© 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. 20200
    Prev
    Next
    INDEX
    Explanations

    corruption-related information involving cheating, bribery, and conspiracy

    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
    87
    +0.21
    1.0%
    40
    +0.10
    0.5%
    711
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    711
    +0.21
    0.03
    87
    +0.10
    -0.03
    323
    +0.10
    0.05
    Negative Logits
    ï¸ı
    -0.75
    uncture
    -0.71
     :(
    -0.67
    ibl
    -0.65
    )!
    -0.65
     ðŁij
    -0.64
     ðŁĺ
    -0.63
     ðŁ
    -0.63
    }}}
    -0.62
    terness
    -0.62
    POSITIVE LOGITS
     prostitutes
    1.04
     pornographic
    0.99
     bribes
    0.98
     allegedly
    0.94
     illicit
    0.93
     cocaine
    0.93
     fictitious
    0.90
     fraudulent
    0.89
     prostitute
    0.88
     purported
    0.88
    Activations Density 0.973%

    No Known Activations