting for GPT2-Small
    5. GPT2-Small
    7. Residual Feature Splitting 49152
    9. 8-RES_FS49152-JB
    11. 37548
    Prev
    Next
    INDEX
    Explanations

    repeated mentions of the term "politics"

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs49152-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    49,152
    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
    288
    +0.16
    0.7%
    546
    +0.13
    0.6%
    679
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    546
    +0.16
    0.02
    288
    +0.13
    0.02
    188
    +0.11
    0.02
    Negative Logits
    actory
    -0.85
    imates
    -0.85
    uran
    -0.78
    amination
    -0.74
     Cancel
    -0.66
    âĢ¢âĢ¢âĢ¢âĢ¢
    -0.65
    Pont
    -0.65
    untary
    -0.64
    orage
    -0.64
    owship
    -0.63
    POSITIVE LOGITS
     correctness
    1.01
    hare
    0.87
    eering
    0.87
    clinton
    0.80
     intrig
    0.79
     Caucus
    0.77
    cape
    0.76
    atism
    0.75
     lawy
    0.74
    icians
    0.71
    Activations Density 0.013%

    No Known Activations