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

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsinterp-engineNEWAPI 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. 24376
    Prev
    Next
    INDEX
    Explanations

    specific terminology related to regulatory standards and classifications

    oai_token-act-pair · gpt-4o-miniTriggered 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
    176
    +0.11
    0.5%
    658
    +0.11
    0.5%
    766
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    575
    +0.11
    0.06
    288
    +0.11
    0.07
    547
    +0.09
    0.03
    Negative Logits
     urge
    -0.66
     interviewed
    -0.61
    \":
    -0.60
    rawdownloadcloneembedreportprint
    -0.60
    athi
    -0.54
     hangs
    -0.54
    rave
    -0.53
    advert
    -0.52
     posters
    -0.52
     itch
    -0.52
    POSITIVE LOGITS
     ASAP
    0.85
    livion
    0.84
     instead
    0.83
     altogether
    0.77
     equivalents
    0.72
    EVA
    0.71
     afterwards
    0.69
     whence
    0.69
    instead
    0.69
     afterward
    0.68
    Activations Density 0.436%

    No Known Activations