© 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. 4712
    Prev
    Next
    INDEX
    Explanations

    positive assessments or performances

    oai_token-act-pair · gpt-3.5-turboTriggered 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
    87
    +0.15
    0.7%
    181
    +0.08
    0.4%
    557
    +0.08
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    608
    +0.15
    0.08
    322
    +0.08
    0.05
    557
    +0.08
    0.04
    Negative Logits
    iru
    -0.67
    inki
    -0.62
    ))))
    -0.61
    lication
    -0.58
    imming
    -0.55
    "},{"
    -0.55
    oyer
    -0.54
    article
    -0.54
     Intermediate
    -0.53
     Tycoon
    -0.53
    POSITIVE LOGITS
     lately
    1.50
     since
    1.25
    since
    1.06
     recent
    0.97
     numerous
    0.88
     recently
    0.87
     countless
    0.84
     steadily
    0.77
     hitherto
    0.74
    Since
    0.74
    Activations Density 0.979%

    No Known Activations