© 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. 15360
    Prev
    Next
    INDEX
    Explanations

    mentions of different countries and their governmental, economic, or social issues

    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
    373
    +0.19
    0.9%
    430
    +0.14
    0.7%
    679
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    430
    +0.19
    0.05
    6
    +0.14
    0.05
    144
    +0.11
    0.04
    Negative Logits
     maiden
    -0.79
     hiber
    -0.75
     asteroids
    -0.67
     pit
    -0.66
     impulse
    -0.66
     worms
    -0.65
     sucker
    -0.63
     hunted
    -0.63
     unlucky
    -0.63
     tunes
    -0.62
    POSITIVE LOGITS
     REUTERS
    1.32
     Picture
    1.06
     Courtesy
    1.06
     Photo
    1.04
     Photograph
    0.97
    Photo
    0.92
     PHOTO
    0.91
    jpg
    0.91
     Photographer
    0.91
     Images
    0.90
    Activations Density 0.136%

    No Known Activations