© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI Community BlogPrivacy & TermsContact
    1. Home
    3. Joseph Bloom · Feature Splitting for GPT2-Small
    5. GPT2-Small
    7. Residual Feature Splitting 49152
    9. 8-RES_FS49152-JB
    11. 45413
    Prev
    Next
    INDEX
    Explanations

    references to sources or attribution in text

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs49152-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    49,152
    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
    319
    +0.10
    0.5%
    352
    +0.10
    0.5%
    559
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    559
    +0.10
    0.01
    334
    +0.10
    0.01
    319
    +0.10
    0.01
    Negative Logits
    iqueness
    -0.93
    ãĤ¦ãĤ¹
    -0.79
     truths
    -0.73
     propositions
    -0.72
     sugars
    -0.71
    iencies
    -0.70
     diarr
    -0.69
    nosis
    -0.68
    heartedly
    -0.68
    resistant
    -0.67
    POSITIVE LOGITS
     Shutterstock
    1.49
     Wikimedia
    1.24
     Flickr
    1.19
     REUTERS
    1.01
     screenshot
    1.01
     YouTube
    0.97
     Courtesy
    0.96
     Youtube
    0.90
     Pinterest
    0.88
     Screen
    0.83
    Activations Density 0.016%

    No Known Activations