© 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. 3634
    Prev
    Next
    INDEX
    Explanations

    keywords related to text formatting and non-linguistic elements

    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
    336
    +0.11
    0.5%
    481
    +0.11
    0.5%
    517
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    517
    +0.11
    0.04
    336
    +0.11
    0.03
    114
    +0.09
    0.04
    Negative Logits
    clus
    -0.73
    ibles
    -0.66
     chronological
    -0.66
    racuse
    -0.59
    Ļ½
    -0.57
    CLOSE
    -0.57
    ãĥ´ãĤ¡
    -0.54
     administrator
    -0.53
    ãĥ¥
    -0.53
     artif
    -0.53
    POSITIVE LOGITS
    âķ
    0.80
    *.
    0.79
    */
    0.74
    thumbnails
    0.70
    requires
    0.69
    }.
    0.68
    ------------------------------------------------
    0.68
    .............
    0.66
    },
    0.65
    into
    0.64
    Activations Density 0.114%

    No Known Activations