© 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. 9671
    Prev
    Next
    INDEX
    Explanations

    mentions of prospective opportunities or potential investments

    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
    575
    +0.14
    0.7%
    311
    +0.12
    0.6%
    506
    +0.11
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    575
    +0.14
    0.03
    506
    +0.12
    0.02
    271
    +0.11
    0.02
    Negative Logits
    cation
    -0.80
    ipper
    -0.72
    capacity
    -0.69
    dos
    -0.69
    cas
    -0.68
    aration
    -0.68
    tein
    -0.67
    ped
    -0.67
    ç
    -0.67
    DCS
    -0.66
    POSITIVE LOGITS
     prospects
    1.02
     prospect
    0.93
     futures
    0.89
     Scouting
    0.88
    ures
    0.80
    ively
    0.78
    ors
    0.77
     Prospect
    0.71
     horizon
    0.71
    urity
    0.70
    Activations Density 0.088%

    No Known Activations