© 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. 9673
    Prev
    Next
    INDEX
    Explanations

    phrases involving emphasis or intensification, especially with the word "hell."

    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
    674
    +0.11
    0.5%
    627
    +0.10
    0.5%
    284
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    284
    +0.11
    0.02
    596
    +0.10
    0.02
    212
    +0.10
    0.02
    Negative Logits
    ãĤ¢ãĥ«
    -0.74
    idas
    -0.73
    Dialog
    -0.72
    eters
    -0.69
    xit
    -0.63
    odox
    -0.63
    CONCLUS
    -0.62
    actionDate
    -0.60
     Guid
    -0.60
    xtap
    -0.59
    POSITIVE LOGITS
    shit
    0.76
    orld
    0.74
    buck
    0.74
    ucks
    0.68
     fuckin
    0.67
    atin
    0.67
     darn
    0.65
     lot
    0.65
     mole
    0.64
     Dynamo
    0.64
    Activations Density 0.107%

    No Known Activations