© 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. 22406
    Prev
    Next
    INDEX
    Explanations

    conditional phrases or statements that involve "if" and variations of "to be"

    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
    373
    +0.28
    1.4%
    94
    +0.09
    0.4%
    245
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    373
    +0.28
    -0.02
    168
    +0.09
    0.04
    671
    +0.09
    0.03
    Negative Logits
    rift
    -0.65
    quartered
    -0.61
    furt
    -0.59
    rome
    -0.58
    engers
    -0.57
    PRES
    -0.57
    doors
    -0.56
    DOC
    -0.55
    placed
    -0.55
    fab
    -0.55
    POSITIVE LOGITS
     recap
    0.86
     disclaimer
    0.82
     disclaim
    0.78
     parting
    0.75
     paraph
    0.74
     congr
    0.73
     reiterate
    0.71
     spoiler
    0.70
     summarize
    0.68
    Sorry
    0.66
    Activations Density 0.215%

    No Known Activations