© 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. 23858
    Prev
    Next
    INDEX
    Explanations

    phrases that indicate a sense of belonging or inclusion

    oai_token-act-pair · gpt-4o-miniTriggered 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
    387
    +0.12
    0.6%
    430
    +0.12
    0.5%
    481
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    312
    +0.12
    0.02
    348
    +0.12
    0.02
    596
    +0.11
    0.02
    Negative Logits
    agra
    -0.74
    arag
    -0.69
    pload
    -0.68
    elle
    -0.66
    SB
    -0.65
    raltar
    -0.65
    deep
    -0.62
    ja
    -0.62
    ham
    -0.61
    Cart
    -0.61
    POSITIVE LOGITS
     nowadays
    0.71
     unchanged
    0.69
     behaved
    0.68
     boils
    0.67
     harmless
    0.67
     consisted
    0.67
     intact
    0.65
     consists
    0.65
     respects
    0.63
     adhere
    0.63
    Activations Density 0.063%

    No Known Activations