© 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. 15892
    Prev
    Next
    INDEX
    Explanations

    words related to changes, modifications, or impacts

    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
    92
    +0.10
    0.5%
    206
    +0.10
    0.5%
    151
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    151
    +0.10
    0.04
    307
    +0.10
    0.03
    45
    +0.10
    0.02
    Negative Logits
    ifully
    -0.87
    afety
    -0.82
    tein
    -0.79
    sburgh
    -0.78
    elsen
    -0.74
    tale
    -0.71
    icip
    -0.70
    atu
    -0.67
     Memories
    -0.67
    ventus
    -0.67
    POSITIVE LOGITS
     impacted
    0.96
     effected
    0.94
     reduces
    0.92
     differ
    0.91
     impacting
    0.91
     improves
    0.89
     reduce
    0.89
     altered
    0.88
     outper
    0.87
     reduced
    0.87
    Activations Density 0.043%

    No Known Activations