© 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 24576
    9. 8-RES_FS24576-JB
    11. 20204
    Prev
    Next
    INDEX
    Explanations

    locations or settings

    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
    16
    +0.10
    0.5%
    614
    +0.09
    0.4%
    118
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    88
    +0.10
    0.04
    123
    +0.09
    0.04
    16
    +0.09
    0.04
    Negative Logits
    DragonMagazine
    -0.74
    Query
    -0.72
    659
    -0.68
    612
    -0.67
     Cosponsors
    -0.67
    642
    -0.66
    263
    -0.61
    sem
    -0.61
    652
    -0.61
    382
    -0.61
    POSITIVE LOGITS
     existence
    1.15
     terms
    1.02
    efficiency
    0.97
     history
    0.95
    clusions
    0.94
     Europe
    0.91
     America
    0.90
    herent
    0.89
    ducers
    0.86
     antiquity
    0.85
    Activations Density 0.122%

    No Known Activations