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

    Neuronpedia

    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. 23988
    Prev
    Next
    INDEX
    Explanations

    official statements and related information in announcements

    oai_token-act-pair · gpt-4o-miniTriggered 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
    640
    +0.13
    0.6%
    314
    +0.12
    0.5%
    263
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    640
    +0.13
    0.09
    6
    +0.12
    0.07
    160
    +0.11
    0.05
    Negative Logits
     traps
    -0.86
     shallow
    -0.83
     dab
    -0.83
     messing
    -0.82
     buckets
    -0.81
     chopping
    -0.81
     churn
    -0.80
     unpredict
    -0.80
     setups
    -0.79
     cooking
    -0.78
    POSITIVE LOGITS
    Dear
    1.39
    Statement
    1.29
     hereby
    1.18
     GENERAL
    1.14
    Letter
    1.14
     UNITED
    1.11
    Effective
    1.11
    WASHINGTON
    1.10
    andum
    1.09
    Attorney
    1.08
    Activations Density 0.394%

    No Known Activations