© 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 EvalsExportsinterp-engineNEWAPI 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. 17052
    Prev
    Next
    INDEX
    Explanations

    instances of strict rules or regulations being discussed and enforced

    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
    365
    +0.11
    0.5%
    178
    +0.09
    0.4%
    440
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    365
    +0.11
    0.01
    658
    +0.09
    0.01
    67
    +0.09
    0.01
    Negative Logits
    ilater
    -0.86
    ovember
    -0.78
    assies
    -0.77
    igate
    -0.74
    NetMessage
    -0.72
    ********
    -0.70
    okemon
    -0.69
     Peninsula
    -0.68
    ̶
    -0.67
    ****
    -0.66
    POSITIVE LOGITS
    ures
    1.24
     adherence
    1.16
    ness
    0.94
    est
    0.90
     strict
    0.89
     adherent
    0.89
    nesses
    0.87
     limits
    0.86
    bred
    0.86
     restrictive
    0.83
    Activations Density 0.021%

    No Known Activations