© 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. 14995
    Prev
    Next
    INDEX
    Explanations

    terms related to political and economic contexts, such as governments, debts, and ratings

    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
    527
    +0.10
    0.5%
    325
    +0.10
    0.5%
    401
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.10
    0.09
    401
    +0.10
    0.07
    325
    +0.10
    0.06
    Negative Logits
    Published
    -0.70
     Belfast
    -0.59
    \",
    -0.58
     Published
    -0.58
     Gw
    -0.57
     Shutterstock
    -0.57
     Latest
    -0.57
     Posted
    -0.57
     Meng
    -0.56
    ,,,,,,,,
    -0.55
    POSITIVE LOGITS
     becomes
    0.93
     remains
    0.91
     disappears
    0.90
     hasn
    0.86
     tends
    0.85
     cannot
    0.84
     stays
    0.84
     wouldn
    0.83
     would
    0.82
     survives
    0.82
    Activations Density 0.923%

    No Known Activations