© 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. 3055
    Prev
    Next
    INDEX
    Explanations

    titles or roles within organizations, such as 'director' or 'president', especially with high activation values

    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
    373
    +0.13
    0.6%
    534
    +0.10
    0.5%
    121
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    104
    +0.13
    0.04
    204
    +0.10
    0.04
    264
    +0.10
    0.04
    Negative Logits
     surgeons
    -0.68
     idiots
    -0.67
     stret
    -0.66
     unbeliev
    -0.65
    eneg
    -0.62
    rums
    -0.62
    Attach
    -0.59
     whim
    -0.59
    uder
    -0.57
     lie
    -0.57
    POSITIVE LOGITS
     emer
    1.06
    itatively
    0.78
    director
    0.77
     of
    0.76
    veland
    0.69
     extraord
    0.69
     Ny
    0.67
    uably
    0.66
     director
    0.66
     curator
    0.65
    Activations Density 0.079%

    No Known Activations