© 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. 15380
    Prev
    Next
    INDEX
    Explanations

    sentences with a critical or analytical tone towards political and social issues

    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
    66
    +0.12
    0.6%
    230
    +0.12
    0.5%
    393
    +0.12
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    66
    +0.12
    0.04
    284
    +0.12
    0.04
    456
    +0.12
    0.04
    Negative Logits
    ULTS
    -0.82
    ESE
    -0.82
    ICES
    -0.81
    YS
    -0.78
     Lines
    -0.74
    arks
    -0.74
     Jackets
    -0.72
    verts
    -0.72
     Walls
    -0.71
    ynthesis
    -0.70
    POSITIVE LOGITS
    rogen
    1.09
     efficient
    1.01
     unpre
    1.01
     economical
    0.98
     unpredictable
    0.97
     imaginative
    0.96
     powerful
    0.95
     lucrative
    0.95
     eclectic
    0.95
     sophisticated
    0.94
    Activations Density 0.100%

    No Known Activations