© 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 24576
    9. 8-RES_FS24576-JB
    11. 1754
    Prev
    Next
    INDEX
    Explanations

    conjunctions like 'and' used frequently in news articles

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs24576-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_24576
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    24,576
    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
    136
    +0.13
    0.6%
    228
    +0.11
    0.5%
    384
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    384
    +0.13
    0.04
    398
    +0.11
    0.04
    120
    +0.11
    0.04
    Negative Logits
    Lago
    -0.79
    :[
    -0.70
    *:
    -0.68
    ({
    -0.67
    BIL
    -0.66
    ©¶æ¥µ
    -0.65
    [[
    -0.65
    RGB
    -0.65
     Roof
    -0.65
    pps
    -0.65
    POSITIVE LOGITS
    rage
    0.77
     fury
    0.71
    ERY
    0.69
     ammunition
    0.69
     diplomacy
    0.67
    politics
    0.66
     commerce
    0.66
     diplomatic
    0.66
     medicine
    0.65
    science
    0.64
    Activations Density 0.121%

    No Known Activations