© 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 EvalsExportsinterp-engineNEWAPI 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. 4784
    Prev
    Next
    INDEX
    Explanations

    references to improving or working on codebase and software development tasks

    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
    167
    +0.14
    0.6%
    739
    +0.12
    0.6%
    615
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    211
    +0.14
    0.02
    352
    +0.12
    0.02
    137
    +0.10
    0.02
    Negative Logits
     Flavoring
    -0.97
     Wynne
    -0.80
    iscopal
    -0.77
    iatus
    -0.77
    abouts
    -0.72
    Ń·
    -0.71
     Bever
    -0.71
    iversal
    -0.70
    ESA
    -0.69
    esville
    -0.66
    POSITIVE LOGITS
     snippet
    1.36
    base
    1.23
     snippets
    1.21
     snipp
    0.96
     repository
    0.93
    code
    0.92
    velop
    0.89
     fragment
    0.87
     reuse
    0.85
     code
    0.85
    Activations Density 0.023%

    No Known Activations