© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Jacobian LensNEW
    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI 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. 86
    Prev
    Next
    INDEX
    Explanations

    technical details and specifications related to electronic devices

    oai_token-act-pair · gpt-3.5-turboTriggered 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
    Not in Any Lists

    No Comments

    Neuron Alignment
    Index
    Value
    % of L₁
    288
    +0.14
    0.6%
    620
    +0.11
    0.5%
    612
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    288
    +0.14
    0.04
    326
    +0.11
    0.03
    289
    +0.10
    0.01
    Negative Logits
     Homeless
    -0.78
     Shelter
    -0.76
     Yelp
    -0.70
     Ethnic
    -0.68
     Orioles
    -0.68
    fur
    -0.67
     volunt
    -0.67
     Cecil
    -0.67
    pees
    -0.65
     Velvet
    -0.65
    POSITIVE LOGITS
     powering
    1.26
     chipset
    1.25
     CPUs
    1.19
     processors
    1.18
    onductor
    1.18
     processor
    1.15
     architectures
    1.10
     GPUs
    1.08
     chip
    1.07
     cores
    1.04
    Activations Density 0.086%

    No Known Activations