© 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. 10846
    Prev
    Next
    INDEX
    Explanations

    mentions of intelligence agencies and investigations in a political context

    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
    288
    +0.17
    0.8%
    679
    +0.15
    0.7%
    616
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.17
    0.03
    328
    +0.15
    0.02
    679
    +0.11
    0.03
    Negative Logits
    nered
    -0.78
    pain
    -0.73
    itol
    -0.69
    anan
    -0.68
    à
    -0.68
    ikan
    -0.68
    zl
    -0.67
    Benz
    -0.65
    uberty
    -0.64
    dule
    -0.64
    POSITIVE LOGITS
     briefings
    1.11
    elligence
    0.91
     quot
    0.90
     analyst
    0.90
     gathering
    0.87
     agencies
    0.87
     Agency
    0.85
    agency
    0.83
     bulletin
    0.82
     briefing
    0.81
    Activations Density 0.061%

    No Known Activations