© 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. 20326
    Prev
    Next
    INDEX
    Explanations

    words related to time periods, particularly "then" which signifies a specific point in the past

    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
    293
    +0.11
    0.5%
    451
    +0.10
    0.5%
    542
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    488
    +0.11
    0.03
    141
    +0.10
    0.02
    677
    +0.10
    0.03
    Negative Logits
    imir
    -0.79
    apixel
    -0.74
    Sing
    -0.69
    arij
    -0.66
    uras
    -0.64
    artifacts
    -0.64
    ibaba
    -0.63
    bys
    -0.61
    ARI
    -0.61
    abis
    -0.60
    POSITIVE LOGITS
     2025
    0.75
     exhaustion
    0.72
     standards
    0.70
    rox
    0.68
     2030
    0.67
     arriving
    0.64
    claw
    0.64
     2050
    0.63
    reau
    0.62
    stocks
    0.62
    Activations Density 0.070%

    No Known Activations