© 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. 3859
    Prev
    Next
    INDEX
    Explanations

    phrases relating to personal beliefs or political discussions

    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
    579
    +0.13
    0.6%
    393
    +0.12
    0.5%
    572
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    0
    +0.13
    0.05
    393
    +0.12
    0.05
    703
    +0.09
    0.04
    Negative Logits
    ullivan
    -0.81
    ICES
    -0.76
    aults
    -0.75
    astical
    -0.75
    ulkan
    -0.74
    oward
    -0.72
    uta
    -0.72
    aido
    -0.72
    uto
    -0.71
    ventory
    -0.71
    POSITIVE LOGITS
     therefore
    1.46
     hence
    1.32
     thus
    1.19
     consequently
    1.15
     secondly
    1.05
     furthermore
    0.94
     moreover
    0.93
     deserves
    0.89
     foremost
    0.88
     vice
    0.87
    Activations Density 0.237%

    No Known Activations