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

    Neuronpedia

    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 12288
    9. 8-RES_FS12288-JB
    11. 10622
    Prev
    Next
    INDEX
    Explanations

    descriptions related to fashion items and do-it-yourself (DIY) projects

    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
    +0.29
    1.4%
    481
    +0.17
    0.8%
    548
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    326
    +0.29
    0.09
    288
    +0.17
    0.10
    678
    +0.11
    0.07
    Negative Logits
    bably
    -0.76
    76561
    -0.75
    judicial
    -0.73
    putable
    -0.72
    arthy
    -0.70
    NetMessage
    -0.69
     intervening
    -0.68
    eded
    -0.68
     redacted
    -0.67
     waning
    -0.67
    POSITIVE LOGITS
     kit
    0.93
     DIY
    0.90
    washer
    0.86
     tutorial
    0.82
     planner
    0.81
     kits
    0.79
     Recipe
    0.79
     recipe
    0.78
     calculator
    0.76
     decor
    0.76
    Activations Density 0.839%

    No Known Activations