© 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. 13017
    Prev
    Next
    INDEX
    Explanations

    political figures and dynamics, particularly related to Senate leadership

    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
    147
    +0.11
    0.5%
    556
    +0.11
    0.5%
    288
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    147
    +0.11
    0.02
    556
    +0.11
    0.02
    136
    +0.10
    0.02
    Negative Logits
    metic
    -0.76
    uckland
    -0.76
    worldly
    -0.75
    tered
    -0.74
    iceps
    -0.70
    tsy
    -0.69
    autical
    -0.68
    icular
    -0.68
    kers
    -0.67
    neys
    -0.67
    POSITIVE LOGITS
     Leader
    1.17
     Whip
    1.15
     Majority
    1.08
     Secretary
    0.86
     Caucus
    0.85
     Speaker
    0.85
     Minority
    0.84
    Leader
    0.83
     Floor
    0.82
     Clerk
    0.82
    Activations Density 0.013%

    No Known Activations