© 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. 15605
    Prev
    Next
    INDEX
    Explanations

    photographs or image captions

    oai_token-act-pair · gpt-3.5-turboTriggered 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
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    373
    +0.18
    0.8%
    609
    +0.10
    0.4%
    564
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    609
    +0.18
    0.01
    334
    +0.10
    0.01
    319
    +0.09
    0.01
    Negative Logits
    stood
    -0.68
     retard
    -0.65
     nonviolent
    -0.62
    comed
    -0.59
     tant
    -0.59
    anus
    -0.58
     demol
    -0.57
     tense
    -0.57
     multipl
    -0.57
     submer
    -0.57
    POSITIVE LOGITS
     Photograph
    1.04
     Courtesy
    0.94
     ©
    0.88
     Thumbnails
    0.87
    archive
    0.87
     caption
    0.83
    ograph
    0.80
     Photo
    0.80
     IMAGES
    0.79
     courtesy
    0.79
    Activations Density 0.016%

    No Known Activations