© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    1. Home
    3. Joseph Bloom · Feature Splitting for GPT2-Small
    5. GPT2-Small
    7. Residual Feature Splitting 768
    9. 8-RES_FS768-JB
    11. 663
    Prev
    Next
    INDEX
    Explanations

    words related to the military, military actions, and military personnel

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs768-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_768
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    768
    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
    +0.15
    0.7%
    152
    +0.13
    0.6%
    85
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.15
    0.23
    152
    +0.13
    0.15
    109
    +0.11
    0.09
    Negative Logits
    âĶģ
    -0.87
     Rowling
    -0.81
     Seller
    -0.79
    Lot
    -0.78
    Publisher
    -0.74
    apest
    -0.73
    à
    -0.71
    lihood
    -0.70
    flush
    -0.69
    Consumer
    -0.69
    POSITIVE LOGITS
     stationed
    1.37
     fatig
    1.30
     barracks
    1.28
     colonel
    1.23
     Corps
    1.22
     garrison
    1.18
     commanders
    1.16
     commander
    1.16
    utenant
    1.10
     patrols
    1.09
    Activations Density 5.245%

    No Known Activations