© 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. 12774
    Prev
    Next
    INDEX
    Explanations

    individuals with specific professions or roles

    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
    575
    +0.12
    0.6%
    447
    +0.12
    0.6%
    291
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    575
    +0.12
    0.07
    291
    +0.12
    0.06
    247
    +0.10
    0.04
    Negative Logits
    izens
    -0.69
    ouls
    -0.68
    itars
    -0.66
     Lans
    -0.65
     Soldiers
    -0.63
     Hath
    -0.63
    laws
    -0.62
     assumptions
    -0.62
    ppings
    -0.62
     judgments
    -0.61
    POSITIVE LOGITS
    digy
    0.97
     extraord
    0.96
    ess
    0.90
     specializing
    0.87
     himself
    0.82
     herself
    0.79
     someday
    0.77
    alyst
    0.73
     unto
    0.73
     fundraiser
    0.73
    Activations Density 0.194%

    No Known Activations