© 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. 16306
    Prev
    Next
    INDEX
    Explanations

    sentences related to societal issues, particularly regarding moral responsibility and societal criticism

    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
    +0.20
    0.9%
    687
    +0.09
    0.4%
    762
    +0.09
    0.4%
    Correlated Neurons
    Index
    P. Corr.
    Cos Sim.
    87
    +0.20
    -0.04
    762
    +0.09
    0.09
    86
    +0.09
    0.06
    Negative Logits
    toggle
    -0.73
    Enlarge
    -0.72
     Finder
    -0.68
    aceae
    -0.67
     peaks
    -0.64
    estones
    -0.63
    ãĥ´ãĤ¡
    -0.63
    sie
    -0.62
    catentry
    -0.61
    knit
    -0.60
    POSITIVE LOGITS
     hypocrisy
    1.28
     arrogance
    1.18
     irresponsible
    1.17
     hypocritical
    1.14
     disgrace
    1.12
     disrespect
    1.08
     disregard
    1.07
     hypoc
    1.05
     incompetence
    1.05
     disingen
    1.04
    Activations Density 1.456%

    No Known Activations