© 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. 9452
    Prev
    Next
    INDEX
    Explanations

    words related to technology specifications, such as power supply and wattage

    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
    523
    +0.10
    0.5%
    522
    +0.09
    0.4%
    755
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    78
    +0.10
    0.03
    523
    +0.09
    0.03
    259
    +0.09
    0.03
    Negative Logits
     Von
    -0.77
    roit
    -0.69
    eryl
    -0.68
     Bei
    -0.68
     Niet
    -0.66
    eret
    -0.65
    rine
    -0.65
    ally
    -0.64
    OOL
    -0.62
    romeda
    -0.62
    POSITIVE LOGITS
    houses
    1.17
    plant
    0.98
    train
    0.97
    lifting
    0.97
    stroke
    0.95
     outage
    0.93
    ups
    0.85
    station
    0.85
    puff
    0.84
    grid
    0.83
    Activations Density 0.066%

    No Known Activations