© 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. 22328
    Prev
    Next
    INDEX
    Explanations

    words related to beauty or aesthetics

    oai_token-act-pair · gpt-4o-miniTriggered 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
    276
    +0.11
    0.5%
    406
    +0.09
    0.4%
    579
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    406
    +0.11
    0.04
    284
    +0.09
    0.03
    404
    +0.09
    0.03
    Negative Logits
    yip
    -1.01
    ooter
    -0.96
    icz
    -0.95
    oto
    -0.89
    uilt
    -0.87
    icago
    -0.84
    uling
    -0.84
    razil
    -0.83
    rew
    -0.83
    ricks
    -0.82
    POSITIVE LOGITS
     Ãł
    1.01
     de
    0.92
     du
    0.91
     des
    0.86
     cout
    0.84
     Lafayette
    0.83
     Franç
    0.83
     Royale
    0.82
     Ãī
    0.80
     é
    0.79
    Activations Density 0.127%

    No Known Activations