© 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. 23956
    Prev
    Next
    INDEX
    Explanations

    instances of the word "part" and its variations, particularly in the context of separation or departure

    oai_token-act-pair · gpt-4o-miniTriggered 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
    127
    +0.10
    0.5%
    571
    +0.10
    0.5%
    355
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    571
    +0.10
    0.01
    304
    +0.10
    0.01
    147
    +0.09
    0.01
    Negative Logits
    trak
    -0.78
    arya
    -0.78
    multi
    -0.73
    ggle
    -0.69
    creation
    -0.68
    eq
    -0.67
    ensitivity
    -0.66
    ency
    -0.66
    reys
    -0.66
    cycles
    -0.66
    POSITIVE LOGITS
     parting
    0.82
     parted
    0.81
    onymous
    0.78
    onym
    0.77
    uary
    0.73
    ophone
    0.70
     sep
    0.69
    nered
    0.67
    emn
    0.64
    oming
    0.63
    Activations Density 0.022%

    No Known Activations