© 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 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. 17954
    Prev
    Next
    INDEX
    Explanations

    exclamations of surprise or emphasis

    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
    255
    +0.12
    0.6%
    240
    +0.11
    0.5%
    604
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    572
    +0.12
    0.02
    604
    +0.11
    0.02
    345
    +0.09
    0.02
    Negative Logits
    perature
    -0.69
    -+-+
    -0.68
     Construct
    -0.61
    ciplinary
    -0.61
     Luxem
    -0.59
    inction
    -0.58
    ":[{"
    -0.57
    iership
    -0.57
     prevailing
    -0.57
    apers
    -0.56
    POSITIVE LOGITS
    hhhh
    1.03
    hhh
    0.95
     dear
    0.92
     yeah
    0.91
    anian
    0.90
    hh
    0.86
     yea
    0.84
    oho
    0.84
    yeah
    0.84
     hey
    0.82
    Activations Density 0.015%

    No Known Activations