© 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. 2856
    Prev
    Next
    INDEX
    Explanations

    phrases related to social justice and activism, specifically around empowering marginalized groups and addressing discrimination

    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
    330
    +0.11
    0.5%
    393
    +0.11
    0.5%
    583
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    583
    +0.11
    0.04
    330
    +0.11
    0.04
    327
    +0.11
    0.04
    Negative Logits
    ob
    -0.74
    bang
    -0.68
    opoly
    -0.66
    onis
    -0.66
    OB
    -0.63
    PB
    -0.63
     Serpent
    -0.62
    osate
    -0.62
     Resolution
    -0.61
    heit
    -0.61
    POSITIVE LOGITS
     wishing
    1.03
     who
    1.00
     pesky
    0.90
    who
    0.86
     affected
    0.85
     victimized
    0.84
     kinds
    0.83
     whom
    0.83
     interested
    0.83
     entrusted
    0.81
    Activations Density 0.062%

    No Known Activations