© 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. 18028
    Prev
    Next
    INDEX
    Explanations

    references to the fairytale character Cinderella

    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
    481
    +0.18
    0.8%
    447
    +0.18
    0.8%
    724
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.18
    0.04
    326
    +0.18
    0.03
    481
    +0.11
    0.03
    Negative Logits
    iness
    -0.88
    lasses
    -0.87
    erate
    -0.86
    enegger
    -0.85
    leave
    -0.83
    ability
    -0.81
    orno
    -0.77
    glas
    -0.76
    tops
    -0.76
    riter
    -0.76
    POSITIVE LOGITS
     Reborn
    0.80
     Metatron
    0.78
     Centauri
    0.78
     Revelations
    0.76
     Unleashed
    0.75
     Rising
    0.72
     Rebirth
    0.72
     Dreams
    0.71
     incarn
    0.71
     Pictures
    0.70
    Activations Density 0.082%

    No Known Activations