© 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. 13859
    Prev
    Next
    INDEX
    Explanations

    phrases related to financial transactions and business dealings

    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
    358
    +0.10
    0.5%
    656
    +0.10
    0.5%
    264
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    358
    +0.10
    0.03
    677
    +0.10
    0.03
    31
    +0.10
    0.03
    Negative Logits
    icone
    -0.83
    progress
    -0.76
    eh
    -0.73
     differe
    -0.71
    jriwal
    -0.69
    ogens
    -0.68
    mith
    -0.67
    gaard
    -0.65
    swer
    -0.65
    dayName
    -0.65
    POSITIVE LOGITS
     assets
    0.89
     Rupert
    0.83
     farmland
    0.76
     bankrupt
    0.74
     property
    0.71
     Chrysler
    0.71
     bonds
    0.71
     holdings
    0.70
     shareholders
    0.69
     subsidiaries
    0.68
    Activations Density 0.103%

    No Known Activations