© 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 12288
    9. 8-RES_FS12288-JB
    11. 9475
    Prev
    Next
    INDEX
    Explanations

    resolutions or decisions made in a formal context to address specific issues or topics

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs12288-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    12,288
    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
    274
    +0.12
    0.6%
    195
    +0.10
    0.4%
    744
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    274
    +0.12
    0.02
    288
    +0.10
    0.02
    744
    +0.09
    0.02
    Negative Logits
    inen
    -0.79
    elong
    -0.77
    spring
    -0.74
    rium
    -0.73
    FactoryReloaded
    -0.73
    endon
    -0.71
    raltar
    -0.70
    aquin
    -0.69
    heon
    -0.69
    anooga
    -0.68
    POSITIVE LOGITS
     resolution
    0.93
     Resolution
    0.92
     resolutions
    0.91
     authorizing
    0.83
    ignty
    0.78
     resolves
    0.77
    olution
    0.75
    ancing
    0.75
    mary
    0.71
    witz
    0.71
    Activations Density 0.083%

    No Known Activations