© 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. 10937
    Prev
    Next
    INDEX
    Explanations

    phrases related to ethics, misconduct, and accountability in social and professional contexts

    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
    338
    +0.10
    0.5%
    481
    +0.10
    0.5%
    237
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    338
    +0.10
    0.06
    720
    +0.10
    0.05
    285
    +0.10
    0.03
    Negative Logits
    enhagen
    -0.98
    rosso
    -0.78
     clustered
    -0.77
    selected
    -0.71
    ebus
    -0.71
     Puzzles
    -0.70
    azines
    -0.69
     branching
    -0.69
    toggle
    -0.68
    ramids
    -0.67
    POSITIVE LOGITS
     disgrace
    1.39
     unacceptable
    1.31
     despicable
    1.18
     intolerable
    1.16
     disrespectful
    1.15
     remorse
    1.11
     irresponsible
    1.10
     shame
    1.10
     deserve
    1.10
     disrespect
    1.09
    Activations Density 0.622%

    No Known Activations