© 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. 22795
    Prev
    Next
    INDEX
    Explanations

    phrases that contain the conjunction "and" emphasizing connections or relationships

    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
    718
    +0.12
    0.5%
    756
    +0.11
    0.5%
    416
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    0
    +0.12
    0.05
    746
    +0.11
    0.03
    484
    +0.10
    0.04
    Negative Logits
    DragonMagazine
    -0.82
    anmar
    -0.72
    FN
    -0.67
    ļéĨĴ
    -0.66
    BIL
    -0.65
    lance
    -0.65
    mma
    -0.63
     Pigs
    -0.62
     Leilan
    -0.61
    ho
    -0.61
    POSITIVE LOGITS
     thereby
    1.01
     thus
    0.98
     preferably
    0.96
     therefore
    0.93
     deserve
    0.93
     consequently
    0.92
     hopefully
    0.90
     effortlessly
    0.88
     efficiently
    0.88
     then
    0.87
    Activations Density 0.186%

    No Known Activations