© 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 49152
    9. 8-RES_FS49152-JB
    11. 45472
    Prev
    Next
    INDEX
    Explanations

    references to digital information and data representation

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs49152-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    49,152
    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
    30
    +0.13
    0.6%
    420
    +0.13
    0.6%
    678
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    30
    +0.13
    0.01
    678
    +0.13
    0.01
    326
    +0.11
    0.01
    Negative Logits
     Physicians
    -0.73
     Sharks
    -0.65
    ammad
    -0.65
    ordan
    -0.61
    WAY
    -0.61
     finance
    -0.61
     Patron
    -0.61
     Submission
    -0.60
     Trop
    -0.59
     causation
    -0.59
    POSITIVE LOGITS
    hift
    0.99
    cake
    0.99
    meal
    0.98
    wana
    0.95
    terness
    0.93
    wagon
    0.92
    buck
    0.90
    hered
    0.86
    umen
    0.86
    bits
    0.82
    Activations Density 0.009%

    No Known Activations