© 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. 16018
    Prev
    Next
    INDEX
    Explanations

    proper nouns and specific names related to global geographical locations and individuals, especially in a geopolitical context

    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
    447
    +0.21
    1.0%
    481
    +0.21
    1.0%
    65
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    752
    +0.21
    0.08
    57
    +0.21
    0.06
    342
    +0.10
    0.06
    Negative Logits
    edin
    -0.88
    Ó
    -0.81
    xxx
    -0.77
    same
    -0.75
    vable
    -0.75
    lished
    -0.74
    imeo
    -0.73
    arkin
    -0.73
    urable
    -0.73
    pection
    -0.72
    POSITIVE LOGITS
     officials
    1.08
     commentators
    1.06
     diplomats
    1.06
     historians
    1.04
     scholars
    1.02
     negotiators
    1.02
     politicians
    1.00
     policymakers
    0.99
     observers
    0.99
     authorities
    0.96
    Activations Density 0.237%

    No Known Activations