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

    Neuronpedia

    Jacobian LensNEW
    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. 9038
    Prev
    Next
    INDEX
    Explanations

    phrases related to legal and constitutional terms, as well as political statements

    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
    373
    +0.11
    0.5%
    214
    +0.10
    0.4%
    100
    +0.10
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    100
    +0.11
    0.03
    239
    +0.10
    0.02
    567
    +0.10
    0.02
    Negative Logits
    elight
    -0.87
    ettel
    -0.72
    ospace
    -0.69
     underestimated
    -0.69
     skew
    -0.66
     Alchemist
    -0.65
     Ank
    -0.65
    andowski
    -0.65
    isks
    -0.64
     Quartz
    -0.64
    POSITIVE LOGITS
    shall
    1.29
     shall
    1.19
     hereby
    1.16
     lawful
    0.99
     proclamation
    0.93
    uties
    0.91
     SHALL
    0.91
     lawfully
    0.91
     solemn
    0.90
     Notwithstanding
    0.89
    Activations Density 0.246%

    No Known Activations