© 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. 15557
    Prev
    Next
    INDEX
    Explanations

    phrases emphasizing the importance or lack of importance of certain actions or situations

    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
    420
    +0.10
    0.5%
    439
    +0.10
    0.5%
    507
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    750
    +0.10
    0.02
    242
    +0.10
    0.02
    271
    +0.10
    0.02
    Negative Logits
    ocally
    -0.76
    visor
    -0.75
    ccording
    -0.73
    yrinth
    -0.70
    ria
    -0.70
    ©¶æ
    -0.69
    urat
    -0.68
    Compan
    -0.68
    odan
    -0.67
    apons
    -0.66
    POSITIVE LOGITS
     anymore
    1.29
     whatsoever
    1.01
     whether
    0.84
     if
    0.82
    usercontent
    0.76
     anytime
    0.74
     anyway
    0.73
     anyways
    0.73
     any
    0.73
     either
    0.71
    Activations Density 0.053%

    No Known Activations