© 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. 10861
    Prev
    Next
    INDEX
    Explanations

    relevant information or facts in a text

    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
    763
    +0.10
    0.5%
    27
    +0.10
    0.5%
    616
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    27
    +0.10
    0.04
    362
    +0.10
    0.04
    616
    +0.10
    0.04
    Negative Logits
    ibaba
    -0.92
    abi
    -0.76
    etsk
    -0.69
    ovsky
    -0.69
    ashtra
    -0.62
    agascar
    -0.61
     propagate
    -0.60
    burgh
    -0.59
    ario
    -0.58
    iatrics
    -0.57
    POSITIVE LOGITS
     margins
    0.93
     virtue
    0.93
     standards
    0.84
     means
    0.83
     Means
    0.82
    committee
    0.79
     Numbers
    0.77
     margin
    0.76
    products
    0.72
     sheer
    0.70
    Activations Density 0.293%

    No Known Activations