© 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. 1424
    Prev
    Next
    INDEX
    Explanations

    descriptions of physical processes or phenomena

    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
    481
    +0.20
    0.9%
    373
    +0.12
    0.6%
    82
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    362
    +0.20
    0.08
    608
    +0.12
    0.07
    703
    +0.10
    0.04
    Negative Logits
    ardless
    -0.79
    blems
    -0.71
    bernatorial
    -0.69
    ogether
    -0.68
    ctuary
    -0.68
    tackle
    -0.67
    allo
    -0.66
    uge
    -0.65
    verage
    -0.65
     Coverage
    -0.65
    POSITIVE LOGITS
     predetermined
    0.79
     imagination
    0.76
     oneself
    0.71
     subconscious
    0.70
     varying
    0.68
     humans
    0.68
     arcane
    0.66
     invention
    0.66
     periphery
    0.66
     peripheral
    0.65
    Activations Density 0.790%

    No Known Activations