© 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. 3056
    Prev
    Next
    INDEX
    Explanations

    punctuation

    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
    55
    +0.10
    0.5%
    766
    +0.10
    0.5%
    45
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    29
    +0.10
    0.04
    140
    +0.10
    0.04
    297
    +0.10
    0.04
    Negative Logits
    ertodd
    -0.73
     fiat
    -0.67
    ensical
    -0.64
    overe
    -0.64
    aeda
    -0.64
    itals
    -0.63
    heartedly
    -0.62
    gaming
    -0.62
    isy
    -0.62
    aples
    -0.61
    POSITIVE LOGITS
     wherein
    0.93
     starring
    0.93
     which
    0.92
     LLC
    0.89
     aka
    0.87
     featuring
    0.86
     Inc
    0.86
     Volume
    0.85
     Remastered
    0.84
    feat
    0.82
    Activations Density 0.175%

    No Known Activations