© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW 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. 15078
    Prev
    Next
    INDEX
    Explanations

    references to North Korea and related events/news

    oai_token-act-pair · gpt-3.5-turbo
    Triggered 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
    Neuron Alignment
    Index
    Value
    % of L₁
    346
    +0.12
    0.5%
    57
    +0.12
    0.5%
    496
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    57
    +0.12
    0.03
    67
    +0.12
    0.03
    346
    +0.11
    0.03
    Negative Logits
     "$:/
    -0.86
    TABLE
    -0.81
    NRS
    -0.75
    natureconservancy
    -0.70
    TED
    -0.70
    0200
    -0.68
    xxxxxxxx
    -0.66
    aminer
    -0.66
    PLIC
    -0.65
    reviewed
    -0.65
    POSITIVE LOGITS
     Korea
    1.34
     Koreans
    1.32
     Korean
    1.17
    uclear
    0.88
    ongyang
    0.88
     Vietnamese
    0.88
     Carolina
    0.85
    ampton
    0.85
    orea
    0.83
     Pole
    0.83
    Activations Density 0.022%

    No Known Activations