© 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 24576
    9. 8-RES_FS24576-JB
    11. 5890
    Prev
    Next
    INDEX
    Explanations

    steps or processes in a specific order or sequence

    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
    685
    +0.13
    0.6%
    450
    +0.11
    0.5%
    174
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    450
    +0.13
    0.04
    554
    +0.11
    -0.00
    685
    +0.10
    0.02
    Negative Logits
     lately
    -0.86
    violence
    -0.73
     underestimated
    -0.69
    endon
    -0.67
    sect
    -0.66
     underestimate
    -0.66
     lacked
    -0.65
    danger
    -0.64
    bole
    -0.63
     disliked
    -0.63
    POSITIVE LOGITS
     whichever
    1.06
     optionally
    0.97
     prest
    0.95
     automatically
    0.92
     vo
    0.91
     outputs
    0.90
    Submit
    0.90
     Repeat
    0.90
     thereafter
    0.86
     then
    0.85
    Activations Density 0.600%

    No Known Activations