© 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. 739
    Prev
    Next
    INDEX
    Explanations

    phrases related to government policies and regulations

    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
    480
    +0.46
    2.3%
    203
    +0.11
    0.6%
    86
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    665
    +0.46
    0.10
    480
    +0.11
    0.00
    641
    +0.10
    0.08
    Negative Logits
    Downloadha
    -0.69
    ighed
    -0.66
    quartered
    -0.60
     appeared
    -0.60
     reportedly
    -0.60
     nicknamed
    -0.59
     boasted
    -0.59
    typically
    -0.58
     famously
    -0.57
    oof
    -0.57
    POSITIVE LOGITS
     ourselves
    1.17
     tomorrow
    1.03
     responsibly
    1.02
     yourselves
    0.94
     someday
    0.81
     ASAP
    0.81
     our
    0.74
     JUSTICE
    0.72
    cipled
    0.71
     wherever
    0.71
    Activations Density 1.420%

    No Known Activations