© 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 24576
    9. 8-RES_FS24576-JB
    11. 18893
    Prev
    Next
    INDEX
    Explanations

    European countries and some specific countries or regions

    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
    462
    +0.11
    0.5%
    241
    +0.10
    0.5%
    659
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    461
    +0.11
    0.04
    462
    +0.10
    0.04
    241
    +0.10
    0.04
    Negative Logits
     lik
    -0.60
     meanings
    -0.56
     qualities
    -0.56
    Redditor
    -0.56
     behavi
    -0.55
    blem
    -0.55
    ACTION
    -0.55
    Reviewer
    -0.55
    76561
    -0.55
    20439
    -0.53
    POSITIVE LOGITS
    ania
    0.78
     Indies
    0.74
    thia
    0.72
     Arabia
    0.72
     etc
    0.72
    anches
    0.69
    ,
    0.69
     Territories
    0.69
     coasts
    0.68
     Philippines
    0.67
    Activations Density 0.105%

    No Known Activations