© 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. 10345
    Prev
    Next
    INDEX
    Explanations

    mentions of specific locations, particularly South Africa

    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
    496
    +0.14
    0.7%
    232
    +0.11
    0.5%
    238
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    496
    +0.14
    0.05
    232
    +0.11
    0.04
    238
    +0.11
    0.04
    Negative Logits
     "$:/
    -0.79
    inventoryQuantity
    -0.71
    TED
    -0.69
     captcha
    -0.65
    INESS
    -0.65
    ATURE
    -0.65
    ACA
    -0.65
    */(
    -0.63
    ovie
    -0.62
     juggling
    -0.62
    POSITIVE LOGITS
    wark
    1.48
    western
    1.31
     Carolina
    1.20
    ward
    1.10
    ampton
    1.08
     Dakota
    1.05
     Africa
    1.04
    side
    1.03
     Korea
    1.03
    wind
    1.01
    Activations Density 0.051%

    No Known Activations