© 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. 4829
    Prev
    Next
    INDEX
    Explanations

    phrases related to entertainment or television show discussions

    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
    480
    +0.20
    0.9%
    314
    +0.13
    0.6%
    447
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    332
    +0.20
    0.03
    598
    +0.13
    0.01
    480
    +0.11
    -0.00
    Negative Logits
     vulner
    -0.68
    artifacts
    -0.67
     contrace
    -0.65
     Leilan
    -0.65
    */
    -0.64
    conservancy
    -0.62
    etheless
    -0.60
    partial
    -0.60
     fruitful
    -0.59
     manpower
    -0.59
    POSITIVE LOGITS
     quotes
    0.94
    yrics
    0.89
     quote
    0.88
     referring
    0.86
     quotation
    0.86
     phrases
    0.84
     Quote
    0.83
    Said
    0.80
     exclaimed
    0.80
     exclaim
    0.80
    Activations Density 0.191%

    No Known Activations