© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    APIAssistant AxisNEWCircuit TracerNEWSteerSAE EvalsExports Community BlogPrivacy & TermsContact
    1. Home
    3. Joseph Bloom · Feature Splitting for GPT2-Small
    5. GPT2-Small
    7. Residual Feature Splitting 12288
    9. 8-RES_FS12288-JB
    11. 845
    Prev
    Next
    INDEX
    Explanations

    individuals' ages and their roles or positions, such as high school seniors or registered nurses

    oai_token-act-pair · gpt-3.5-turboTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs12288-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_12288
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    12,288
    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
    752
    +0.11
    0.5%
    314
    +0.10
    0.5%
    373
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    752
    +0.11
    0.07
    711
    +0.10
    0.04
    152
    +0.10
    0.06
    Negative Logits
     iPads
    -0.87
    ravings
    -0.81
     Scenes
    -0.79
    zees
    -0.78
     Puzzles
    -0.78
     interfaces
    -0.77
    rooms
    -0.75
    isms
    -0.75
     Beasts
    -0.72
     VIDEOS
    -0.72
    POSITIVE LOGITS
     member
    1.07
     lawyer
    1.00
     accountant
    1.00
     graduate
    1.00
     supporter
    0.97
     biologist
    0.94
     educator
    0.94
     attorney
    0.94
     researcher
    0.94
     consultant
    0.94
    Activations Density 0.389%

    No Known Activations