© 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. 13707
    Prev
    Next
    INDEX
    Explanations

    phrases related to specific places in news articles

    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
    613
    +0.11
    0.5%
    517
    +0.10
    0.5%
    704
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    232
    +0.11
    0.05
    659
    +0.10
    0.04
    409
    +0.10
    0.04
    Negative Logits
     Spike
    -0.81
     acron
    -0.73
     Naz
    -0.73
     Stud
    -0.71
     Sim
    -0.70
     Lad
    -0.70
     Brawl
    -0.70
     DX
    -0.69
     Tsu
    -0.69
    bows
    -0.69
    POSITIVE LOGITS
    ILL
    0.97
    ill
    0.93
    illa
    0.88
    illon
    0.83
    illed
    0.82
    illus
    0.81
    rite
    0.79
    illing
    0.79
    lly
    0.78
     Grimes
    0.78
    Activations Density 0.282%

    No Known Activations