© 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. 3182
    Prev
    Next
    INDEX
    Explanations

    important phrases or details within news articles, potentially related to policies, investigations, crime, or current events

    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
    +0.11
    0.5%
    307
    +0.10
    0.5%
    61
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    729
    +0.11
    0.01
    61
    +0.10
    0.01
    412
    +0.10
    0.01
    Negative Logits
    wcs
    -0.89
    interstitial
    -0.82
    士
    -0.76
    paralle
    -0.74
    artifacts
    -0.73
    phase
    -0.72
    itol
    -0.72
    cffffcc
    -0.72
    decl
    -0.71
    orum
    -0.70
    POSITIVE LOGITS
     Shutterstock
    1.03
     commenter
    1.03
     Huffington
    1.00
     Erik
    0.93
     Inquis
    0.91
     Gawker
    0.89
     Geek
    0.89
     Dave
    0.86
     Heidi
    0.85
     @
    0.85
    Activations Density 0.035%

    No Known Activations