© 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. 849
    Prev
    Next
    INDEX
    Explanations

    phrases related to civil unrest and protests

    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
    8
    +0.12
    0.6%
    742
    +0.12
    0.6%
    583
    +0.12
    0.6%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    8
    +0.12
    0.09
    737
    +0.12
    0.05
    131
    +0.12
    0.04
    Negative Logits
     attributed
    -0.80
     inappropriately
    -0.74
     somewhat
    -0.74
     incorrectly
    -0.74
     Originally
    -0.73
    ensical
    -0.72
     errone
    -0.72
     Presumably
    -0.72
     anecd
    -0.72
    imilar
    -0.72
    POSITIVE LOGITS
     FIGHT
    0.87
     ain
    0.84
    Ñĭ
    0.76
     destiny
    0.76
     YOU
    0.75
     YOUR
    0.75
     patri
    0.74
     DEM
    0.74
    venge
    0.74
     Promise
    0.74
    Activations Density 1.254%

    No Known Activations