© 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. 3131
    Prev
    Next
    INDEX
    Explanations

    conditional statements where a specific action or outcome is contingent upon a prior condition

    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
    406
    +0.10
    0.5%
    590
    +0.10
    0.5%
    426
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    590
    +0.10
    0.02
    747
    +0.10
    0.02
    406
    +0.10
    0.02
    Negative Logits
    ãĥ´
    -0.74
    eries
    -0.74
    ery
    -0.73
    entry
    -0.68
    08
    -0.67
     vul
    -0.63
    eny
    -0.63
    ARC
    -0.63
    ulate
    -0.63
    iquette
    -0.63
    POSITIVE LOGITS
     otherwise
    1.07
     someone
    0.75
     somebody
    0.74
     you
    0.73
    soever
    0.72
    fy
    0.72
     expressly
    0.69
     corrective
    0.69
     explicitly
    0.68
    rr
    0.67
    Activations Density 0.023%

    No Known Activations