© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    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. 22163
    Prev
    Next
    INDEX
    Explanations

    occurrences of words related to legal or formal processes, particularly in the context of documentation and compliance

    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
    100
    +0.13
    0.6%
    393
    +0.11
    0.5%
    31
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    43
    +0.13
    0.06
    100
    +0.11
    0.07
    546
    +0.10
    0.07
    Negative Logits
    Enlarge
    -0.71
    ctuary
    -0.63
    orest
    -0.62
    veland
    -0.61
    eki
    -0.61
    hedral
    -0.61
    iny
    -0.61
    lier
    -0.61
    zn
    -0.60
    zzy
    -0.60
    POSITIVE LOGITS
     respectively
    1.27
     including
    1.25
     irrespective
    1.21
     pursuant
    1.15
     regardless
    1.14
     INCLUD
    1.13
     namely
    1.12
     etc
    1.06
     excluding
    1.06
     thereby
    1.05
    Activations Density 0.320%

    No Known Activations