© 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 49152
    9. 8-RES_FS49152-JB
    11. 37232
    Prev
    Next
    INDEX
    Explanations

    phrases indicating regularity or frequency, particularly involving the word "every" followed by a number

    oai_token-act-pair · gpt-4o-miniTriggered by @bot
    New Auto-Interp
    Top Features by Cosine Similarity
    Comparing With GPT2-SMALL @ 8-res_fs49152-jb
    Configuration
    jbloom/GPT2-Small-Feature-Splitting-Experiment-Layer-8/blocks.8.hook_resid_pre_49152
    Prompts (Dashboard)
    24,576 prompts, 128 tokens each
    Dataset (Dashboard)
    Skylion007/openwebtext
    Features
    49,152
    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
    493
    +0.12
    0.5%
    616
    +0.12
    0.5%
    686
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    686
    +0.12
    0.02
    616
    +0.12
    0.02
    82
    +0.11
    0.02
    Negative Logits
    phabet
    -0.79
    çīĪ
    -0.68
     Viz
    -0.68
     anonymity
    -0.67
    dict
    -0.66
    ez
    -0.64
     Sakuya
    -0.64
     devils
    -0.62
     labels
    -0.61
    ãĥ¯
    -0.59
    POSITIVE LOGITS
     conceivable
    0.93
    THING
    0.92
    single
    0.85
     inch
    0.82
     imaginable
    0.81
    where
    0.75
     hour
    0.75
    ursday
    0.75
     month
    0.74
     single
    0.74
    Activations Density 0.019%

    No Known Activations