© 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. 4058
    Prev
    Next
    INDEX
    Explanations

    statements of positions and responses from governmental institutions or officials

    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
    629
    +0.10
    0.5%
    679
    +0.10
    0.5%
    256
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    6
    +0.10
    0.09
    145
    +0.10
    0.07
    679
    +0.10
    0.08
    Negative Logits
     lur
    -0.73
     dummy
    -0.73
     nifty
    -0.72
     cram
    -0.71
     clustered
    -0.70
     evapor
    -0.70
     quir
    -0.70
     hooked
    -0.70
     somewhere
    -0.69
     handy
    -0.69
    POSITIVE LOGITS
    Therefore
    1.26
     Additionally
    1.25
     Therefore
    1.24
     Accordingly
    1.22
     However
    1.20
     Furthermore
    1.20
    Furthermore
    1.14
    Additionally
    1.09
     Moreover
    1.08
     Please
    1.07
    Activations Density 0.518%

    No Known Activations