© 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 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. 41630
    Prev
    Next
    INDEX
    Explanations

    instances of systemic issues or failures within government or societal structures

    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
    +0.25
    1.2%
    481
    +0.22
    1.0%
    679
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    679
    +0.25
    0.07
    561
    +0.22
    0.05
    393
    +0.11
    0.06
    Negative Logits
    ãĥij
    -0.79
    ãĤ¨ãĥ«
    -0.75
    dies
    -0.73
    iris
    -0.73
    peed
    -0.72
    Ö
    -0.70
    rette
    -0.69
    orb
    -0.68
     mates
    -0.68
    worm
    -0.68
    POSITIVE LOGITS
     empirical
    0.99
     successive
    0.98
     neoliberal
    0.97
     scandals
    0.95
     policymakers
    0.94
     scholars
    0.93
     scholarly
    0.93
     majorities
    0.92
     methodological
    0.91
     historians
    0.91
    Activations Density 0.433%

    No Known Activations