© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI 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. 18010
    Prev
    Next
    INDEX
    Explanations

    explicit mentions of goals or objectives

    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
    247
    +0.14
    0.7%
    16
    +0.11
    0.5%
    679
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    247
    +0.14
    0.04
    16
    +0.11
    0.03
    355
    +0.10
    0.03
    Negative Logits
    oute
    -0.76
    ols
    -0.71
    CV
    -0.68
    olson
    -0.64
    Edited
    -0.62
    windows
    -0.61
     Solitaire
    -0.61
    uddin
    -0.60
    greg
    -0.60
    avi
    -0.60
    POSITIVE LOGITS
    posts
    1.07
     Goal
    1.02
    Goal
    0.96
    goal
    0.93
     goal
    0.87
     goals
    0.83
     accomplished
    0.81
     scorer
    0.79
    setting
    0.78
    lessly
    0.76
    Activations Density 0.050%

    No Known Activations