© 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. 14943
    Prev
    Next
    INDEX
    Explanations

    numbers or numeric words

    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
    447
    +0.20
    1.0%
    477
    +0.10
    0.5%
    679
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    679
    +0.20
    0.06
    477
    +0.10
    0.04
    724
    +0.10
    0.05
    Negative Logits
    racuse
    -0.72
    rency
    -0.63
    rick
    -0.63
    Delivery
    -0.62
     Courier
    -0.61
     Blaster
    -0.60
    ctor
    -0.56
    Dub
    -0.55
     Conversation
    -0.54
     srf
    -0.54
    POSITIVE LOGITS
    hips
    1.12
     belong
    1.00
    mith
    0.96
     are
    0.94
    hip
    0.93
    paces
    0.89
     aren
    0.89
    hops
    0.89
     exist
    0.88
    folk
    0.88
    Activations Density 0.225%

    No Known Activations