© 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. 657
    Prev
    Next
    INDEX
    Explanations

    information related to scientific studies and research

    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
    481
    +0.16
    0.7%
    393
    +0.11
    0.5%
    253
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    463
    +0.16
    0.05
    703
    +0.11
    0.04
    393
    +0.10
    0.04
    Negative Logits
    zb
    -0.79
    ieties
    -0.78
    ogy
    -0.77
    isse
    -0.73
    edia
    -0.72
    rued
    -0.72
    igg
    -0.70
    ied
    -0.69
    apy
    -0.68
    inct
    -0.68
    POSITIVE LOGITS
     formerly
    1.01
     adjunct
    1.01
     grandson
    0.95
     vice
    0.95
     scourge
    0.93
     longtime
    0.93
     honorary
    0.91
     outspoken
    0.90
     chairman
    0.87
     affiliated
    0.86
    Activations Density 0.242%

    No Known Activations