© 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. 18938
    Prev
    Next
    INDEX
    Explanations

    the name "Ko" followed by a number

    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
    659
    +0.15
    0.7%
    48
    +0.12
    0.6%
    343
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    659
    +0.15
    0.02
    343
    +0.12
    0.01
    232
    +0.10
    0.01
    Negative Logits
    senal
    -0.80
     Creed
    -0.77
    glass
    -0.74
    ب
    -0.73
     narrator
    -0.73
     à¨
    -0.73
    Ö¼
    -0.71
    天
    -0.69
    IBLE
    -0.68
    ingham
    -0.66
    POSITIVE LOGITS
    zzi
    1.18
    osta
    1.10
    zy
    0.98
    essler
    0.96
    jo
    0.96
    etter
    0.96
    unin
    0.95
    eln
    0.95
    ppa
    0.95
    pps
    0.94
    Activations Density 0.015%

    No Known Activations