© 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. 15044
    Prev
    Next
    INDEX
    Explanations

    technical terms related to university studies, aviation equipment, cloud computing services, artistic endeavors, transgender issues, home automation, and stock market analysis

    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
    481
    +0.18
    0.8%
    447
    +0.17
    0.8%
    450
    +0.12
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.18
    0.06
    163
    +0.17
    0.04
    67
    +0.12
    0.04
    Negative Logits
    ĸļ
    -1.00
     Archdemon
    -0.71
    ILA
    -0.71
    thora
    -0.68
     Beir
    -0.67
     TAMADRA
    -0.67
    xit
    -0.66
     Heidi
    -0.64
     Heard
    -0.64
    senal
    -0.64
    POSITIVE LOGITS
     societies
    0.96
     observers
    0.95
     wat
    0.89
     analysts
    0.86
     organizations
    0.86
     users
    0.85
     enthusiasts
    0.84
     scholars
    0.84
     households
    0.84
     respondents
    0.83
    Activations Density 0.281%

    No Known Activations