© 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. 7438
    Prev
    Next
    INDEX
    Explanations

    phrases related to specific legal or moral issues, possibly related to the law or rules violation

    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
    79
    +0.10
    0.4%
    725
    +0.09
    0.4%
    601
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    79
    +0.10
    0.02
    206
    +0.09
    0.02
    665
    +0.09
    0.02
    Negative Logits
    orah
    -0.73
    lights
    -0.72
    arily
    -0.72
    mother
    -0.71
    ario
    -0.70
    esa
    -0.69
    ended
    -0.68
    til
    -0.65
    ãĥ¼ãĥ³
    -0.64
    2019
    -0.64
    POSITIVE LOGITS
     conversation
    0.98
     vigorous
    0.95
     hostilities
    0.92
     meaningful
    0.92
     continual
    0.89
     dialogue
    0.88
     conversations
    0.87
     discussions
    0.87
     mutual
    0.84
     risky
    0.84
    Activations Density 0.082%

    No Known Activations