© 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 EvalsExportsinterp-engineNEWAPI 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. 16346
    Prev
    Next
    INDEX
    Explanations

    words related to legal and policy matters in the context of political discourse

    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
    314
    +0.25
    1.2%
    680
    +0.11
    0.5%
    160
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    314
    +0.25
    0.02
    152
    +0.11
    0.05
    154
    +0.10
    0.04
    Negative Logits
     plur
    -0.70
     starters
    -0.69
     redress
    -0.67
     umb
    -0.67
     progress
    -0.66
     exper
    -0.66
     reproduction
    -0.65
     cir
    -0.65
     adulthood
    -0.65
     semifinals
    -0.65
    POSITIVE LOGITS
    inspired
    1.29
    induced
    1.28
    approved
    1.28
    era
    1.25
    appointed
    1.24
    owned
    1.24
    backed
    1.23
    sponsored
    1.22
    branded
    1.19
    themed
    1.19
    Activations Density 0.081%

    No Known Activations