© 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. 9466
    Prev
    Next
    INDEX
    Explanations

    words related to government agencies, investigations, and law enforcement activities, particularly focusing on the Department of Homeland Security

    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
    +0.14
    0.7%
    422
    +0.12
    0.5%
    665
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.14
    0.02
    422
    +0.12
    0.02
    146
    +0.10
    0.02
    Negative Logits
    代
    -0.65
     toget
    -0.64
     gou
    -0.63
     BM
    -0.63
    ISTER
    -0.63
     Pall
    -0.62
    orem
    -0.62
    ãģ®é
    -0.61
    poons
    -0.60
    heet
    -0.60
    POSITIVE LOGITS
     Security
    1.19
    security
    1.15
     security
    0.98
    istan
    0.87
     Defense
    0.86
    elligence
    0.85
    gency
    0.84
     Homeland
    0.81
    Security
    0.81
     DHS
    0.77
    Activations Density 0.083%

    No Known Activations