© 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. 24568
    Prev
    Next
    INDEX
    Explanations

    references to unwanted or harmful events and their prevention

    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
    456
    +0.11
    0.5%
    233
    +0.10
    0.4%
    574
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    284
    +0.11
    0.05
    233
    +0.10
    0.04
    456
    +0.09
    0.04
    Negative Logits
     Cyrus
    -0.73
     Schumer
    -0.72
     Formation
    -0.71
     Ruler
    -0.69
     Tycoon
    -0.68
     Cohn
    -0.67
     Fiesta
    -0.66
    YS
    -0.65
     Oaks
    -0.64
     Reeves
    -0.64
    POSITIVE LOGITS
    rogens
    1.21
    rogen
    1.08
     unexplained
    0.93
     overlapping
    0.91
     unexpl
    0.90
     conflicting
    0.89
    nam
    0.89
     unrelated
    0.89
     unmarked
    0.88
     contradictory
    0.86
    Activations Density 0.122%

    No Known Activations