© 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 49152
    9. 8-RES_FS49152-JB
    11. 25946
    Prev
    Next
    INDEX
    Explanations

    uppercase letters, possibly indicating acronyms or important names

    oai_token-act-pair · gpt-4o-mini
    Triggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs49152-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    49,152
    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
    373
    +0.16
    0.8%
    599
    +0.10
    0.5%
    314
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    659
    +0.16
    0.04
    314
    +0.10
    0.01
    67
    +0.09
    0.03
    Negative Logits
     Resistance
    -0.65
     Limits
    -0.58
    iencies
    -0.56
     Saving
    -0.55
     Revis
    -0.55
     Rampage
    -0.54
    problem
    -0.54
    alties
    -0.54
     ______
    -0.54
     Save
    -0.54
    POSITIVE LOGITS
    +)
    1.02
    utterstock
    0.95
    )(
    0.93
    )/
    0.90
    )
    0.90
    )</
    0.90
    ENN
    0.88
    NS
    0.87
    )--
    0.84
    NW
    0.83
    Activations Density 0.045%

    No Known Activations