© 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 12288
    9. 8-RES_FS12288-JB
    11. 502
    Prev
    Next
    INDEX
    Explanations

    phrases related to personal experiences, interactions, and emotional responses

    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
    447
    +0.20
    0.9%
    636
    +0.10
    0.5%
    10
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    49
    +0.20
    0.08
    499
    +0.10
    0.05
    55
    +0.10
    0.08
    Negative Logits
    selves
    -0.87
     selves
    -0.83
     millenn
    -0.77
    results
    -0.68
    Higher
    -0.67
    OTAL
    -0.64
     collective
    -0.64
     mutual
    -0.63
     oppos
    -0.59
    yg
    -0.59
    POSITIVE LOGITS
     himself
    1.19
     Himself
    1.03
     assassinated
    0.93
     veto
    0.80
     his
    0.76
     herself
    0.74
    hedon
    0.67
    panic
    0.66
    hirt
    0.65
     sidelined
    0.63
    Activations Density 1.010%

    No Known Activations