© 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. 4409
    Prev
    Next
    INDEX
    Explanations

    expressions related to moral judgment and disapproval, particularly focusing on the concept of shame

    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
    35
    +0.11
    0.5%
    326
    +0.11
    0.5%
    526
    +0.10
    0.5%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    334
    +0.11
    0.01
    35
    +0.11
    0.01
    326
    +0.10
    0.01
    Negative Logits
    enhagen
    -0.86
    ersive
    -0.82
    aeda
    -0.79
    ighters
    -0.75
    agall
    -0.74
    ancial
    -0.74
    agnetic
    -0.73
    IFE
    -0.70
    ichick
    -0.70
    女
    -0.68
    POSITIVE LOGITS
    fully
    1.36
    faced
    1.35
    imaru
    1.04
    face
    1.02
    fulness
    0.97
    ful
    0.97
     shame
    0.94
    ously
    0.90
     Shame
    0.90
     judgement
    0.89
    Activations Density 0.028%

    No Known Activations