© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it
    Neuronpedia
    Jacobian LensNEW
    Natural Language
    Autoencoders
    MAY 26    Assistant AxisJAN 26Circuit TracerAPR 26SteerSAE EvalsExportsinterp-engineNEW 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. 14906
    Prev
    Next
    INDEX
    Explanations

    actions or events related to criminal activity

    oai_token-act-pair · gpt-3.5-turbo
    Triggered 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
    Neuron Alignment
    Index
    Value
    % of L₁
    373
    +0.11
    0.5%
    379
    +0.11
    0.5%
    92
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    70
    +0.11
    0.04
    761
    +0.11
    0.04
    665
    +0.10
    0.04
    Negative Logits
    DragonMagazine
    -0.75
    inger
    -0.69
    Reporting
    -0.67
    respond
    -0.65
    ilogy
    -0.65
    fortunately
    -0.65
    erm
    -0.64
     thanking
    -0.63
    endif
    -0.63
    Acknowled
    -0.63
    POSITIVE LOGITS
     improper
    0.87
    BuyableInstoreAndOnline
    0.83
     illegal
    0.82
     underage
    0.75
     illegally
    0.73
     insensitive
    0.70
     irregularities
    0.69
     unsafe
    0.69
     endangered
    0.69
     insufficient
    0.69
    Activations Density 0.188%

    No Known Activations