© 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. 5152
    Prev
    Next
    INDEX
    Explanations

    statements involving future intentions or aspirations

    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
    241
    +0.12
    0.5%
    583
    +0.11
    0.5%
    465
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    465
    +0.12
    0.03
    241
    +0.11
    0.03
    687
    +0.10
    0.03
    Negative Logits
    cies
    -0.88
    lete
    -0.86
    acas
    -0.82
    ament
    -0.77
    ombat
    -0.77
    iya
    -0.76
    iak
    -0.75
    ieve
    -0.74
    ocaust
    -0.73
    ken
    -0.73
    POSITIVE LOGITS
     someday
    1.23
     even
    0.90
     SOME
    0.80
     subconscious
    0.75
     kidding
    0.74
     sooner
    0.70
     EVEN
    0.69
     thats
    0.68
     unsurprisingly
    0.67
     tempted
    0.66
    Activations Density 0.033%

    No Known Activations