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

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports 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. 9750
    Prev
    Next
    INDEX
    Explanations

    time-related references, such as mentions of decades, years, months, and halves

    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
    223
    +0.12
    0.5%
    238
    +0.12
    0.5%
    255
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    651
    +0.12
    0.06
    657
    +0.12
    0.06
    238
    +0.10
    0.05
    Negative Logits
    ettings
    -0.66
     corridors
    -0.66
    ãĤ¢ãĥ«
    -0.62
    ashtra
    -0.62
    acted
    -0.61
    estine
    -0.60
    ports
    -0.60
    arters
    -0.59
    uctions
    -0.58
     flyers
    -0.57
    POSITIVE LOGITS
    long
    1.20
     ago
    0.99
    glass
    0.99
     apiece
    0.91
    ozy
    0.88
    gap
    0.85
     Ago
    0.83
    fuck
    0.78
    cake
    0.78
    nn
    0.77
    Activations Density 0.165%

    No Known Activations