© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI 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. 8920
    Prev
    Next
    INDEX
    Explanations

    political and societal manipulation and control-related phrases

    oai_token-act-pair · gpt-3.5-turboTriggered 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
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    87
    +0.28
    1.5%
    295
    +0.08
    0.4%
    448
    +0.08
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.28
    0.09
    87
    +0.08
    -0.03
    698
    +0.08
    0.05
    Negative Logits
    ITNESS
    -0.75
    ifted
    -0.58
    VIDIA
    -0.57
     Canaver
    -0.56
    uana
    -0.55
    utz
    -0.55
     confir
    -0.54
     largeDownload
    -0.51
    omorphic
    -0.50
    ertodd
    -0.50
    POSITIVE LOGITS
     itch
    0.67
     onto
    0.65
    ebted
    0.64
     into
    0.63
    into
    0.63
     prematurely
    0.62
    til
    0.60
     alike
    0.57
     goodbye
    0.57
    onto
    0.57
    Activations Density 0.895%

    No Known Activations