© Neuronpedia 2026
    Privacy & TermsBlogGitHubSlackTwitterContact
    Neuronpedia logo - a computer chip with a rounded viewfinder border around it

    Neuronpedia

    Natural Language
    Autoencoders
    NEW    Assistant AxisNEWCircuit TracerUPDATESteerSAE EvalsExportsAPI 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. 18362
    Prev
    Next
    INDEX
    Explanations

    words related to governmental and political matters, including policies, commissions, and legal issues

    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
    167
    +0.12
    0.6%
    616
    +0.10
    0.5%
    134
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    616
    +0.12
    0.03
    134
    +0.10
    0.03
    705
    +0.10
    0.03
    Negative Logits
    edIn
    -0.78
     Seym
    -0.70
    Reloaded
    -0.64
     Lauder
    -0.63
     Frie
    -0.63
     unborn
    -0.61
     Anger
    -0.59
     Greenwald
    -0.59
     wiser
    -0.59
    theless
    -0.59
    POSITIVE LOGITS
    imposed
    1.40
    nova
    1.30
    visory
    1.28
    visor
    1.23
    charged
    1.17
    charg
    1.17
    powers
    1.16
    cedes
    1.11
    cell
    1.11
    computer
    1.09
    Activations Density 0.019%

    No Known Activations