© 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. 7896
    Prev
    Next
    INDEX
    Explanations

    references to famous personalities and entities such as politicians, celebrities, and sports figures in news articles

    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
    481
    +0.15
    0.7%
    302
    +0.10
    0.5%
    373
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    204
    +0.15
    0.04
    571
    +0.10
    0.04
    91
    +0.10
    0.04
    Negative Logits
    mbuds
    -0.74
    ipedia
    -0.73
    noon
    -0.71
    fo
    -0.64
     Helpful
    -0.63
    veyard
    -0.61
     Crimean
    -0.60
    english
    -0.60
    Duration
    -0.60
     Females
    -0.58
    POSITIVE LOGITS
     opted
    0.90
     underwent
    0.87
     joked
    0.85
     survived
    0.84
     admits
    0.83
     tweeted
    0.82
     insists
    0.80
     penned
    0.79
     endured
    0.78
     wore
    0.78
    Activations Density 0.168%

    No Known Activations