© 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 49152
    9. 8-RES_FS49152-JB
    11. 25976
    Prev
    Next
    INDEX
    Explanations

    references to companies or organizations, particularly those with "Inc" in their names

    oai_token-act-pair · gpt-4o-mini
    Triggered 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
    Neuron Alignment
    Index
    Value
    % of L₁
    180
    +0.11
    0.5%
    627
    +0.10
    0.5%
    735
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    180
    +0.11
    0.02
    88
    +0.10
    0.01
    251
    +0.09
    0.01
    Negative Logits
    buster
    -0.76
     differe
    -0.68
     doub
    -0.67
     bottleneck
    -0.65
     armies
    -0.65
     cue
    -0.62
     mercy
    -0.62
     Emin
    -0.61
     policemen
    -0.57
     Halls
    -0.57
    POSITIVE LOGITS
    orporated
    2.02
    arcer
    1.36
    orpor
    1.29
    endiary
    1.22
    idents
    1.22
    .,
    1.17
    redibly
    1.15
    orp
    1.10
    orrect
    1.09
    ./
    1.01
    Activations Density 0.008%

    No Known Activations