© 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. 45692
    Prev
    Next
    INDEX
    Explanations

    mentions of Reddit and its associated activities or discussions

    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
    447
    +0.14
    0.7%
    496
    +0.12
    0.6%
    520
    +0.12
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    520
    +0.14
    0.02
    346
    +0.12
    0.01
    683
    +0.12
    0.01
    Negative Logits
     fidelity
    -0.65
     blinded
    -0.63
    xon
    -0.63
     accur
    -0.61
     leukemia
    -0.60
     Bethlehem
    -0.59
     warranty
    -0.59
    ³³³³³³³³³³³³³³³³
    -0.59
     dozen
    -0.59
     Lauder
    -0.59
    POSITIVE LOGITS
    reddits
    0.99
    Reddit
    0.96
     Username
    0.95
    ors
    0.94
    username
    0.89
     admins
    0.82
     AMA
    0.82
    pmwiki
    0.82
    icum
    0.79
     Reddit
    0.79
    Activations Density 0.007%

    No Known Activations