© 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. 14961
    Prev
    Next
    INDEX
    Explanations

    web links and email addresses

    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
    640
    +0.15
    0.7%
    764
    +0.09
    0.4%
    513
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    284
    +0.15
    0.04
    107
    +0.09
    0.03
    515
    +0.09
    0.03
    Negative Logits
     conclud
    -0.81
     FN
    -0.73
     basis
    -0.72
     grounding
    -0.71
     refres
    -0.67
     ADA
    -0.66
     whims
    -0.66
     ambush
    -0.66
     withd
    -0.64
     aggregation
    -0.64
    POSITIVE LOGITS
    edu
    1.47
    exe
    1.25
    github
    1.21
    dll
    1.17
    jp
    1.14
    blogspot
    1.12
    wikipedia
    1.11
    png
    1.10
    jpg
    1.10
    net
    1.07
    Activations Density 0.065%

    No Known Activations